Proceedings of International Conference On Power Electronics and Renewable Energy Systems

Lecture Notes in Electrical Engineering 795

C. Subramani
K. Vijayakumar
Brayima Dakyo
Subhransu Sekhar Dash Editors

Proceedings
of International
Conference on Power
Electronics and
Renewable Energy
Systems
ICPERES 2021
Lecture Notes in Electrical Engineering

Volume 795
Series Editors
Leopoldo Angrisani, Department of Electrical and Information Technologies Engineering, University of Napoli
Federico II, Naples, Italy
Marco Arteaga, Departament de Control y Robótica, Universidad Nacional Autónoma de México, Coyoacán,
Mexico
Bijaya Ketan Panigrahi, Electrical Engineering, Indian Institute of Technology Delhi, New Delhi, Delhi, India
Samarjit Chakraborty, Fakultät für Elektrotechnik und Informationstechnik, TU München, Munich, Germany
Jiming Chen, Zhejiang University, Hangzhou, Zhejiang, China
Shanben Chen, Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Tan Kay Chen, Department of Electrical and Computer Engineering, National University of Singapore,
Singapore, Singapore
Rüdiger Dillmann, Humanoids and Intelligent Systems Laboratory, Karlsruhe Institute for Technology,
Karlsruhe, Germany
Haibin Duan, Beijing University of Aeronautics and Astronautics, Beijing, China
Gianluigi Ferrari, Università di Parma, Parma, Italy
Manuel Ferre, Centre for Automation and Robotics CAR (UPM-CSIC), Universidad Politécnica de Madrid,
Madrid, Spain
Sandra Hirche, Department of Electrical Engineering and Information Science, Technische Universität
München, Munich, Germany
Faryar Jabbari, Department of Mechanical and Aerospace Engineering, University of California, Irvine, CA,
USA
Limin Jia, State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing, China
Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland
Alaa Khamis, German University in Egypt El Tagamoa El Khames, New Cairo City, Egypt
Torsten Kroeger, Stanford University, Stanford, CA, USA
Yong Li, Hunan University, Changsha, Hunan, China
Qilian Liang, Department of Electrical Engineering, University of Texas at Arlington, Arlington, TX, USA
Ferran Martín, Departament d’Enginyeria Electrònica, Universitat Autònoma de Barcelona, Bellaterra,
Barcelona, Spain
Tan Cher Ming, College of Engineering, Nanyang Technological University, Singapore, Singapore
Wolfgang Minker, Institute of Information Technology, University of Ulm, Ulm, Germany
Pradeep Misra, Department of Electrical Engineering, Wright State University, Dayton, OH, USA
Sebastian Möller, Quality and Usability Laboratory, TU Berlin, Berlin, Germany
Subhas Mukhopadhyay, School of Engineering & Advanced Technology, Massey University,
Palmerston North, Manawatu-Wanganui, New Zealand
Cun-Zheng Ning, Electrical Engineering, Arizona State University, Tempe, AZ, USA
Toyoaki Nishida, Graduate School of Informatics, Kyoto University, Kyoto, Japan
Federica Pascucci, Dipartimento di Ingegneria, Università degli Studi “Roma Tre”, Rome, Italy
Yong Qin, State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing, China
Gan Woon Seng, School of Electrical & Electronic Engineering, Nanyang Technological University,
Singapore, Singapore
Joachim Speidel, Institute of Telecommunications, Universität Stuttgart, Stuttgart, Germany
Germano Veiga, Campus da FEUP, INESC Porto, Porto, Portugal
Haitao Wu, Academy of Opto-electronics, Chinese Academy of Sciences, Beijing, China
Walter Zamboni, DIEM - Università degli studi di Salerno, Fisciano, Salerno, Italy
Junjie James Zhang, Charlotte, NC, USA
The book series Lecture Notes in Electrical Engineering (LNEE) publishes the
latest developments in Electrical Engineering - quickly, informally and in high
quality. While original research reported in proceedings and monographs has
traditionally formed the core of LNEE, we also encourage authors to submit books
devoted to supporting student education and professional training in the various
fields and applications areas of electrical engineering. The series cover classical and
emerging topics concerning:
• Communication Engineering, Information Theory and Networks
• Electronics Engineering and Microelectronics
• Signal, Image and Speech Processing
• Wireless and Mobile Communication
• Circuits and Systems
• Energy Systems, Power Electronics and Electrical Machines
• Electro-optical Engineering
• Instrumentation Engineering
• Avionics Engineering
• Control Systems
• Internet-of-Things and Cybersecurity
• Biomedical Devices, MEMS and NEMS
For general information about this book series, comments or suggestions, please
contact [email protected].
To submit a proposal or request further information, please contact the
Publishing Editor in your country:
China
Jasmine Dou, Editor ([email protected])
India, Japan, Rest of Asia
Swati Meherishi, Editorial Director ([email protected])
Southeast Asia, Australia, New Zealand
Ramesh Nath Premnath, Editor ([email protected])
USA, Canada:
Michael Luby, Senior Editor ([email protected])
All other Countries:
Leontina Di Cecco, Senior Editor ([email protected])
** This series is indexed by EI Compendex and Scopus databases. **
More information about this series at https://link.springer.com/bookseries/7818

C. Subramani · K. Vijayakumar · Brayima Dakyo ·
Subhransu Sekhar Dash
Editors
Proceedings of International
Conference on Power
Electronics and Renewable
Energy Systems
ICPERES 2021
Editors
C. Subramani K. Vijayakumar
Department of Electrical and Electronics Department of Electrical and Electronics
Engineering Engineering
SRM Institute of Science and Technology SRM Institute of Science and Technology
Kattankulathur, Tamil Nadu, India Kattankulathur, Tamil Nadu, India
Brayima Dakyo Subhransu Sekhar Dash

Department of Electrical and Electronics Department of Electrical and Electronics
Engineering Engineering
University of Le Havre Government College of Engineering
Le Havre, France Keonjhar, Odisha, India
ISSN 1876-1100 ISSN 1876-1119 (electronic)

Lecture Notes in Electrical Engineering
ISBN 978-981-16-4942-4 ISBN 978-981-16-4943-1 (eBook)
https://doi.org/10.1007/978-981-16-4943-1
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
This ICPERES 2021 volume contains the papers presented at the International
Conference on Power Electronics and Renewable Energy (ICPERES 2021) held
during April 21–23, 2021, at SRM Institute of Science and Technology, Kattanku-
lathur, India. ICPERES 2021 is the International conference aiming at bringing
together the researchers from academia and industry to report and review the latest
progresses in the cutting-edge research on various research areas of power electronics,
renewable energy and its applications to create awareness about these domains to a
wider audience of practitioners.
ICPERES 2021 received 325 paper submissions, including seven foreign countries
across the globe. All the papers were peer-reviewed by the experts in the area in India
and abroad, and comments have been sent to the authors of accepted papers. Finally,
63 papers were accepted for oral presentation at the conference and for the publication
in Lecture Notes in Electrical Engineering (LNEE) series. This corresponds to an
acceptance rate of around 20% and is intended to maintain the high standards of
the conference proceedings. The papers included in this Springer volume cover a
wide range of topics in power electronics, renewable energy and their real-time
applications in problems from diverse domains of science and engineering.
The conference was inaugurated by Dean (CET), SRMIST, on April 21, 2021. The
conference featured distinguished keynote speakers as follows: Dr. S. Akhtar Kalam,
Head of External Engagement Leader—Smart Energy Research Unit, Victoria
University, Australia; Dr. Sidhartha Panda, Professor, Electrical Engineering, Veer
Surendra Sai University of Technology, Odisha; Dr. S. S. Dash, GCE, Keonjhar,
Odisha, India; Dr. Vivekananda Mukherjee, Professor, Electrical Engineering, Indian
Institute of Technology Jharkhand; Dr. Ramazan Bayindir, Professor, Electrical Engi-
neering, Gazi University Technology, Turkey; and Dr. Anup Kumar Panda, Professor,
Electrical Engineering, National Institute of Technology Rourkela.
We take this opportunity to thank the authors of the submitted papers for their
hard work, adherence to the deadlines, and patience with the review process. The
quality of a referred volume depends mainly on the expertise and dedication of
the reviewers. We are indebted to the technical committee members, who produced
excellent reviews in short time frames.
v
vi Preface
First, we are indebted to the honorable Dr. T. R. Paarivendhar, Chancellor, Mr. Ravi
Pachamoothoo, Pro-chancellor (Admin), Dr. P. Sathyanarayanan, Pro-chancellor
(Academics), and Dr. R. Shivakumar, Vice President, SRMIST, wholeheartedly for
the confidence they entrusted on us for organizing this international conference
ICPERES 2021. We sincerely thank our beloved interim Vice Chancellor for their
continuous support and guidance in organizing the conference. Our heartfelt thanks to
the Registrar, HODs, Professors, and the staff members of SRMIST, Kattankulathur,
for their valuable support for the success of this program. We thank the International
Advisory Committee members for providing valuable guidelines and inspiration
to overcome various difficulties in the process of organizing this conference. We
would also like to thank the participants of this conference. The members of faculty
and students of SRM Institute of Science and Technology, Chennai, deserve special
thanks because, without their involvement, we would not have been able to face the
challenges of our responsibilities. Finally, we thank all the volunteers who made
great efforts in meeting the deadlines and arranging every detail to make sure that
the conference could run smoothly. We hope the readers of these proceedings find
the papers inspiring and enjoyable.
Kattankulathur, India Prof. K. Vijayakumar

Kattankulathur, India Dr. C. Subramani
Le Havre, France Brayima Dakyo
Keonjhar, India Subhransu Sekhar Dash
May 2021
Conference Committee
Organizing Chairs
Prof. K. Vijayakumar
Dr. C. Subramani
Chief Patrons
Dr. T. R. Paarivendhar, Founder Chancellor

Shri. Ravi Pachamoothoo, Pro-Chancellor (Admin)
Dr. P. Sathyanarayanan, Pro-Chancellor (Academic)
Patrons
Dr. C. Muthamizhchelvan, Vice Chancellor, SRMIST

Dr. T. V. Gopal, Dean (Engineering and Technology), SRMIST
General Chairs
Dr. Ramazan Bayindir, Gazi University, Turkey

Dr. Florian Misoc, UWM, USA
Dr. Brayima Dakyo, University of Le Havre, France
vii
viii Conference Committee
Program Chairs
Dr. Valentina Emilia Balas, UOA, Romania

Dr. G. S. Chae, Baekseok University, South Korea
Dr. S. S. Dash, Government College of Engineering, Keonjhar
Organizing Chairs
Prof. K. Vijayakumar, SRMIST

Dr. C. Subramani, SRMIST
Co-organizing Chairs
Dr. A. Rathinam, SRMIST

Dr. R. K. Pongiannan, SRMIST
Dr. N. Chellammal, SRMIST
Dr. C. S. Boopathi, SRMIST
Organizing Committee
Dr. Y. Jeyashree, SRMIST

Dr. D. Suchitra, SRMIST
Dr. R. Sridhar, SRMIST
Dr. D. Sattianadan, SRMIST
Dr. C. Bharatiraja, SRMIST
Dr. J. Preetha Roselyn, SRMIST
Dr. K. Saravanan, SRMIST
Dr. M. Jagabar Sathik, SRMIST
Dr. K. Mohanraj, SRMIST
Dr. Arun Noyal Doss, SRMIST
Dr. S. Padmini, SRMIST
International Advisory Committee
Dr. Akhtar Kalam, Victoria University, Australia

Dr. Arunachalam Sundaram, Jubail Industrial College, Kingdom of Saudi Arabia
Conference Committee ix
Dr. C. A. Vaithilingam, Taylor University, Malaysia

Dr. Diego Bellan, Politecnico di Milano, Italy
Dr. Dirman Hana, Universiti Tun Hussien Onn Malaysia (UTHM), Malaysia
Dr. Eliathamby Ambikairajah, University of New South Wales, Australia
Dr. Jahangir Hossain, Macquarie University, Australia
Dr. Jayashri Ravishankar, University of New South Wales, Australia
Dr. Naayagi Ramasam, Newcastle University, Singapore
Dr. Narottam Das, CQUniversity, Australia
Dr. Prahlad Vadakkepat, National University of Singapore
Dr. Raj Jain, Washington University, St. Louis
Dr. Ramani Kannan, Universiti Teknologi Petronas, Malaysia
Dr. S. Umashankar, Prince Sultan University, UAE
Dr. Ruzairi Abdul Rahim, Universiti Tun Hussein Onn Malaysia (UTHM), Malaysia
Dr. Saad Mekhilef, University of Malaya, Malaysia
Dr. P. Sanjeevikumar, Aalborg University, Denmark
Dr. Siti Hawa Ruslan, Universiti Tun Hussein Onn Malaysia (UTHM), Malaysia
National Advisory Committee
Dr. Kumar Pradhan, IIT Kharagpur

Dr. Bhim Singh, IIT Delhi
Dr. Ganapati Panda, IIT Bhuvaneshwar
Dr. S. Jeevananthan, PEC, Pondicherry
Dr. Debashish Jena, NITK Surathkal
Dr. N. P. Padhy, IIT Roorkee
Dr. S. Chandramohan, Anna University
Dr. G. Uma, Anna University
Dr. K. Udhayakumar, Anna University
Dr. M. Somasundaram, Anna University
Dr. M. Arun Bhaskar, VEC, Chennai
Dr. S. Raghuraman, VEC, Chennai
Dr. D. Kalpana, MIT, Anna University
Dr. M. Jegadeeshkumar, Sairam Institute of Technology
Dr. G. Prakash, Sairam Institute of Technology, Chennai
Dr. Sidhartha Panda, VSSUT EC, Burla
Dr. M. Mohanthy, SOA University, Odisha
Dr. K. Sureshkumar, VEC, Chennai
Paper Selection and Publication Process
Dr. A. Geetha, SRMIST

x Conference Committee
Dr. T. M. Thamizh Thentral, SRMIST

Dr. S. Usha, SRMIST
Dr. A. Dominic Savio, SRMIST
Dr. R. Brindha, SRMIST
Mr. B. Vinothkumar, SRMIST
Contents
Deep CNN Depth Decision in Intra Prediction . . . . . . . . . . . . . . . . . . . . . . . . 1

Helen K. Joy and Manjunath R. Kounte
Investigation of Electrical Characteristics for Different
Geometrical Dimensions of 11 kV Insulators Under Finite Element
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
R. Bharanidharan and R. V. Maheswari
Non-isolated Multilevel Zeta Converter for MLI Application . . . . . . . . . . 21
Marikannu Marimuthu, Subramanian Vijayalakshmi, B. Paranthagan,
R. Venugopal, S. Srinithi, B. Yuvaraj, R. Soundarajan,
and S. K. Vasantha Kumar
Corrupted Image Enhancement Through WaveNet: A Hybrid
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
C. Vimala and P. Aruna Priya
Modeling and Control of Cascaded Multilevel Inverter
for Harmonics Mitigation of Induction Motor Drive . . . . . . . . . . . . . . . . . . 39
Ranganathan Selvarasu, C. Kannan, S. Priyadharsini,
and Dagne Alemayehu Shiferaw
Assessment of Various Vector Control Schemes for PMSM Drive
Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Kodumur Meesala Ravi Eswar, Chokkalingam Bharatiraja,
and Jayakumar Vinoth
Design and Implementation of Discrete Controller-Based Zeta
Converter for Solar Power Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
R. Uthirasamy, V. Kumar Chinnaiyan, U. S. Ragupathy,
and S. Vishnu Kumar
xi
xii Contents
Structural Optimization of LMS Adaptive Filter Using Multi-stage

Cascaded Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
S. Hannah Pauline, Samiappan Dhanalakshmi, and R. Kumar
THD Optimization with Low Switching Frequency Control
for 15-Level Reduced Switch Asymmetric Multilevel Inverter . . . . . . . . . . 81
Gireesh Kumar Devineni, Aman Ganesh, Neerudi Bhoopal,
and D. S. N. M. Rao
Signal Conditioning Circuits for GMR Sensor in Biomedical
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
G. Anand, T. Thyagarajan, B. Aashique Roshan, L. Rajeshwar,
and R. Shyam Balaji
Modeling and Simulation of SPVGC System Using Conventional
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
P. Chandra Babu, B. Venkata Prasanth, and P. Sujatha
Study of the Operating Parameters of a Two-Bed Adsorption
System: Comparison Between the Silica–Gel–Water and Zeolite–
Water Pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Mouhamadou Lamine Cisse and Biram Dieng
Investigation of Hybrid Fault Ride Through Control Strategies
in Grid-Connected Wind Energy Conversion Systems . . . . . . . . . . . . . . . . . 137
J. Preetha Roselyn, C. R. Raghavendran, and D. Devaraj
Performance Analysis of Single-Axis Solar Tracker Using IoT
Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
S. Usha, A. Geetha, T. M. Thamizh Thentral, C. Subramani,
R. Ramya, and C. S. Boopathi
Wireless Power Transfer for Electric Vehicles with Better Ground
Clearance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
A. Geetha, S. Usha, T. M. Thamizh Thentral, C. Subramani,
J. Santha Kumar, and C. S. Boopathi
Five-Level PUC Inverter-Based Shunt Active Power Filter
for Harmonic Elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
T. M. Thamizh Thentral, K. Vijayakumar, S. Usha, A. Geetha,
and C. S. Boopathi
Performance of 100-kW Rooftop PV Plant in Library
Building—A Case Study in SRMIST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
J. Divya Navamani, Tanmay Padhi, Aditi Kumari, and A. Lavanya
Design and Implementation of a Low-Cost Mini Heliostat Solar
Tracking System in West Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Marie Pascaline Sarr, Ababacar Thiam, Biram Dieng,
and El Hadji Ibrahima Cisse
Contents xiii
Power-Domain NOMA for Massive Connectivity in Smart Grid

Communication Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
M. Jayachandran and C. Kalaiarasy
Power Management in DC Microgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
B. Balaji and S. Ganesan
Embedded System for Lethal Gas Leakage Exposure
and Forewarning System Using Arm Processor . . . . . . . . . . . . . . . . . . . . . . . 227
M. Nirmala and S. Durga Shree
A Trasformerless Buck-Boost Converter as Maximum Power Point
Tracker for Battery Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
C. Balaji, O. Hemakesavulu, A. Dominic Savio, B. Vinothkumar,
S. Sakthi, and P. Sivaperumal
A Feed-Forward Neural Network Based MPPT Controller
for PEMFC System with Ultra High Step Up Converter . . . . . . . . . . . . . . . 247
B. Karthikeyan, D. Karthikeyan, V. P. Arumbu, K. Sundararaju,
R. Palanisamy, and P. Divya
Predictive Maintenance of Industrial Equipment’s Using IOT . . . . . . . . . 257
S. Nithya, K. Vijayalakshmi, and M. Parimala Devi
A Simplified Beginner’s Guidelines for Design and Fabrication
of Prototype Electrical Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
P. Ramesh Babu, P. Vigneshwar, R. Udaya Simha,
S. Tanweer Ahamed, S. Vengatesh, and V. Vijay
Performance Analysis of IPT with DC to DC Converter
for E-Vehicle Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
V. Senthil Nayagam and L. Premalatha
Operating Cost Analysis of Microgrid Including Renewable
Energy Sources and a Battery Under Dynamic Pricing . . . . . . . . . . . . . . . . 291
Hephzibah Jose Queen, J. Jayakumar, and T. J. Deepika
A Comparative Study of Field Distribution Properties of Different
Types of Oils Using Comsol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
M. Divya Bharathi, R. V. Maheswari, and S. Senthil Kumar
Managing the Smart Grid with Demand Side Management Using
AntLion Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Banala Venkatesh and S. Padmini
Comparison of PI Based and ANN Based Dynamic Voltage
Restorer Controller for Voltage Sag Mitigation in Distribution
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
T. Jane Tracy and N. Rathina Prabha
xiv Contents
Design of Quadratic Boost Converter for Renewable Applications . . . . . . 331

B. Abinayalakshmi, S. Muralidharan, and J. Gnanavadivel
Recognition of Partial Discharge Signal Using Deep Learning
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
J. Ashmin Sugaji, M. Ravindran, and R. V. Maheswari
Design and Analysis of LK Model Based FEFET Memories . . . . . . . . . . . . 353
S. S. Vijayavelu, K. Mariammal, M. Adhitya Narayan,
and P. Subash Rathinam
Convolutional Encoder–Decoder Architecture for Speech
Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Utkarsh Maheshwari, Piyush Goel, R Annie Uthra,
Vinay Vasanth Patage, Sourabh Tiwari, and Saksham Goyal
Tuning of MIMO PID Controller Using HCLPSO Algorithm . . . . . . . . . . 377
T. Jeyaraman, D. Joelpraveenkumar, M. Kaliraj, M. Krishna Chandar,
and M. Willjuice Iruthayarajan
Design Guide for Small-Scale Grid-Connected PV System Using
PVsyst Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
A. Lavanya, Kushagra Bhatia, J. Divya Navamani, A. Geetha,
and K. Vijayakumar
Evaluation of Various Machine Learning Algorithms for Detection
of Attacks in 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
C. Arul Stephen, R. Mathesh, L. Venkat, B. Ebenezer Abishek,
and A. Vijayalakshmi
Performance Analysis of Slope-Compensated Current Controlled
Universal PV Battery Charger for Electric Vehicle Applications . . . . . . . . 407
S. Ramprasath, R. Abarna, G. Anjuka, K. Deva Priya, S. Iswarya,
and C. Krishnakumar
Comparative Analysis of Solar PV Production in Durban to Other
Cities in South Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Sanjeeth Sewchurran, Innocent E. Davidson, and Elutunji Buraimoh
An Experimental Analysis of the Impact of a Grid-Tied
Photovoltaic System on Harmonic Distortion . . . . . . . . . . . . . . . . . . . . . . . . . 431
Abayomi A. Adebiyi, Ian J. Lazarus, Akshay K. Saha, and Evans E. Ojo
Genetic Algorithm Based Energy Management in Microgrid . . . . . . . . . . 445
M. Sadees, Aishwarya Raju, Utsahan Mukherjee, and K. Vijayakumar
Improvement of Voltage Stability in Micro Grid System Using
Hybrid Power Flow Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
R. Suganya and M. Anitha
Contents xv
An Intelligent Fuzzy Controlled Microgrid Fed by Distribution

Generation Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Md. Aijaz, T. Muthamizhan, and T. Venkateswarlu
Multiresolution Representation of SONAR Pipeline Image Using
Pyramidal Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
R. Kumudham, P. Sathish Kumar, V. Rajendran, M. S. Jagan Mugesh,
and U. Charan Raj
Dual Motor Power Management Strategy for Plug-in Hybrid
Electric Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
Vinoth Kumar Balan and P. Avirajamanjula
Investigation on Power System Stability Improvement Using Facts
Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
Gajana Penchalaiah and R. Ramya
Power Quality Enhancement Using Interline Dynamic Voltage
Restorer in Renewable Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
T. Ahilan, P. Suresh, S. Elam Cheren, and G. Ramya
Enhanced Hybrid Touch Screen Display for Industrial Applications . . . . 517
A. Vignesh Babu, J. Ajay Daniel, V. N. Ganesh, S. Balaji, and G. Ramya
Reliability Study on the Distribution System Integrated with Wind
Generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
S. B. Aruna and D. Suchitra
Retrofitting of Internal Combustion Engine Vehicles with DC Motor . . . 537
Femi Robert, Muskan Puri, Ashay Kumar Thakur,
and Gajendran Marimuthu
Analysis of Cascaded Multilevel Inverters for Smart Grid
Connected PV Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
R. L. Josephine, M. Yoogesh Kumar, and S. Harishankar
Voltage Stability Improvement of PV and Battery-Based Sliding
Mode-Controlled Microgrid System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
V. N. Ganesh and S. Manivannan
Electric Propulsion System with Dual-Motor Power Management
Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
Vinoth Kumar Balan, P. Avirajamanjula, and A. Dominic Savio
Realization and Implementation of Peak Energy Management . . . . . . . . . 579
P. Kanakaraj and L. Ramesh
Novel Template Protection Scheme for Multimodal Data . . . . . . . . . . . . . . 593
C. Hari Akhilesh, M. Gayathri, and C. Malathy
xvi Contents
Performance Optimization of Stepper Motor Using MATLAB . . . . . . . . . 603

R. Gopalakrishnan, A. Dominic Savio, K. Dhayalini,
and Nithya Rani Navaneethan
Performance and Analysis of Voltage Compensation
in Transmission Line Using SMES-Based IDVR . . . . . . . . . . . . . . . . . . . . . . 613
M. Jagadeesh Kumar, T. Muthamizhan, P. Rathnavel, G. Ezhilarasan,
and T. Eswaran
Multiple Regression-Based Adaptive Protection Scheme
for Microgrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
Aindrila Dey, Moumita Dhang, Rangit Ray, and C. Vaithilingam
Power Quality Improvement in Distribution Network Using
Levenberg–Marquardt (LM) Algorithm-Based Controller
for Active Current Harmonic Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
P. Balamurugan, N. Senthil Kumar, and C. Vaithilingam
Modified Multiverse Optimization, Perturb and Observer
Algorithm-Based MPPT for Grid-Connected Photovoltaic System . . . . . 647
N. K. Rayaguru and S. Sekar
IOT-Monitored EV Charging Stations Using DC–DC Converter
with Integrated Split Battery Energy System . . . . . . . . . . . . . . . . . . . . . . . . . 659
M. Arun Noyal Doss, R. Brindha, A. Ananthi Christy,
and Viswanathan Ganesh
Implementation of TLBO Optimised PIλ Dμ Controller for LFC
of Nonlinear Reheat Thermal Power System . . . . . . . . . . . . . . . . . . . . . . . . . 671
Nimai Charan Patel, Binod Kumar Sahu, and Ramesh Chandra Khamari
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689

About the Editors
Dr. C. Subramani received Master of Engineering

(2006) from Anna University (India), doctoral degree
(2012) from SRM University (India) and is Postdoctoral
Fellow (2016) from Tshwane University of Technology
(South Africa). Professional position: He has joined
the SRM University as Assistant Professor from 2009
to till date, in the Department of Electrical and Elec-
tronics Engineering, Faculty of Engineering and Tech-
nology. He has more than 15 years of teaching experi-
ence. Research: Professor Subramani has published and
co-authored more than 90 papers in scientific journals
and conference proceedings. He has been Advisor of 11
Ph.D. and numerous postgraduate students. He has been
acting as Reviewer for many national and international
journals. Research areas: His broad areas of research are
electric vehicle, energy management control strategy,
fuzzy logic controller, particle swarm optimizations, etc.
Prof. K. Vijayakumar has received B.E. degree in

Electrical and Electronic Engineering from Annamalai
University, M.Tech. (Power Systems) from Annamalai
University, and Ph.D. from SRM Institute of Science and
Technology. He is currently working as Professor and
Head in the Department of Electrical and Electronics
Engineering, SRM Institute of Science and Technology
(India). His area of research is power electronic appli-
cations in power system. His current research is focused
on optimizations, hybrid energy systems, and deregula-
tion. He has more than 23 years of teaching experience.
Research: Professor Vijayakumar has published and co-
authored more than 100 papers in scientific journals and
conference proceedings. He has been Advisor of eight
xvii
xviii About the Editors
Ph.D. and numerous postgraduate students. He has been

acting as Reviewer for many national and international
journals. He is also one of the editors of few of the books
titled Basic Electrical Engineering and Electrical Engi-
neering Practice Lab Manual of Vijay Nicole Imprints
Private Limited.
Prof. Brayima Dakyo received Engineer Dipl. (1984)

and Doctor in Engineering (1987) degrees from Dakar
University (Senegal) and Doctoral degree (1988) and
Habilitation (1997) from University of Le Havre
(France). Professional position: He has joined the
University of Le Havre as Assistant Professor (MdC)
from 1989 to 1999, in the Faculty of Sciences and
Technology. He was Founder and Director of Electrical
and Automatic research Laboratory (GREAH) 1999.
From 2000 to now, he is Full Professor of Electrical
Engineering and Director of Electrical and Automatic
Research Team of Le Havre (GREAH) (15 full profes-
sors and assistant professors plus 20 Ph.D. students
among whom six are averagely involved in renewable
energies field). Professor Dakyo is General Advisor of
the electrical engineering department, Master’s degree
and bachelor’s degree. Research: Professor Dakyo has
published and co-authored more than 180 papers in
scientific journals and conference proceedings since
1987. He has been Advisor of 20 Ph.D. and numerous
M.Sc. students and Co-Author of three books (power
electronic, marine and wind energy, and power manage-
ment).
Dr. Subhransu Sekhar Dash is presently Professor

and Head in the Department of Electrical Engineering,
Government College of Engineering, Keonjhar, Odisha,
India. He received his Ph.D. degree from College of
Engineering, Guindy, Anna University, Chennai, India.
He has more than 22 years of research and teaching
experience. His research areas are AI techniques appli-
cation to power system, modeling of FACTS controller,
power quality, and smart grid. He is Visiting Professor
at Francois Rabelais University, POLYTECH, France.
He has published more than 220 research articles in
peer-reviewed international journals and conferences.
Professor Dash has played a role as Convenor, Program,
or General Chair for many International Conferences.
About the Editors xix
He is also one of the editors of few of the books titled

Introduction to FACTS and Power Quality Management
of IRD Publication and Vijay Nicole Imprints Private
Limited, respectively.
Deep CNN Depth Decision in Intra
Prediction
Helen K. Joy and Manjunath R. Kounte
Abstract The video and its compression become prominent with the emergence of
digital video technology and common use of video acquisition devices. The tradi-
tional video compression needs upgradation with artificial intelligence, machine
learning, neural network, and deep learning. Apart from normal signal processing
the deep learning technologies are advantages as they can deal with content analysis
than dealing only with neighboring pixels. The initial steps in video compression,
intra/inter frame prediction provide a better percentage in overall compression. The
computational complexity of existing intra prediction method is more. This paper
proposes a deep learning based intra prediction method using CNN. This deep depth
prediction algorithm trains the network to provide depth of the CTU with reduced
computation and less time. The experimental results show a dip in the encoding time,
about 71.3% compared to existing method.
Keywords Deep learning · Video compression · HEVC · CNN
1 Introduction
Various compression methods became so prominent when the transfer of data at a

low bit rate became a need, but the quality of the image was also a concern. So,
video compression and its variance became an eminent one in the research area.
Digging back to the history of video compression the traditional methods started
from basic coding like Huffman coding followed by frequency transformation to
recent techniques like hybrid video processing techniques, HEVC, and its advanced
version. Based on the need of the user and various other features there is various
compression.
H. K. Joy (B)
School of Electronics and Communication, REVA University, Rukmini Knowledge Park,
Kattigenahalli, Yelahanka, Bengaluru, Karnataka 560064, India
M. R. Kounte
Department of Electronics and Computer Engineering, School of Electronics and
Communication, REVA University, Rukmini Knowledge Park, Kattigenahalli, Yelahanka,
Bengaluru, Karnataka 560064, India
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 1
C. Subramani et al. (eds.), Proceedings of International Conference on Power Electronics
and Renewable Energy Systems, Lecture Notes in Electrical Engineering 795,
https://doi.org/10.1007/978-981-16-4943-1_1
2 H. K. Joy and M. R. Kounte
1.1 Evolution of Video Compression Development

by ISO/IEC/MPEG
“Motion JPEG (M-JPEG or MJPEG) is an initial video compression technique in

which each frame in the video or interlaced field of the video in digitalized sequence
is compressed separately as an image in jpeg format. It was developed focusing on
multimedia personal computer applications, M-JPEG is now used by video-capture
devices like digital cameras, webcams IP cameras, etc. and by non-linear video
editing systems too. It is supported by QuickTime Player, the PlayStation console,
and web browsers such as Microsoft Edge Safari, Google Chrome, Mozilla Firefox,
etc.” [1]. MPEG-1 standard is the initial version of the MP3 audio format introduced,
i.e., best-known part of this format. MPEG-2 came out of the drawbacks of MPEG-
1 [2]. “MPEG-2 is used widely in digital television signal which is broadcast by
terrestrial, and DBS (direct broadcast satellite television system) followed by MPEG
4, MPEG H has several “parts”—each of them can be considered a different standard
[3]. These include a video compression standard, media transport protocol standard,
a digital file format container standard, an audio compression standard, three sets of
conformance testing standards, set of three reference software packages, and related
technologies and technical reports.
1.2 Evolution of Video Compression Development

by ITU/VCEG
ITU-T video compression standard introduced H.261 is the year November 1988. In
the H.26 × family, first member, H.261 IN “video coding standards in the domain
of the VCEG [4] (ITU-T Video Coding Experts Group) then Specialists Group on
Coding for Visual Telephony” [1]. MPEG-2 consists of “three different kinds of
coded frames: I-frame/intra-coded frames, P -frame/predictive-coded frames, and
B-frame/bidirectionally-predictive-coded frames” [1, 5]. H.263 is a member of the
H.26 × family of video coding standards in the domain, ITU-T. Like other H.26 ×
standards, H.263 is also based on (DCT) discrete cosine transform video compres-
sion. It was later advanced to add different additional enhanced features in the years
1998 and 2000.H.264 mostly called AVC (Advanced video coding), its block segmen-
tation based, motion compensated with DCT technique. The aim behind AVC was to
transfer video at low bit rate with better efficiency for UHD videos. “High Efficiency
Video Coding, is also known as HEVC or H.265 [6], is the step in this evolution.
It builds off a lot of the techniques used in AVC/H.264 to make video compression
even more efficient.
Deep CNN Depth Decision in Intra Prediction 3
1.3 Evolution of Video Compression Development by SMPTE
“SMPTE 421 M technology was developed with contributions from a number of

companies, with the majority of patent contributions from Microsoft, Panasonic,
LG Electronics, and Samsung Electronics.VC-1 is supported in the now deprecated
Microsoft Silverlight, the now discontinued HD DVD, and in the Blu-ray Disc” [1].
DNxHD is a video coding format intended to be used as both an intermediate format
good to use in editing and in presentation format. VC3 data is typical can be stored in
an MXF container, still, it can also be saved in a QuickTime container. 1.4 Evolution
of video compression development by On2 True Motion.
On2 True Motion’s VP3 is a lossy video compression format that is royalty-free
and a video codec that focus on web activities. It is an incarnation done by “True
Motion video codec; a set of video codecs created by On2 Technologies. There is no
standard.
specification for the VP3 bitstream format above the VP3 source code done by
On 2 Technologies. It was given to the public as open source, and On2 irrevocably
disclaimed all rights to it, granting a royalty-free license grant for any patent claims
it might have over the software and any derivatives, allowing anyone to use any VP3-
derived codec for any purpose” [1]. VP4 codec and the RealSystem iQ architecture
became the only streaming media platform capable of delivering the VP4 codec” [7].
VP7 codec was used by Move Media Player plugin for Internet Explorer, Firefox,
and for the streaming of full network shows used by ABC and Fox. VP9 is a royalty-
free and open video coding format developed by Google that was mainly used on
Google’s video platform YouTube. VP9 support is common among web browsers,
Android, etc. [1].
Currently, the video compression is mainly focusing on smart compression that
is the incorporation of the goodness of Artificial intelligence (AI), Machine learning
(ML), Neural network (NN), and Deep Learning (DL). As the scope of machine
learning based systems to process the pixels are not bounded to the neighboring
pixels, as in general signal processing, the content-based analysis can be considered.
This is a major breakthrough in all image and video processing.
Video compression procedure can be mainly made into some general steps that
all compression standard follows. The methodology or approach changes but steps
remain the same and it follows so, sampling the frames, intra frame prediction,
inter frame prediction, Transform, scaling, quantization, entropy coding, and loop
filtering.
This paper mainly focuses on how a deep learning based intra frame prediction can
be done with less computational complexity. the paper starts with the introduction
evolution and steps in video compression. Then the focus will be streamlined to
intra frame prediction using neural network and deep learning application with the
proposed algorithm and results and concluded finally pointing how deep learning
technique can reduce computational complexity in intra frame prediction.
2 Intra Prediction in Video Compression
Intra prediction provides a major role in the compression of video and it’s the
topic to be taken care when intelligent VC is taken into consideration. Intra predic-
tion is defined as predicting the redundant information within the frame, In recent
video systems and commonly used compression tech, the role of intra prediction to
compression is good. As the frame is divided into macro blocks in H.264 or CTU in
H265, the technique of comparison is of frames similarities. Comparing and encoding
same CU helps in better compression efficiency. The recent VC technique, HEVC
[8] in market is using Quadra tree structure CTU. CTU is of variable size from 64 ×
64 CV to 8 × 8 that is, processing variable depth from larger coding unit (LCU) to
smaller coding unit. PU (prediction unit) is defining CTU split is required or not, so
PU is the basic deciding unit for intra/inter prediction.
HEVC has proved its capability to reduce bit rate by more than half percentage
than its previous encoder [8]. The bit rate reduction HEVC could afford the bit
rate saving because of the new encoding method of CTU splitting. But the impact
of this is affected on computation, complexity of computing increased as the new
CTU encoding was introduced. So, the recent focus is to reduce the complexity by
maintaining other parameters. Results have proved that complexities in calculation
or more observed in inter/intra prediction, it’s 3.2 times complex than H.264. CTU
partitioning and prediction are observed as the reason for the encoding. It can be
noted that working the process to reduce the CTU partitioning can help the compu-
tational complexity and thus fastening up the total encoding process. In HEVC the
RD cost of various coding units from LCU is 64*64 to SCV 8*8 is calculated in each
CTU with optimism depth. All CU [9] poses a prediction unit that holds the info of
intra/inter prediction. Taking into consideration the better compression efficiency,
the rate distortion cost is calculated separately and for all the PU modes of depth.
This makes the calculation complex but helps to attain better compression than the
previous method.
Algorithms are proposed to reduce the complexity in calculation with better
encoding efficiency minimum encoding efficiency. To reduce the computational
complexity many algorithms were proposed that can be mainly of learning based
algorithms or manually tuned algorithms considering various parameters to check
complexity reduction. Cho et al. used a Bayesian decision rule that uses the RD cost
of full and low complexity CTUS to develop a better CTU splitting of purring. Based
on the no of high frequency key points in CU, Kim et. all proposed a CTU spieling
algorithm.
Some algorithms were proposed to reduce the comparing the size of PU to have
a fast PU six [10] decision. As it is developed manually the trial and error method
works more here that is a major disadvantage noted. Taking into consideration one
of the parameters to work manually may affect the other dependent parameters and
may affect the total RD performance [11]. The next is learning based algorithms.
Learning based algorithms for the intra prediction helps to reduce video coding
complexity by learning using the proposed by learning using dataset extensivity and
building a model, Here classification algorithm with logistic regression. Lu et al.

SVM support vector margin for classification [12]. These algorithms wanted some
features, i.e., inbuilt/manually extracted features to predict the CTU partitioning.
Extracting the features may help in future for CNN and deep learning approach to
predict actual CTU structure. The CNN tried its best to predict CTU perfectly but
the shallow net has its limitation to extract enough features to accurately predict the
CTU partition.
The paper proposes a deep layer of learning with an extensive dataset to extract the
underlying features to partition the CTU in an effective way using deep CNN network.
The proposed system focuses on deep CNN based intra prediction that efficiently
predicts the CTU should be split on non split using deep CNN network learning
methods. The data base should be created effectively, an efficient database that covers
all the 4 quantization parameter ranges are chosen for this analysis. It should be
trained effectively to extract the deeper information and multiple information and
multiple patterns in the CTU splitting. The model used is deep architect use that uses
all suitable kernel sizes to match the CTU splitting from 64*64 to 16*16. Comparing
to the computational complexity in HEVC the system architecture will properly
predict the inter CTU split/non split by a reduced vector of length 16*1 with the
total depth information of the CTU. This will effectively reduce the time taken for
encoding and make the total system fast and efficient.
3 Data Set and CTU Splitting
The data set used here contains coding unit and its corresponding intra prediction
depth. In HEVC intra frame prediction 64X64 coding tree units and its CU /PU, that
constitute the basic block of intra frame prediction can be represented by 16 × 1
vector. In HEVC it is represented by a 16 × 16 matrix with depth values 0,1,2,3.
The dataset used is YUV dataset where Y represents the luminance and U and V
represent the chrominance of blue and red. The YUV data set has a set of image files
of different sizes extracted from video. From that information, the 64 × 64, 32 ×
32, 16 × 16, etc. can be extracted. If image files are into 64 × 64 CTUs, the size of
the train dataset is around 110 K images. The size of the validation dataset is around
40 K images. Testing validation and training of these datasets will be done as an
initial step.
3.1 Idea of CTU Splitting
According to CTU splitting in HEVC, it can be split or non split, i.e., it’s a binary
classifier. But the 64 × 64 CTU can be split again to 32 × 32, 16 × 16, and 8 ×
8 based on split or non split binary consent. So, it’s a kind of binary classification
at various levels. Let’s discuss the CTU split, a 64 × 64 CTU can be regarded to 2
Fig. 1 Concept depth decision algorithm using CNN
levels, split/non split. The spilled one will be four, 32 × 32 cu’s. Out of this four,32
× 32 cu, each can be split/non split there are 42 ways to split. Furthermore, if we
split 43 split ways are there. Now let’s consider the available patterns by this split &
non split based on its depth. As it goes deeper to the CTU split for a 16 × 16 cu if we
consider it can have only 2 patterns available split/non split pattern. For 32 × 32 its
24 + 1 = 17 possible patterns and for the next level its 174 + 1 = 83,522 possible
patterns are available (Fig. 1).
By observing the depth value many information regarding CTU split can be
extracted. Compared to the time taken procedure in the calculation of depth in HEVC,
deep learning can train the system in a better way to evaluate it. As in RDO process, it
checks each time the cost, compares, and sends back, which consumes time. Instead,
it can be segregated into three levels and provide depth. Each level possesses various
depth and based on the depth value, the split and non-split can be decoded. For
example, if depth is 0 it is non split that is 64 × 64 matrices. If depth is 1 it will
possess CU of 32 × 32 in that D1 level will check it is split or non split by a binary.
classifier. In each level the process is repeated, the depth/level of classification
can be represented as [H d]3.d=0 for all d = 0, i.e., non split.
so, this case can be ignored and simplify to [H d]3.d=1 . Each depth can be split or
non- split.
H1 can be split/non split and need to be represented as H1(x) x = 0 or 1.
H2(x) have more information so H2(x) should be rewritten as
H2(xi) x = 0 or 1
i =0−3
H3(x i,j) with i,j = 0–3 and can be mathematically represented as
Yd =Hd(x) d=1
Hd(xi) d = 2, i = (0 − 3)
Hd(xi, j) d = 3, i, j = (0 − 3)
So, Yd can give the information about the depth of CU as split or non split. The
value gives information about depth and whether it’s split or not with a calculation
time.
If Yd = 0 it says that the depth d is non split and if Yd = 1 the depth d is split. As
we train the network with this information the calculation complexity reduces that
saves time too.
3.2 Deep Depth Decision Algorithm Using Deep Learning
The deep depth decision algorithm is based on convolutional neural network. Here
the input is CU of various size from 32 × 32, 16 × 16, and so on. The 64 × 64 CTU
is by default with depth 0 and is not considered, which saves calculation time. The
YUV input with 3 channels is given as input to CNN, the convolution layer convert
it to 5 × 5 followed by a max pooling to 2 × 2 with activation function as ReLU.
The primitive features are extracted with various kernels of size 5, 6, etc. by the deep
CNN layers and the training helps it to provide better results. The extracted features
are then combined and reduce by the fully connected network to 256 then to 64 and
to a vector of 16 elements that provide the information if total CTU split as shown
in Fig. 2.
The steps for this will have this work flow initial step is to build a data set, the data
set used here are YUV data set and it is available for research purpose. The selected
data set are preprocessed by splitting it into image frames with label. These frames
and labels are trained and validated using the deep CNN network. The model can be
improved by training it again until an optimum is reached. The Convolution process
in CNN layer can extract the fine features by using various kernels to achieve better
output.
Fig. 2 Deep depth decision algorithm with CNN

Fig. 3 Bit rate comparison
Fig. 4 Time for encoding

comparison to analyze
complexity
4 Experimental Results
The table shows sample input and its intraframe encoding time using HEVC and the
proposed deep depth decision algorithm. A noticeable dip in encoding time can be
observed here focusing on the last two columns in the table and the graph in Figs. 3
and 4. The average encoding time reduction is: 71.5%. This point says clearly that
the prediction of deep depth decision algorithm is fast and reduced the computational
complexity compared to existing technology.
Table 1 showing the input and its intraframe encoding time using HEVC and the
proposed deep depth decision algorithm.
5 Conclusion
The video and its compression are a matter of interest with the latest video acquisition
devices and the internet. The video processing and its compression were updating its
technology with the emergence of artificial intelligence, machine learning, NN, and
Table 1 Bitrate comparison table for various resolutions

YUV sequence with Bitrate HEVC Bitrate CNN Time-HEVC Time-Deep depth
resolution and frame decision CNN
rate
Keiba_416 × 240_30 539.8007 582.1690 1281.129 356.524
Mobile_352 × 288_20 1853.5339 1955.5392 1629.696 666.098
RaceHorses_832 × 2594.1504 2719.9232 5440.564 1513.008
480_30
ParkScene_1920 × 7361.9354 7702.8979 13,238.985 3591.274
1080_24
deep learning networks (DL). Intra perdition that is the initial step in video compres-
sion helps in a major part of compression in total video compression process. The
proposed method trains the deep CNN network with varities of dataset of different
dimensions and frame rates. The trained data is validated and tested. This network
reduces the computational complexity of intra prediction, if input CTU is given it
provides the depth information as a vector of length 16 that provides the total infor-
mation with less time. The encoding time is calculated and compared with HEVC
reference software. The results show 71.3% saving in encoding time. The results can
be extended by trying with other deep learning networks like ResNet and compare
results.
References
1. Sullivan GJ, Ohm J, Han W-J, Wiegand T (2012) Overview of the high efficiency video coding
(HEVC) standard. IEEE TCSVT 22(12):1649–1668
2. Correˆa G, Assuncao P, Agostini L, da Silva Cruz LA (2012) Performance and computational
complexity assessment of high efficiency video encoders. IEEE TCSVT 22(12):1899–1909
3. Shen X, Yu L, Chen J (2012) Fast coding unit size selection for HEVC based on Bayesian
decision rule. In: PCS
4. Kiho C, Jang ES (2012) Early TU decision method for fast video encoding in high efficiency
video coding. Electron Lett 48(12):689–691
5. Yoo HM, Suh JW (2013) Fast coding unit decision algorithm based on inter and intra prediction
unit termination for HEVC. In: ICCE
6. Miyazawa K, Murakami T, Minezawa A, Sakate H (2012) Complexity reduction of in- loop
filtering for compressed image restoration in HEVC. In: PCS
7. Peixoto E, Shanableh T, Izquierdo E (2014) H.264/AVC to HEVC video transcoder based on
dynamic thresholding and content modeling. IEEE TCSVT 24(1):99–112
8. Correˆa G, Assuncao PA, Agostini LV, da Silva Cruz LA (2015) Fast HEVC encoding decisions
using data mining. IEEE TCSVT 25(4):660–673
9. Vanne J, Viitanen M, Ham al ainen TD, Hallapuro A (2012) Comparative rate- distortion-
complexity analysis of hevc and avc video codecs. IEEE TCSVT 22(12):1885–1898
10. Kim N, Jeon S, Shim HJ, Jeon B, Lim SC, Ko H (2016) Adaptive keypoint-based CU depth
decision for HEVC intra coding. In: IEEE BMSB
11. Khan MUK, Shafique M, Henkel J (2013) An adaptive complexity reduction scheme with fast
prediction unit decision for HEVC intra encoding. In: IEEE ICIP
12. Hu Q, Shi Z, Zhang X, Gao Z (2016) Fast HEVC intra mode decision based on logistic
regression classification. In: IEEE BMSB
Investigation of Electrical Characteristics
for Different Geometrical Dimensions
of 11 kV Insulators Under Finite Element
Method
R. Bharanidharan and R. V. Maheswari
Abstract In the power system, the occurrence of a short circuit in transmission

lines is the major challenge in planning, designing, and operating with an excessive
high-voltage. The new model incorporates the actual geometrical two dimensions of
insulators, property of materials, and self-adaptive measurement of the components.
11-kV non-ceramic insulator designs are simulated using ANSYS software. The
insulators are free from contamination. Finite element method (FEM) is a numerical
method with 2–3 space variables to solve partial differential equations is employed to
identify the electrical stress issues in high-voltage fields. For all surface conditions,
the electric field distribution, maximum electrical field strength and its location,
voltage distribution, and heat generated on the surface are examined. The simulation
findings were introduced and analyzed for non-ceramic insulator designs.
Keywords 11 kV polymeric insulator · Finite element method · Electric field

distribution · Voltage distribution
1 Introduction
For the performance of an electric power system, the reliability and consistency
of the power system and equipment are very important. Because of damp weather
conditions such as dew fog or drizzle, the emission layer becomes conductive, and
a leakage current flows through it when combined with moisture [1, 4] Due to their
lighter weight, greater mechanical intensity, better anti-pollution performance, and
lower maintenance requirements, composite insulators are commonly used in the
power system. The distribution and potential of the electrical field can be distorted
by flashovers. Dry bands on the insulator surface [2] may form due to the flow of
the leakage current. Compared to ceramic insulators, the electrical field distribution
of the non-ceramic insulator is distinct. The electric field distribution of a composite
insulator is typically more nonlinear than that of a porcelain insulator [3, 7]. Numer-
ical methods such as the Boundary Element Method, the Charge Simulation Method,
R. Bharanidharan (B) · R. V. Maheswari

Department of EEE, National Engineering College, Kovilpatti, India
https://doi.org/10.1007/978-981-16-4943-1_2
12 R. Bharanidharan and R. V. Maheswari
the Finite Difference Method, and the Finite Element Method can assess the distri-
bution of the field. Some research has been carried to investigate the AC behavior
of hydrophilic and hydrophobic post-insulators [5] while it was raining. Axial and
radial elements of electrical field strength along an insulator were determined under
dry and rain conditions. FEM was used to conduct an investigation of stress control
on polymeric insulators [6]. The optimal installation of an arcing horn close to the
end fittings from the investigation has a reduction in the Electric field stress where it is
maximum and uniform potential distribution and Electric field for the High-voltage
end and earth end Arcing horn installed in a composite insulator. Temperature
vulcanizing (RTV) coated insulators usage was increasing in transmission lines [8].
An oversized spectrum of RTV coating damage is currently observed on square
measure insulators. Three standard modes of RTV coating damage were simulated,
so the characteristics of the electrical discharge emission were investigated. The elec-
tric fields of high-voltage insulators were analyzed. This paper involves designing
and applying multiple long rod insulators with E-Field analysis. Multiple materials
such as ceramic, porcelain, fiberglass, and silicon rubber [9] have been used to model
long rod insulators. The effects of the Corona ring on composite polymer insulator
strings insulation performance. For the strings, which include corona rings of the
respective design specifications, and also for the ring-free strings, experimental
investigations were carried out. Test results were discussed from the point of view
of discharge development physics and the strings were optimally configured [10].
2 Finite Element Method of Analysis
A simple way to test the distribution of the electric field is to calculate the electric
potential distribution initially. Calculation of electric field distribution is done by
subtracting the gradient’s electrical potential distribution from it and it is expressed
in Eq. (1).
Electric field = −∇u. (1)
where ‘u’ is the potential gradient. Equation (2) is obtained by applying Maxwell
equation where ρvol is the volume charge density and ‘ε’ is the permittivity of the
medium which is ε = ε0 εr where ‘ε0 ’ is air permittivity and ‘εr ’ is the relative
permittivity of dielectric material.
ρvol
∇(Electric field) = . (2)
ε
ρvol
∇2u = − . (3)
ε
With space charges ρvol = 0, Laplace equation is obtained, and it is expressed in
Eq. (4).
Investigation of Electrical Characteristics for Different … 13
∇ 2 u = −0. (4)
The Cartesian system of coordinates of two-dimensional function F(V ) is

expressed in Eq. (5).
¨ 2 2
1 dV dV
F(V ) = [εe + εf ded f . (5)
2 de df
where ‘εe ’ and ‘εf ’ are the components of the permittivity and the’ ‘e’ and ‘f ’
components of the permittivity of components in the coordinate system of the Carte-
sian system, and V is the electrical potential. Equation (6) is obtained by assuming
the isotropic permittivity distribution ε = εe = εf .

1
F(V ) = ε|∇V |2 ds. (6)
2
Equation (7) represents the effect of conductivity of the pollution layer is

considered on the distribution of the electric field then F(V ) is
¨
1
F(V ) = (σ + jωε)|∇V |2 ds
2 (7)
.
‘ω’ is the angular frequency and ‘σ ’ is the conductivity of pollution layer µs/cm
and ‘F(V )’ is the complex function. Assume the ‘c’ and ‘d’ are the linear depen-
dencies of each sub-domain of ‘V ’. it results in the first order approximately in
Eq. (8).
Va (c, d) = aa1 + aa2 c + aa3 d, (a = 1, 2, . . . , h). (8)
The measurement of the electrical potential at any node in the total network
composed of several triangular elements is performed by minimizing the F function
is obtained in Eq. (9). Figure 1 represents the flowchart of the finite element method
∂ F(r Vi )
= 0, i = 1, 2, . . . , h. (9)
∂ Vi
3 FEM Model of Polymeric Insulator
Polymeric insulators have four major components, they are

a) Fiber-reinforced polymer.
Fig. 1 Flow chart for the FEM
Table 1 Material properties

Properties PEC Polymer Air
of FEM model
Relative Permittivity, (εr ) 4.3 7.2 1
Conductivity, µ s/cm 1 4 0
b) Polymer sheath on the rod.

c) Polymer weather sheds.
d) Metals and fittings.
For the simulation study, an insulator made of polymeric materials was chosen
for the current operation dimensions of the 11 kV polymeric insulators used in the
simulation analysis, as shown in Table 1. The electric field and voltage distribution
(EFVD) of 11 kV polymeric insulators at clean and dry surface conditions are inves-
tigated in this paper. The relative permittivity, conductivity, and thickness or height
of the contaminated layer and dry band of the materials used in the FEM modeling
insulator. Metal fittings are installed at both the line and ground ends of the insulator.
Profiles for Test Specimens are shown in Table 2.
4 Simulation Results and Discussion
The potential calculation and distribution of the electric field (E-Field) along the
insulator have been constructed. The simulation work is done with the help of soft-
ware from the ANSYS ELECTROMAGNETIC suite. E-Field distributions for the
model were plotted by using 2D axisymmetric tools. Electric potential distributions
for the model are plotted by using 2D tools. The relation between the E-Field and
Table 2 Profiles for test specimens

Parameters Type 1 Type2 Type 3 Type 4 Type 5
Diameter (cm) 5.1 6.1 5.1 6.7 6.5
Total height (cm) 38 40.5 3 31 30
Creepage distance (cm) 34 39 34 30 28
Creepage factor 1.6 2.1 1.6 2.4 2.3
Profile factor 6.2 4.2 6.2 3.9 3.7
Form factor 2.1 2.0 1.9 1.4 1.2
Shed slope angle 28 22 28 20 20
leakage path, electric potential, and leakage path was plotted for each model. The
design, operation, and performance of the HV equipment influence the electric field
and electric potential distribution. E-Field and Electric potential distribution are the
Electrical performance of power equipment. To improve electrical performance,
various gases are used. Due to non-uniform electric field distribution, partial and
corona discharges occur in the power equipment.
4.1 Type 1 Insulator
Figure 2 shows the results of the Type 1 insulator 2D simulation, such as the distribu-
tion of electric potential, the relationship of E-Field distribution between the electric
field and the end-to-end arcing distance, the relationship between the electric poten-
tial and the end-to-end arcing distance. Figure 2a indicates that the insulator is 2D
axisymmetric. Figure 2b indicates that the 2D magnitude of the insulator’s distribu-
tion of electrical potential. Figure 2c indicates that the magnitude of the distribution
of the electric field is more localized around the final fitting. Figure 2d shows that
the electrical field distribution vector is more localized around the final fitting. The
relationship between the E-Field and the end-to-end arc distance is shown in Fig. 2e,
which shows that the field is increased at the end fitting and rises when the end-to-end
arc distance approaches the sharp point of the first shed. The relation between the
electric potential and the end-to-end arc distance is shown in Fig. 2f, which shows
that the potential is reduced from live end fitting to ground end fitting along the
insulator.
Figure 3 indicates the effects of the Type 2 insulator 2D simulation, such as the
distribution of electric potential, the relationship of E-Field distribution between the
electric field and the end-to-end arcing distance, and the relationship between the
Fig. 2 Results of the Type 1 Insulator, a 2D axisymmetric of Type 1 insulator, b Potential distri-
bution, c Magnitude of E-Field distribution, d Vector of E-Field distribution, e E-Field versus Arc
length, f Electric potential versus Arc length
Fig. 3 Results of the Type 2 insulator, a 2D axisymmetric of Type 2 insulator, b potential distri-
bution, c magnitude of E-Field distribution, d Vector of E-Field distribution, e E-Field versus arc
length, f electric potential versus arc length
electric potential and the end-to-end arcing distance. Figure 3a indicates that the
insulator is 2D axisymmetric. Figure 3b indicates the 2D magnitude of the distri-
bution of the insulator’s electrical potential. Figure 3c indicates that the magnitude
of distribution of the electric field is more intense around the end fitting. Figure 3d
shows that the electric field distribution vector is more intense around the end fitting.
The relationship between the E-Field and the end-to-end arc distance is shown in
Fig. 3e, which shows that the field is raised at the end fitting and rises when the
end-to-end arc distance approaches the sharp point of the first shed. The relation
between the electric potential and the end-to-end arc distance is shown in Fig. 3f,
which shows that the potential is reduced from live end fitting to ground end fitting
along the insulator.
Figure 4 indicates the effects of the Type 3 insulator 2D simulation, such as the
bution, c magnitude of E-Field distribution, d vector of E-Field distribution, e E-Field versus arc
Figure 5 indicates the results of the Type 4 insulator 2D simulation, such as the
Fig. 5 Results of the Type 4 Insulator, a 2D axisymmetric of Type 4 insulator, b potential distri-
Figure 6 indicates the results of the Type 5 insulator 2D simulation, such as the
5 Conclusion
In this work, E-Field distribution on composite insulators is nonlinear near the end
fitting when subjected to the highest magnitude. Electric Field and Electric potential
distribution are analyzed along the surface of insulators for 5 different configurations
of composite insulators. Type 1 insulators having 3 alternate sheds with ball end
fitting shows lower stress compare to others along with higher profile factor, form
factor, and dry arcing resistance with solid layer coastal region pollution experiments.
Pollution flashover voltages are measured for five kinds of specimen, and it is found
the pollution flashover voltage reduces with increasing salt deposit density. The
maximum E-field stress is significantly reduced when an arcing horn is installed
near the end fittings, according to the analysis.
Acknowledgments The Department of Science and Technology fund for improving S&T infras-
tructure in universities and higher educational institutions is funding this work (DST-FIST)
(SR/FST/College-061/2017) Grant ID and The authors are also grateful to the National Engineering
College, Kovilpatti, Tamilnadu, India, for their support.
References
1. Walyo (2007) Study on leakage current waveforms of porcelain insulator due to various artificial
pollutants. Int J Math Phys Eng Sci 2(1):27–32
2. Andrew J (2008) Electric fields on ac composite transmission line insulators. IEEE Trans Power
Delivery 23(2):823–830
3. Silvester PP, Ferrari RL. Finite elements for electrical engineers, 3rd edn. Cambridge University
Press, Cambridge
4. Chandrasekhar S, Sarathi R, Danikas M (2006) Analysis of surface degradation of silicone
rubber insulation due to tracking under different voltage profiles. Int J Electrical Eng (Springer)
5. Hartings R (1994) Electric fields along a post insulator: AC-measurements and calculations.
IEEE Trans Power Delivery 9(2):912–918
6. Natrajan M, Basharan V, Ganapathyapillai K, Velayutham MR, Mariyasilluvairaj W (2015)
Analysis of stress control on 33-kV Non-ceramic insulators using Finite element method.
Electric Power Components Syst 43(5):566–577
7. Muniraj C, Chandrasekhar S (2012) Finite element modeling for electric field and voltage
distribution along the insulators. World J Modelling Simul 8(4):310–320
8. Zhang Z, Liu X, Jiang X, Hu J, Gao DW (2013) Study on AC flashover performance for different
types of porcelain and glass insulators with non-uniform pollution. IEEE Trans Power Deliv
28(3):1691–1698. https://doi.org/10.1109/tpwrd.2013.2245153
9. Jiang X, Yuan J, Zhang Z, Hu J, Sun C (2007) Study on AC artificial-contaminated flashover
performance of various types of insulators. IEEE Trans Power Delivery 22(4):2567–2574
10. Ilhan S, Ozdemir A (2015) Impacts of corona rings on the insulation performance of composite
polymer insulator strings. IEEE Trans Dielectr Electr Insul 22(3):1605–1612
Non-isolated Multilevel Zeta Converter
for MLI Application
Marikannu Marimuthu, Subramanian Vijayalakshmi, B. Paranthagan,

R. Venugopal, S. Srinithi, B. Yuvaraj, R. Soundarajan,
and S. K. Vasantha Kumar
Abstract A new multilevel zeta converter which converts fixed DC voltage to multi-
level DC output voltage is discussed in this paper. Proposed converter has stable
voltage feedback capacity and produces high gain output voltages with less input
current. Voltage supplied by the PV panel or the fuel cell is at an output of a low
voltage. These output voltages can be interfaced with standalone (or) grid connected
inverter system by employing the proposed converter. By using a single transistor
with the multilevel capacitor geometric structure, the proposed converter is able to
generate an output voltage that is much higher and has ripple free output current with
a higher step-up conversion ratio. It allows for operation at much higher frequencies
for a much longer period of time. Additionally, it does not require for the electrical
transformer for high gain. The output of the zeta converter generates four output
level voltages which are simulated and verified with theoretical results.
Keywords Zeta converter · High voltage gain · Multilevel voltages · Multilevel

Inverter · Self-balancing capacitor
M. Marimuthu (B) · S. Vijayalakshmi

Department of Electrical and Electronics Engineering, Saranathan College of Engineering, Trichy,
India
B. Paranthagan
Associate Professor, Department of Electrical and Electronics Engineering, Saranathan College of
Engineering, Trichy, India
e-mail: [email protected]
R. Venugopal
Assistant Professor, Department of Electrical and Electronics Engineering, Saranathan College of
S. Srinithi
PG Scholar, Department of Electrical and Electronics Engineering, Saranathan College of
B. Yuvaraj · R. Soundarajan · S. K. Vasantha Kumar
UG Student, Department of Electrical and Electronics Engineering, Saranathan College of
https://doi.org/10.1007/978-981-16-4943-1_3
22 M. Marimuthu et al.
1 Introduction
Transportation sector using electric motors is growing exponentially which may be

a less contributor to the greenhouse gas emission when compared to conventional
transportation system. This reduction is seen as important because it can slow the
warming of the planet by lowering pollution and bringing the global temperature
closer to normal [1–5]. The burning of petroleum products like petrol, diesel, CNG,
etc., is the key reason why we are experiencing a rise of in air pollution due to transport
sector, that is why there is some need to modernize the design of the electric vehicles.
Electric vehicles will be zero-carbon emission vehicles which will result in no carbon
pollution.
Hybrid electric vehicles (plug-in hybrid electric vehicles) are becoming more
popular, and more research is being done to develop an efficient, durable, and highly
reliable hybrid energy storage scheme with immense power density, a lower price-
to-weight ratio, and a longer life cycle. The three phase power electronic converters
(PEC) have a significant role to play in regulating the amount of current flowing
through an electric hybrid system (HISS). As of today, depending on the type of
vehicle that they use energy sources like solar, fuel cells, and ultra-capacitors, all
of which are renewable in nature. For utilizing low voltage obtained from renew-
able sources, the proposed converter will be useful for production of voltages with
high gain. Figure 1 shows the overall block diagram of the applications involving
multilevel zeta converter.
The proposed converter provides many advantages, such as: (i) a huge conversion
ratio with much less duty cycle (ii) increased voltage feedback stability (iii) higher
performance and (vi) low EMI noise.
The ultimate goal is to combine the advantages of a zeta converter and a multistage
geometric capacitor structure. This combination makes it possible to increase the
FUEL
SOURCE
(DC)
PV AC LOAD
DC-DC MULTILEVEL MULTILEVEL
SOURCE
ZETA CONVERTER INVERTER
(DC)
SUPER
CAPACITOR
DC LOAD
(DC)
Fig. 1 Application of high gain dc-dc converters

Non-isolated Multilevel Zeta Converter … 23
levels by simply inserting a diode and a capacitor. The unique aspect of this proposed
zeta is achieving the merits of zeta converter along with high voltage gain [6–8].
2 Design and Operation of Multilevel Zeta Converter
The ultimate idea of this converter is to reduce the number of switches and increase
the output voltage gain. In multilevel zeta converter MZC, the circuit is built with
single switch only [9–12]. The circuit diagram of the proposed generalized multilevel
zeta converter for n levels is shown in Fig. 2.
The duty cycle D for a zeta converter is given by,
Vout
D= (1)
Vout − Vin
Vin D
Vout = (2)
1− D
The inductor L 1 and L 2 are designed based on the following equation by assuming
appropriate duty ratio and output power,
Vin D
L1 = (3)
FIL1
D
Cn
D3 LOAD
C2
D2
Ca
D1
S
C L2
Vin
D C1
L1
Fig. 2 Generalized structure of MZC

Vin D
L2 = (4)
FIL2
The capacitor value can be obtained by using following expression,
Vout D
C1 = (5)
F RVC1
Vin D
C2 = (6)
8F L 2 VC2
2
where F = Switching Frequency.

The multilevel zeta converter with four output levels is shown in Fig. 3.
Operates in various modes which are explained as follows,
Mode 1: The Switch is closed.
When the switch is in the on state, the inductor L 1 is connected to the input
voltage and the inductor which is charged in pervious cycle discharges through D2 .
As a result, capacitor C 1 is charged which is shown in Fig. 4a.
Mode 2: The switch is closed.
In mode 2 operation, capacitor voltage C b is clamped to C 2 with the help of diode
D4 which is shown if Fig. 4b.
Mode 3: The Switch is closed.
Fig. 3 MZC circuit with

four multilevel outputs D8
C4
D7
Cd
D6
C3
D5
Cc
D4
C2
D3
Cb
D2
S
Ca L2
Vin
D1 C1
L1
D8 D8
C4 C4
D7 D7
Cd Cd
D6 D6
C3 C3
D5 D5
Cc Cc
D4 D4
C2 C2
D3 D3
Cb Cb
D2 D2
S S
Ca L2 Ca L2
Vin Vin
D1 C1 D1 C1
L1 L1
(a) (b)
D8 D8
C4 C4
D7 D7
Cd Cd
D6 D6
C3 C3
D5 D5
Cc Cc
D4 D4
C2 C2
D3 D3
Cb Cb
D2 D2
S S
Ca L2 Ca L2
Vin Vin
D1 C1 C1
L1 L1 D1
(c) (d)
Fig. 4 a Mode-1, b Mode-2, c Mode-3, d Mode-4, e Mode-5, f Mode-6, g Mode-7, h Mode-8
D8
D8
C4
C4 D7
D7
Cd
Cd
D6
D6 C3
C3
D5
D5
Cc
Cc
D4
D4
C2
C2
D3
D3
Cb
Cb
D2 D2
S S
Ca L2 Ca L2
Vin Vin
C1 D1 C1
L1 D1 L1
(e) (f)
D8
D8 C4
D7
C4
D7 Cd
Cd
D6
C3
D6
C3
D5
D5
Cc
Cc D4
D4 C2
C2
D3
D3
Cb
Cb
D2
D2
S
S Ca L2
Ca L2
Vin Vin
C1 D1 C1
L1 D1 L1
(g) (h)
Fig. 4 (continued)
When the voltage across V cb and V cc is greater than V c2 and V c3 , then capacitor
voltage across C b is clamped to C 2 , and capacitor voltage across C c is clamped to
capacitor voltage C 3 through D6 which is shown in Fig. 4c.
Mode 4: switch is closed.
When the circuit is in the clamped mode, if the voltage across the C b , C c , and C d
are greater than the voltage across C 2 , C 3 , and C 4 , the voltages across C b , C c , and
C d are clamped to those values of those voltages across C 2 , C 3 , and C 4 through the
diode D8 and the key switches S. It can be seen in Fig. 4d.
Mode 5: The switch is open.
During mode 5, Switch S is open condition, and diode D1 is operates. Inductor
L 1 discharges through D1 which is shown in Fig. 4e.
When voltage across C 1 is greater than C b , then voltage between C 1 is clamped
to voltage between C b which is shown in Fig. 4f.
In mode 7, voltage across C 1 and C 2 is greater than C b and C c , and then, the
voltage between C 1 and C 2 is clamped to voltage between C b and C c with the help
of diode D5 is shown in Fig. 4g.
In mode 8, the diode D5 turns ON, hence voltages of C 1 , C 2 , C 3 are greater than
C b , C c , C d . The voltage across capacitors C 1 , C 2 , C 3 are clamped to voltage across
the capacitors C b , C c , C d is shown in Fig. 4h.
The simulation has been implemented in MATLAB/Simulink environment. Figure 5

shows the output voltage obtained for the proposed four-level zeta converter. The
Fig. 5 Load voltage obtained for the proposed MZC

Table 1 Design parameters

Parameter Value
of proposed zeta converter
Input voltage (V in ) 12 V
Switching frequency 5 kHz
Inductor L 1 1.155 mH
Inductor L 2 1.155 mH
Capacitor C a 1.846 mF
Self-balancing capacitor (C b , C c , C d , C 1 , C 2 , C 3 ) 1.731 mF
Duty ratio (D) 60%
converter is mainly used as a conventional zeta converter, yielding a 21 V output

voltage at the first stage. By adding the components to expand the level further, at
each level, the converter showed a considerable step-up in the output voltage like
42 V, 64 V, and 86 V, respectively.
The multilevel voltages convergence hold good for all capacitors in the range
to be controlled to a standard voltage despite any changes in the source loads or
the existence of voltage changes introduced on the board. The parameter values are
shown in Table 1.
Finally, when the multilevel converter is powered using zeta, it generates a desired
output voltages having less voltage overshoot. In addition, the output has less electro-
magnetic noise and self-oscillation. The performance parameters including faster
settling time and no steady-state error have been greatly improved. Many experiments
have been performed to check the supremacy of the zeta is well on the switching
element.
4 Application of Multilevel Zeta Converter to Multilevel

Inverter Operation
The proposed multilevel zeta converter is used to act as input voltage supply for
H-bridge using level circuit as shown in Fig. 6. An H-bridge with level circuit combi-
nation generates 9 degree of AC voltage with a frequency of 50 Hz. The nine level
MLI output voltages are 0 V, +21 V, +42 V, +64 V, +86 V, −21 V, −42 V, −64 V,
−86 V, are shown in Fig. 7. Since these AC waves have less harmonics, they can
be used for medium-power applications like computers, telecom power supplies in
remote areas, electric vehicles, and uninterruptible power supplies.
S4
D8
C4 S3
D7 Ds3
S11 S31
Cd
D6
C3 LOAD Vo, max
S2
D5
Ds2 S21
S41
Cc
D4
C2
S1
D3
Ds1
Cb
D2
S
Ca L2
Vin
D1 C1
L1
Multilevel Zeta converter Lelel Circuit H-Bridge Inverter
Fig. 6 Multilevel zeta converter with H-bridge inverter
Fig. 7 Output voltage waveform of H-bridge inverter
5 Conclusion
When comparing the output voltage of the single-level zeta converter to that of the
multilevel one, it is easier to understand that the number of levels is increased that the
maximum output voltage 86 V with 12 V input. The constant input current, the higher
degree of output voltage gain obtained by assuming a much shorter duty cycle, the
lower current stress, and the necessity of fewer parts, all without the use of a trans-
former, are all “special features” of this multilevel zeta converter. The most important
advantage is that it is able to raise multilevel output voltages without interruption
of the primary elementary circuit. Transistors that are capable of overcoming these
electric vehicle problems go a long way in passing the slew of test results.
References
1. Gaurav P, Dhote VP, Modak. Study and analysis of zeta converter fed by solar photovoltaic
system using PID controller
2. Manikandan K, Sivabalan A, Sundar R, Sury P. A study of landsman, sepic and zeta converter
by particle swarm optimization technique. 978-1-7281-5197-7/20/$31.00 ©2020 IEEE
3. Suresh P, Kirubakaran D (2018) Novel zeta converter with multi level inverter connected to
grid. In: IJEECS, vol 11, no 3, September 2018, pp 814–820
4. Falin J (2010) Designing DC/DC converters based on ZETA topology. Analog Appl J
5. Marimuthu M, Vijayalakshmi S, Shenbagalakshmi R (2020) A novel non-isolated single switch
multilevel cascaded DC-DC boost converter for multilevel inverter application. J Electrical Eng
Technol. The Korean Institute of Electrical Engineers 2020
6. Al Gabri AM, Fardoun AA, Ismail EH (2015) Bridgeless PFC modified SEPIC rectifier with
extended gain for universal input voltage applications. IEEE Trans Power Electron 30(8):4272–
4282
7. Vuthchhay E, Bunlaksananusorn, Hirata H (2008) Dynamic modeling and control of a zeta
converter. In: ECTI-CON 2008, vol 2, 1417 May 2008, pp 969–972
8. Palma L, Todorovic MH, Enjeti P (2005) A high gain transformer less DC-DC Converter for
fuel cell applications. In: IEEE Power Electron Jun 2005, p 2514
9. Padmanaban S, Grandi G, Wheeler PW, Blaabjerg F, Loncarski J (2015) A simple MPPT
algorithm for novel PV power generation system by high output voltage DC-DC boost converter.
In: IEEE (ISIE), pp 214–220
10. Rosas-Caro JC, Sanchez VM, Vazquez-Bautista RF, Morales-Mendoza LJ, Mayo-Maldonado
JC, Garcia-Vite PM, Barbosa R(2016) A novel DC-DC multilevel SEPIC converter for PEMFC
systems. Int J Hydrogen Energy
11. Alishah RS, Hosseini SH, Babaei E, Sabahi M (2017) Optimal design of new cascaded switch-
ladder multilevel inverter structure. IEEE Trans Ind Electron 64(3):2072–2080. https://doi.org/
10.1109/TIE.2016.2627019
12. Taghvae A, Adabi J, Renanejad M (2017) A multilevel Inverter structure based on a combi-
nation of switched capacitor and DC sources. IEEE Tras. Ind. Informatics 13(5):2162–2171.
https://doi.org/10.1109/TII.2017.2710265
Corrupted Image Enhancement Through
WaveNet: A Hybrid Approach
C. Vimala and P. Aruna Priya
Abstract A denoising method for medical images through hybrid technique is

presented in this paper. The hybrid technique is the combination of wavelet and
neural network (WaveNet). Proposed algorithm has been validated through bench-
mark image, and medical images both are degraded by the variety of noise density
through Gaussian noise and visual property. The performances of denoised images
are also analyzed with wavelet techniques and compare the results with proposed
technique. The proposed strategy is developed in MATLAB platform. Simulation
results are evidence for the proposed work
Keywords Wavelet · Neural network · Denoising · Enhancement
1 Introduction
A medical image is corrupted by different reasons, and the removal of the noise is
extremely important for correct diagnosis. The process of removing noise from the
image is called image denoising. Denoising is basically classified into two types one
is filtered method and the other one is transform method. This paper deals with one
of the transform methods like wavelet transform [1]. The wavelet transform-based
denoising method is concentrated more in the literature. Continuous and discrete
are the two different wavelet approaches. The developed version of discrete wavelet
transform is the higher density wavelet transform. The higher density wavelet trans-
form that includes double density DWT, double density dual-tree DWT-based image
denoising technique will provide good quality of the image. The image quality perfor-
mance measure values such as peak signal to noise ratio (PSNR), mean square error
(MSE), and root mean square error (RMSE) are used to identify the best denoising
methods [2]. Digital images are affected by various types of noise. Noise is an
C. Vimala (B) · P. Aruna Priya

Department of Electronics and Communication Engineering, College of Engineering and
Technology, SRM Institute of Science and Technology, Kattankulathur, India
P. Aruna Priya
https://doi.org/10.1007/978-981-16-4943-1_4
32 C. Vimala and P. Aruna Priya
unwanted component, which affects the image. One of the well-known noise that
affect the image is additive white Gaussian noise (AWGN), in which the noise prob-
ability density function is bell-shaped, and it is uniformly distributed. The additive
noise model is represented in Eq. (1) [3].
x(m, n) = s(m, n) + η(m, n) (1)
Removing additive white Gaussian noise (AWGN) from a corrupted signal or

image is very interesting. The wavelet-based denoising techniques are mostly imple-
mented for the removal of AWGN [4]. The random variable for Gaussian y is normal-
ized to have zero mean, and variance value is one. The mathematical expression is
written in Eq. (2) as shown below

1 y2
f y (y) = √ exp − (2)
2π 2
Therefore, the probability distribution function of random variable Gaussian z of

a signal is given in Eq. (3)

1 (z − μ)
p(z) = √ exp − (3)
2π 2σ 2
where z represents the gray level, μ represents mean value, and σ represents
the standard deviation (SD) value. The square of the SD (σ 2 ) represents variance.
The Gaussian noise is normalized to have mean zero and variance one. The standard
normal distribution of Gaussian noise is given by N(0, 1), i.e., N(μ, σ 2 ) it is shown in
Fig. 1. Also, the power spectrum of the noise is constant at all frequencies [5]. Image
denoising is one of the vital essentials in image processing applications. The degraded
images are not useful for any analysis. The researchers have used various methods
to improve the degraded images, with appropriate analysis. While implementing the
denoising techniques, the prior knowledge on the properties of the distorted images
and the noises is assumed [6]. Figure 2 denotes the concept of denoising s(x, y) is
the image subjected to blurring through a linear operation. n(x, y) is the noise signal
Fig. 1 Normal distribution

of Gaussian noise
Corrupted Image Enhancement Through WaveNet … 33
Fig. 2 Denoising process
nique, z(x, y) a restored image is obtained. “Linear operation” considered here can
add or multiply the noises with the original image [3].
2 Higher Density Dual-Tree DWT
The dual-tree DWT and higher density will provide higher density. This dual-tree
at one level of computation has six dilation and translation coefficients [7, 8]. In
this method, the DDDWT technique is applied to the dual-tree DWT. The scaling
and wavelet functions are defined absolutely through the dilation and translation
it is represented in Eq. (4) to Eq. (6 [9, 10]). The system model for denoising in
DDDTDWT is illustrated in Fig. 3.
3 Proposed Denosing Method
Hybrid (wavelet + Neural Network = WaveNet) image denoising technique of

WaveNet network. The WaveNet is a combination of both wavelet and neural
network. Wavelet is a well-known denoising algorithm for image processing along
with this combined neural network, so it is very useful for noise reduction (Fig. 4).
The performance of the hybrid technique is compared with the wavelet-based
denoising method. The various noise level of noisy and denoised PSNR values are
given in Table 1. With the low-noise level taken as 10, the PSNR improves by 30% by
the wavelet method and hybrid method improves PSNR by 35.92%. If the sigma value
is 25, the percentage improvement for hybrid method is 50.11%, but wavelet method
provides only 41.56%. Table 2 represents the MSE performance of hybrid method,
which is compared with DDDTDWT method. The 85.03% of the noise is removed
from noisy image by DDDTDWT method, and by the improved hybrid method,
88.92% of the noise is removed from noisy image. The best training performance
of hybrid method is shown in Fig. 5. The performance value calculated using the
formula Eq. (7)–(9) [11–13].
ψh,i (t), ψg ,i (t), i = 1, 2 (4)

50
100
150
200
250
300
350
400
450
500
100 200 300 400 500
50
100
150
200
250
300
350
400
450
500
100 200 300 400 500
Fig. 3 System model for DDDTDWT denoising method
Fig. 4 Neural network

structure
Table 1 Comparison of performance measure for noisy and denoised images using DDDTDWT
for high noise = 50
Image MSE RMSE PSNR(dB)
corrupted Enhanced corrupted Enhanced corrupted Enhanced
Lena 401.8993 89.2713 20.0474 9.4483 28.1442 34.6783
zelda 398.2987 63.6774 20.0316 7.3819 22.0965 30.7674
Mandrill 400.3713 291.5080 20.0093 7.0736 22.1062 23.4843
Peppers 225.8456 60.8188 15.0282 7.7986 24.5927 31.9732
Table 2 PSNR performance

Noisy DDDWT DDDTDWT DDDTDWT + ANN
comparison
10 29.55 31.7 33.56
15 24.65 30.8 31.37
20 22.13 29.81 30.02
25 20.58 28.81 29.97
Fig. 5 Training performance Best Training Performance is 0.00539 at epoch 16

using hybrid method 102
Train
Mean Squared Error (mse)
Best
Goal
100
-2
10
-4
10
0 2 4 6 8 10 12 14 16
16 Epochs
h,1 (t) ≈ h,2 (t − 0.5) (5)
g,1 (t) ≈ g,2 (t − 0.5) (6)
The test images are selected and added with Gaussian white noise with zero
mean and different noise variance. The quality of the denoised image is measured,
and the values are also tabulated, and the values are compared with DWT and
DDDTDWT. Due to more number of detailed coefficients, the DDDTDWT achieves
high PSNR. The comparisons are made for a different noise variance, where the
denoising methods such as DWT, DDDWT, and DDDTDWT are compared, and the
results show that DDDTDWT performs better than the other two methods.

Max
PSNR = 20 log10 (7)
RMSE(x, y)

N
1
RMSE = (xi − yi )2 (8)
N i=1
1
N
MSE = (xi − yi )2 (9)
N i=1
4 Result and Discussion
The proposed system is for medical images like uterus, tooth, lung, and brain.
The denoised PSNR value shown in Table 2 and the values are compared with
various wavelet denoising algorithms and evident that proposed techniques gives
the commendable PSNR values than the existing techniques. The MSE values of
proposed and existing techniques are calculated, and the values are given in Table
3. The minimum MSE values yield the proposed system. The minimum MSE is one
type of evidence to prove the proposed system. The same measurements are taken for
four different noise density values. For evaluation purpose, noisy image with constant
noise level and the same image size is used for both existing and proposed. It is seen
that the value of MSE is decreased when the noise variance low and increases when
the noise variance is high. The proposed method gives the less MSE, and it is given
in Fig. 6. The proposed ANN is incorporated with all the above-mentioned noisy
medical images through proposed method, and the corresponding denoised MSE
value is shown in Table 3, and the improved PSNR value is shown in Table 2.
Table 3 Noisy and denoised MSE, RMSE value comparison

σ Corrupted image Enhanced image MSE
DWT DDDWT DDDTDWT DDDTDWT + ANN
10 100.631 65 50.763 60.16 44.54
15 224.580 90 60.18 50.21 48.02
20 398.862 124 77.809 65.23 52.48
25 626.233 250 123.786 100.56 88.35
σ=25
700
600
500
400
MSE
300
200
100
0
Noisy DWT DDDWT DDDTDWT Proposed
Fig. 6 Graphical representation of enhanced image MSE value
5 Conclusion
The hybrid technique, which is the combination of wavelet and neural network
(WaveNet), has been implemented for denoising medical images, and the perfor-
mance measures such as PSNR, MSE, and RMSE are computed. Four test medical
images have been analyzed, and denoised images are applied to image quality assess-
ment which shows improvement in terms of the PSNR increase and the mean square
error and RMSE decrease. The obtained results of the proposed method have been
compared with results from other wavelet techniques that confirms that the proposed
method outperforms the on hand denoising methods.
References
1. Guo L, Chen W, Liao Y, Liao h, Li J (2016) An edge-preserved image denoising algorithm

based on local adaptive regularization. J Sensors Article ID 2019569, 6 pages. https://doi.org/
10.1155/2016/2019569
2. Liu XW, Huang LH (2014) A new nonlocal total variation regularization algorithm for image
denoising. Math Comput Simul 97(3):224–233
3. Kittisuwan P (2016) Image enhancement via MMSE estimation of Gaussian scale mixture with
Maxwell density in AWGN. J Innov Optical Health Sci 9(2)
4. Ding Y, Ivan WS (2015) Artifact-free wavelet denoising: Non-convex sparse regularization
convex optimization. IEEE Signal Process Lett 22:1364–1368
5. Vimala C, Aruna Priya P (2019) Artificial neural network based wavelet transform technique
for image quality enhancement. Comput Electrical Eng 76:258–267
6. Aravind N, Suresh KV (2015) A discontinuity adaptive prior for image denoising. Int J Comput
Appl 110:13–19
7. Xu J, Feng A, Hao Y, Zhang X, Han Y (2016) Image deblurring and denoising by an improved
variational model. Int J Electronics Commun 70:1128–1133
8. Yan J, Lu WS (2015) Imaging denoising by generalized total variation regularization and least
squares fidelity. Multidimension Syst Signal Process 20(1):89–97
9. Wang X, Ou X, Chen B-W, Kim M (2015) Image denoising based on improved wavelet
threshold function for wireless camera networks and transmissions. Int J Distributed Sensor
Netw. https://doi.org/10.1155/2015/670216
10. Jaiswala A, Upadhyayb J, Somkuwar A (2014) Image denoising and quality measurements by
using filtering and wavelet based techniques. Int J Electronics Commun 68:699–705
11. Chinnarao B, Madhavilatha M, Pratap NL (2012) Improved image denoising algorithm using
dual tree complex wavelet transform with new thresholding technique. Int J Electronics Eng
4(1):1359
12. Hedaoo P, Swati S, Godbole (2011) Wavelet thresholding approach for image denoising. Int J
Netw Security Appl 3(4):16–21
13. Vimala C, Aruna Priya P (2015) Noise reduction based on double density discrete wavelet
transform. In: IEEE international conference on smart structures and systems, pp 15–18
Modeling and Control of Cascaded
Multilevel Inverter for Harmonics
Mitigation of Induction Motor Drive
Ranganathan Selvarasu, C. Kannan, S. Priyadharsini,

and Dagne Alemayehu Shiferaw
Abstract Nowadays the majority of the industrial drives are employed with induc-
tion motors as a load. Induction motor receives input from the output of MLIs,
which requires quality power in terms of lower or negligible total harmonic distor-
tion (THD). In order to provide quality power, in this paper, modeling of cascaded
multilevel inverter (MLI) for harmonics mitigation of induction motor is proposed.
In recent years, cascaded H-bridge MLI technology has emerged as a crucial to
offer a quality output for industrial drives. In this work, three-phase five- and
seven-level inverters are developed, modeled and simulation carried out using
MATLAB/Simulink. Comparisons of simulation as well as prototype solutions are
presented.
Keywords Induction motor · Total harmonic distortion · PWM technique ·

Cascaded H-bridge multilevel
1 Introduction
In present days, almost ninety percentages of industrial drives are employed with
induction motors due to the advantages offered by them, which includes rugged and
simple in construction, less maintenance, reliable, offers high efficiency, and cost is
less. The variable speed of drive is required in many applications, which is accom-
plished by induction motor. Nowadays due to the development of semiconductor
devices, power electronic devices are playing a vital role for controlling speed of
induction motor. In most of the drives, the induction motor speed is controlled by
output of voltage source inverters.
The output voltage level of inverters is improved by altering structure of MLI
topologies. The conventional inverter offers output voltage with huge total harmonic
R. Selvarasu · D. A. Shiferaw
Adama Science and Technology University, Adama, Ethiopia
C. Kannan (B) · S. Priyadharsini
Arunai Engineering College, Tiruvannamalai, India
https://doi.org/10.1007/978-981-16-4943-1_5
40 R. Selvarasu et al.
distortion. A multilevel inverter offers output voltages with less total harmonic distor-
tion. A survey has been carried out on multilevel inverter topologies, control, and
their applications [1].
A tutorial on multilevel converter has been presented, which includes operating
principle of inverter, various topologies for power circuit, methods of modulation,
technical issues, and applications at industry [2]. Single- and double-source module
for multilevel inverter structure has been proposed to decrease the power electronic
switches, and the results have been compared with different topologies [3]. A triple
inverter topology has been proposed with multilevel operation. Each inverter has been
fed by three isolated PV generators. The simulation and experiment results confirmed
the performance of proposed topology of inverter [4]. PI and fuzzy controller have
been proposed to develop cascaded H-bridge MLI, which produces output voltages
with five levels of reduced total harmonic distortion and the results compared [5]. A
new multilevel electromagnetic interference filter has been developed to suppress
mutual interference. The resistance to thermal and electromagnetic interference
has been improved by the designed filter [6]. The grid-connected PV system relia-
bility has been improved by designing a 27-level cascaded H-bridge MLI. Modified
synchronous optimal PWM technique has used to obtain the minimal THD [7]. Six-
phase induction generator has been modeled for wind energy conversion system,
which uses three-level converters on grid side [8]. Cascaded H-bridge MLI with
modified topology has been proposed to provide a staircase shape of DC voltage to
the bridge inverter, which in turn offers AC voltage with minimal THD with power
loss reduction [9].
A cascaded H-bridge MLI with five levels has been modeled for wind mill applica-
tions, which provides reduced total harmonic distortion [10]. A complete review has
been presented for newly proposed different MLI topologies [11]. In order to reduce
switches and losses, a new tapped source stack succored MLI has been developed
[12].
In this paper, three new types of cascaded H-bridge MLI are proposed with multi-
carrier pulse width modulation. They produce two levels, five levels, and seven
levels output voltages. The proposed topologies are modeled and simulated with
MATLAB/Simulink environment. In addition to that, prototype model is devel-
oped to prove the efficacy of proposed topology. Simulation and prototype result
are presented and compared.
2 Basic Structure of Inverter-Fed Induction Motor
Induction motor is supplied from DC supply using proposed five levels and seven
levels cascaded H-bridge MLI in order to lessen the THD. The multicarrier PWM
scheme is implemented to get the gating signal to control the MOSFET gate. Figure 1
represents the block diagram for induction motor fed by MLI.
Modeling and Control of Cascaded Multilevel Inverter … 41
DC Cascaded
POWER H-Bridge
Multilevel Inverter IM
SUPPLY
Pulse Generator
Fig. 1 Block diagram of induction motor fed by inveter
3 Proposed Modeling of Inverter
Cascaded H-bridge MLI is one of the conventional MLI which possesses numerous
advantages than other types. The foremost merit of cascaded MLI structure is modular
in nature where the entire system that contains identical cell is grouped. However, it
requires isolated DC supply of (2k + 1) voltage level. In this work, three different
structures are developed and modeled.
3.1 Modeling of Two Levels Inverter with Induction Motor
A three-phase two-level inverter is modeled, which comprises six groups of active

switches, S1 to S6 . The MATLAB/Simulink model of two-level inverter is shown in
Fig. 2.
Fig. 2 Simulink model of two-level inverter-fed induction motor

Simulink model of two levels inverter is developed by considering induction motor

as a load with rating of 50 HP, 400 V, 1500 rpm. Multicarrier pulsed width modulation
is adopted for the developed system. All the carrier waves are formed above zero
and arranged in phase with 180°. The stator of induction motor is fed with output of
two levels inverter. Three-phase source and load torque are maintained as possible
constant. Simulation is performed in phasor mode.
3.2 Modeling of Five Levels Inverter with Induction Motor
A three-phase five-level inverter is modeled. Figure 3 indicates Simulink model of

five-level inverter. Induction motor is considered as a load for developed Simulink
model of five levels inverter.
Fig. 3 Simulink model of five-level inverter with induction motor

Fig. 4 Simulink model of seven-level inverter with induction motor
3.3 Modeling of Seven-Level Cascaded Inverter

with Induction Motor
Figure 4 shows Simulink model of a developed three-phase seven-level inverter-fed

induction motor.
4 Results and Discussion
In this work, two, five, and seven levels cascaded H-bridge inverters are modeled. The
modeled inverters are fed with induction motor, which is simulated using MATLAB
software. The voltage and total harmonic distortion of two-level inverter are presented
in Figs. 5 and 6, respectively. Figure 6 shows the two-level inverter harmonic spectrum
analyses with THD of 40.48c/o for modulation Index 1.
Figure 7 presents the simulation and hardware result of phase voltage of five-level
inverter. Figure 8 represents the five-level inverter harmonic spectrum analysis with
THD 35.77c/o for modulation Index 1. Figure 9 indicates speed of induction motor
offered by five-level inverter output.
Figure 10 presents simulation and hardware result of phase voltage of seven-level
inverter. Simulation and hardware result of FFT analysis of seven levels inverter are
Fig. 5 Simulation and hardware result of phase voltage of two-level inverter
Fig. 6 FFT analysis of two-level inverter offered by simulation and prototype
Fig. 7 Simulation and hardware result of phase voltage of five-level inverter
Fig. 8 Simulation and hardware result of FFT analysis of five-level inverter

Fig. 9 Induction motor speed obtained by five-level inverter output
Fig. 10 Simulation and hardware result of phase voltage of seven-level inverter
presented in Fig. 11. In this case, the THD is 11.89c/o for modulation Index 1. It is
clear from the figures that THD in output voltage of seven levels inverter is lower
than the harmonic distortion in output voltage of five levels inverter. Figure 12 shows
the speed of induction motor offered by seven-level inverter output.
Table 1 provides the THD offered by two, five, and seven levels inverter for
various modulation index. It is possible to see from the table that seven-level inverter
provides lesser THD than two- and five-level inverter. The hardware of prototype
model is presented in Fig. 13.
Fig. 11 FFT analysis of seven-level inverter offered by simulation and prototype

Fig. 12 Induction motor speed obtained by seven-level inverter output
Table 1 Comparison of
Modulation THD of THD of THD of
modulation index versus total
index two-level five-level seven-level
harmonic distortion
inverter (%) inverter (%) inverter (%)
1 40.48 35.77 11.89
0.9 43.67 36.64 12.28
0.8 45.98 38.93 14.79
0.7 48.43 41.72 15.74
0.6 51.54 31.39 8.94
0.5 36.74 43.68 13.00
0.4 53.00 46.48 18.36
0.3 55.94 49.87 19.97
0.25 57.24 52.56 21.27
5 Conclusion
In this paper, two-, five-, and seven-level cascaded inverters are developed and
modeled. The modeled inverter topologies fed with induction motor are simulated
using MATLAB software. The performance of developed inverters in terms of output
voltage and THD of modeled inverters is compared with the solutions of prototype
model. The THD of two, five, and seven levels inverter is presented with various
modulation indexes. It is obvious from the above results and discussions that the
seven-level inverter offers low harmonic distortion in their output voltages.
Fig. 13 Hardware proof of concept
References
1. Rodriguez J, Lai J-S, Peng FZ (2002) Multilevel inverters: a survey of topologies, controls,
and applicatiosn. IEEE Trans Ind Electron 49:724–738
2. Rodriguez J, Frequelo LG, Kouro S, Leon JI, Portillo RC, Prats MAM, Perez M (2009)
Multilevel converters: an enabling technology for high-power applications. IEEE Proc
97(11):1786–1817
3. Vinoth Kumar N, Cinnaiyan VK (2017) Multilevel inverter topology using single source and
double source module with reduced power electronic components. J Eng 3:467–480
4. Fernão Pires V, Cordeiro A, Foito D, Fernando Silva J (2018) Three-phase multilevel inverter for
grid-connected distributed photovoltaic systems based in three three-phase two-level inverters.
Solar Energy 174:1026–1034
5. Kannan C, Mohanty NK, Selvarasu R (2017) A new topology for cascaded H-bridge multilevel
inverter with PI and Fuzzy control. Energy Proc 117:917–926
6. Xu S-Z, Peng Y-F, Li S-Y (2019) Suppression effectiveness research on multi-level EMI filter
in thermal electromagnetic interactive filed of explosion-proof three-level NPC converter. Case
Stud Thermal Eng 15
7. Gayathri Devi B, Keshavan BK (2019) A novel hybrid phase shifted-modified synchronous
optimal pulse width modulation based 27-level inverter for grid-connected PV system. Energy
178:309–317
8. Chinmaya KA, Singh GK (2019) Modeling and experimental analysis of grid-connected six-
phase induction generator for variable speed wind energy conversion system. Electric Power
Syst Res 166:151–162
9. Garapati DP, Jegathesan V, Veerasam M (2018) Minimization of power loss in newfangled
cascaded H-bridge multilevel inverter using in-phase disposition PWM and wavelet transform
based fault diagnosis. Ain Shams Eng J 9:1381–1396
10. Palanivel P, Selvarasu R, Barani Sundaram B, Alemayehu H (2018) Cascaded hybrid device
multilevel converters for wind mill applications. In: International conference on advances of
science and technology, pp 93–101
11. Chandrasekaran K, Mohanty NK, Sahoo AK. Performance analysis of modular vs non modular
multilevel inverter. In: International Conference on Renewable Energy Integration into Smart
Grids: A Multidisciplinary Approach to Technology Modelling and Simulation (ICREISG),

IEEE, pp 61–67.
12. Chandrasekaran K, Mohanty NK. A flexible rung ladder structured multilevel inverter. Tehnički
vjesnik 27(4):1044–1049
Assessment of Various Vector Control
Schemes for PMSM Drive Application
Kodumur Meesala Ravi Eswar, Chokkalingam Bharatiraja,

and Jayakumar Vinoth
Abstract Currently, the operation of Finite Control Set Model Predictive Control
(FCS-MPC) schemes in electric motor drives domain is emerging. Its distinguished
features are: intuitive, simple and easy insertion of multi-objectives. The application
of FCS-MPC in permanent magnet synchronous motor (PMSM) drive has several
benefits in terms of improving the performance when compared with direct torque
control (DTC)-operated PMSM drive. The FCS-MPC scheme is categorized into
two: predictive torque control (PTC) and predictive current control (PCC). This
paper deals with comparative evaluation of these control schemes for PMSM drive
operation concerning flux and torque ripple, and current THD. Therefore, the gained
results are analyzed for PMSM drive highlighting the benefits of FCS-MPC over
DTC.
Keywords Direct torque control (DTC) · Finite Control Set Model Predictive
Control (FCS-MPC) · Permanent magnet synchronous motor (PMSM) · Predictive
current control (PCC) · Predictive torque control (PTC)
1 Introduction
Considering support of fast microprocessor technologies, the AC motor drives are

powering the future of industry [1]. Nowadays, PMSM drives are emerging, which
is having special advantages in machine point of view [2]. The notable advantages
are stated as: high efficiency, high power density and torque-to-weight ratio. To
achieve high dynamic performance for motor drives, vector control schemes are
K. M. Ravi Eswar (B) · C. Bharatiraja · J. Vinoth

Department of Electrical and Electronics Engineering, SRM Institute of Science and Technology,
Kattankulathur, India
C. Bharatiraja
J. Vinoth
https://doi.org/10.1007/978-981-16-4943-1_6
50 K. M. Ravi Eswar et al.
preferred. The vector control schemes for motor drives started with introduction of
field-oriented control (FOC) in the year 1968 [3]. In beginning, it is implemented
for induction motor (IM) drives. Later, its application is extended for PMSM drives
[4]. The brief features of FOC operation for PMSM drives are: implementation
in rotating reference frame necessitates coordinate frame transformations, indirect
control of flux and torque through current components and involvement of current
PI regulators. Therefore, the FOC operation is stated to be complex for PMSM drive
[5]. Continuing the developments in motor drives, in the year 1986, direct torque
control (DTC) is introduced. In [6], DTC is implemented for IM drive. Considering
the special benefits of PMSM, DTC operation is applied for PMSM drives [7]. Its
operational advantages [7] over FOC are provided as: simple, nonappearance of
current PI controllers, performance in stationary frame of reference, straight regu-
lation of flux and torque. However, hysteresis controller existence in digital stage
of DTC operation consequences poor steady-state response of flux and torque for
PMSM drive [5]. The detailed implementation of DTC and its operational results for
PMSM drive are analyzed in this paper.
On the other side, FCS-MPC schemes are providing as promising technology
in the domain of motor drives [8]. This modeling-based control is accomplished in
stationary/rotating reference frame. The vital operational stages in predictive control
are considered as [9]: data measuring and estimating, prediction and cost function
(CF) evaluation. The FCS-MPC is classified into two [9], which are stated as PCC and
PTC. In both of these control techniques, the mentioned operational stages remain
same. However, the control variables differ. In PCC technique [10], stator current
is considered as basic control variable, whereas in PTC technique, the basic control
variables are flux and torque [11, 12]. In this paper, FCS-MPC (i.e., PCC and PTC)
technique implementation and operational results are examined for PMSM drive.
The overall conclusions are drawn from various control responses obtained for
PMSM drive. Therefore, in this paper detailed investigations are made for the
mentioned control schemes applied to PMSM drive.
2 PMSM Drive Operation
The PMSM drive supplied with two-level voltage source inverter (VSI) is presented
in Fig. 1a. Voltage space vector (E s ) is indicated based on switching combinations.
Its generalized form is given by (1). As it is familiar that the possible switching
combinations are 8, and there are overall 8 voltage vectors (VVs). Their realization
is depicted in Fig. 1b.

2E dc + 2π 4π
Es = Sa + Sb+ e j ( /3) + Sc+ e j ( /3) (1)
3
Assessment of Various Vector Control Schemes … 51
(0, 1, 0) E3 E2 (1, 1, 0)
Sa Sb Sc
Edc
(0, 1, 1) E4 E1 (1, 0, 0)
Sa Sb Sc (0, 0, 0) E0 E7 (1, 1, 1)
(0, 0, 1) E5 E6 (1, 0, 1)
PMSM
(a) (b)
Fig. 1 a PMSM supplied from two-level VSI and b possible VVs of two-level VSI
2.1 Direct Torque Control (DTC)
PMSM drive is executed with DTC technique in stationary (αβ) reference frame.
Its operational block diagram is mentioned in Fig. 2. In first step of its operation,
measurement and estimation of PMSM drive variables are required. It is possible
to have direct measurement of PMSM speed, current phasor and inverter DC link
voltage. The process of present state (k) stator flux estimation is given by (2), where
k − 1 is previous state data and T s is sample period. Torque estimation is given
by (3), where is denotes stator current variable. DTC operation requires stator flux
sector information. Among the probable six sectors (S 1 − S 6 ) having 60° span as
provided in Fig. 3, the stator flux sector information can be gained by (4). In Fig. 3a,
an instant is presented where flux space vector (λs ) falls in sector 2. The reference
flux (|λs * |) during control operation of PMSM drive is set as rated flux value. With
Fig. 2 Block model of PMSM drive operated with DTC technique

E3 E2
β-axis (Tem λs )
λs E3 S2
S3 S2
(Tem λs )
E4 E1
E4 S4 S1 E1 (Tem λs )
E0 E7
λs
E6
(Tem λs )
S5 S6
α-axis
E5 E6
(a) (b)
Fig. 3 a Possible sectors of voltage space vector plane for two-level VSI and b flux space vector
in sector-2 and selection of VV for flux and torque control
these, necessary inputs required for operation of DTC technique are fulfilled.
λs (k) = λs (k − 1) + (E s − Rs i s )Ts (2)
P
Tem = 1.5 imag λs ∗ i s (3)
2

λβs
θ = tan−1 (4)
λαs
During control operation, the speed error (i.e., difference of reference and actual
speed) is fed to speed PI controller, which produces required reference torque value
(T em * ) to minimize speed error. Torque and flux errors are attained by comparing
their reference value with actual.
The torque error is provided to three-level torque hysteresis controller, whose
output (H T ) may be −1 or 0 or +1, based on its error value crossing boundary
conditions. Likewise, two-level flux hysteresis controller is supplied with flux error,
whose output (H λ ) may be −1 or +1, based on its error value crossing boundary
conditions.
The predefined look-up table is framed for VV application in a sample time, which
is based on stator flux sector, outputs of flux and torque hysteresis controller. This
choice of VV is to reduce errors of flux and torque. In view of flux space vector in
sector-2, Fig. 3b shows the effect of VV on flux and torque response. Considering
same analysis, Table 1 represents complete look-up table development for application
of VV (E opt ) in a sample time of DTC operation. Therefore, gaining direct control
over flux and torque of PMSM drive.
Table 1 VV selection for flux and torque control in DTC

Sector identification Hysteresis torque error (H T )
+1 0 −1
Hysteresis flux error (H λ )
+1 −1 +1/−1 +1 −1
S1 E2 E3 E0 E6 E5
S2 E3 E4 E0 E1 E6
S3 E4 E5 E0 E2 E1
S4 E5 E6 E0 E3 E2
S5 E6 E1 E0 E4 E3
S6 E1 E2 E0 E5 E4
2.2 Finite Control Set-Model Predictive Control (FCS-MPC)
At the present time, FCS-MPC techniques drawing more attention in the field of
electrical drives. As the control scheme rely on system model, it is essential to deliver
system dynamic model during its operation. Therefore, dynamic equations of PMSM
used in FCS-MPC operation are given by (5–8), where λf is rotor magnetic flux
linkage along d-axis, ωm is mechanical speed of rotor, L s and Rs are stator inductance
and resistance. These dynamic equations are presented in rotating reference frame
(dq).
dλs
E s = Rs i s + + jλs ωr (5)
dt
λs = L s i s + λ f (6)
P
Tem = 1.5 imag λs ∗ i s (7)
2
dωm
Tem − Tl = J (8)
dt
As control variables are required to be predicted for next sample intervals, predic-
tion model has to be developed based on PMSM dynamic equations. By means of
Euler’s methodology, prediction equations for control variables are framed at (k + 2)
sample instant. (k + 2) signifies two-step ahead prediction of control variables which
obviate sample delay issues during control operation. To execute FCS-MPC, the vital
control stages are: measuring and estimation, prediction, and CF optimization. As
it is stated that FCS-MPC is categorized into two (i.e., PTC and PCC), the part of
first control stage (i.e., measurement) remains same for both. Measuring variables
includes PMSM drive speed (ωr ), current phasor (is ) and DC link voltage (E dc ).
The functioning of remaining stages varies based on two subcontrol operations of

FCS-MPC. These are presented in detail below.
2.2.1 Predictive Torque Control (PTC)
Using flux and torque as basic control variables, PTC technique for PMSM drive
is developed as presented in Fig. 4. PTC functioning is implemented in stationary
reference frame to circumvent complexity of coordinate frame transformation. The
estimated flux and current phasor at (k + 1) sample are given by (9–10), which are
utilized in two-step ahead (k + 2) prediction of flux (11), current (12) and thereby,
torque (13) of PMSM drive is attained. Here, er signifies back emf which is λf ωr ejθ ,
where ωr and θ are electrical rotor speed and angle, respectively. The predictions at
(k + 2) are gained with possible 7 (n) VVs of two-level VSI supplying to PMSM
drive. Therefore, estimation and prediction of control variables fulfilled two stages
of PTC operation.
λs (k + 1) = λs (k) + (E opt − Rs i s (k))Ts (9)

Ts Rs Ts
i s (k + 1) = i s (k) 1 − + (E opt − er ) (10)
Ls Ls
Prediction of control variables:
λs (k + 2)n = λs (k + 1) + ((E s )n − Rs i s (k + 1) )Ts (11)
Fig. 4 Block model of PMSM drive operated with PTC technique


Ts Rs Ts
i s (k + 2)n = i s (k + 1) 1 − + ((E s )n − er ) (12)
Ls Ls
P
Tem (k + 2)n = 1.5 imag λs (k + 2) ∗ i s (k + 2) (13)
2
The final stage involves in CF optimization, which acts crucial role in optimal VV
choice. The basic CF design with control objectives of flux and torque is specified
by (14). The predicted variables of flux and torque are compared with their refer-
ences. The optimal VV serving predicted variable closeness to reference ensures the
minimum CF value. Hence, CF optimization is achieved. This evaluated optimal VV
is chosen for control action, therefore realizing predictive torque and flux control of
PMSM drive.
∗
X n = Tem − Tem (k + 2)n + W λ∗s − |λs (k + 2)n | (14)
In (14), W is termed as weighting factor, creating relative balance between two

dissimilar control objectives (flux and torque).
2.2.2 Predictive Current Control (PCC)
In this control technique, flux and torque of PMSM drive are indirectly controlled
by current components in rotating (dq) reference frame. Therefore, basic control
variable in PCC is stator current phasor. Its functioning block diagram is presented
in Fig. 5. Till the rated speed of PMSM drive, demagnetization of rotor flux is not
necessary. Therefore, d-axis reference stator current (ids * ) is set to zero. The q-axis
reference stator current (iqs * ) is set by speed PI controller according to speed error
minimization. In first stage, current phasor estimation at (k + 1) is attained by (15).
Fig. 5 Block model of PMSM drive operated with PCC technique

With the presented 7 (n) VVs of two-level VSI, measured and estimated data, the
current prediction at (k + 2) is obtained in second stage as given by (16).

Ts Rs Ts
i s (k + 1) = i s (k) 1 − − jωr Ts + E opt − jλ f ωr (15)
Ls Ls
Prediction of control variable:

Ts Rs Ts
i s (k + 2)n = i s (k + 1) 1 − − jωr Ts + (E s )n − jλ f ωr (16)
Ls Ls
In PCC, CF (17) is established with elementary control objectives of stator current

components. The optimal VV satisfying closeness of predicted current components
with their references guarantees minimum CF value and optimization at final stage
of PCC operation. The evaluated optimal VV serves predictive current control and
thereby, indirect flux and torque control of PMSM drive.
∗ ∗
X n = i ds − i ds (k + 2)n + i qs − i qs (k + 2)n (17)
From this entire discussion, it is summarized that PMSM drive operation is

elucidated with various control techniques such as DTC, FCS-MPC: PTC and PCC.
Using mathematical analysis presented in part-2, MATLAB/Simulink simulations

are conducted on existing PMSM drive whose ratings are listed in Table 2. The
discussed control techniques are applied for PMSM drive operation.
Table 2 PMSM drive

Parameters Value
parameters
Stator resistance (Rs ) 1.12
Stator inductance (L s ) 10.5 mH
Rated stator flux (|λs * |) 0.705 Wb
Rated power 5 HP
Rated torque 23.5 N m
Poles (P) 4
Inertia 0.061 kg m2
Rotor flux (λf ) 0.7 Wb
Proportional gain 0.2
Integral gain 0.05
Flux (Wb) Speed (rpm) 60 60
Flux (Wb) Speed (rpm)

50 50
40 40
30 30
20 20
0.72 0.72
0.71 0.71
0.7 0.7
0.69 0.69
4 4
Torque (Nm)
Torque (Nm)
0 0
-4 -4
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
Time (Sec) Time (Sec)
(a) (b)
60
Torque (Nm) Flux (Wb) Speed (rpm)
50
40
30
20
0.72
0.71
0.7
0.69
4
-4
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
Time (Sec)
(c)
Fig. 6 Response of PMSM drive during low speed (50 rpm) condition a DTC, b PTC and c PCC
In PTC, CF (14) is assigned with weighting factor value of 75, which is desig-
nated from heuristic tuning, whereas PCC operation is independent from weighting
factor. Considering this simulating environment, various control operation results
are presented and discussed in this part of paper.
The simulations are executed on PMSM drive operated with DTC, PTC and
PCC techniques. Transition-free response is captured during PMSM drive opera-
tion showing speed, flux and torque in Fig. 6. In Fig. 6, low-speed reference (50 rpm)
is selected for PMSM drive while observing response of DTC, PTC and PCC tech-
niques. From these tests, it is evident that PTC and PCC techniques exhibits overall
improved control response when compared to DTC. In DTC-operated PMSM drive,
the existence of hysteresis regulators for flux and torque control is responsible for
lack of better response. On the other side, two step ahead prediction, intuitive nature
with the consideration of CF optimization in every sample instant made predictive
control techniques gaining the benefit of improved motor drive response.
The transition response is analyzed for PMSM drive with sudden variations in
speed. In Fig. 7, the reference speed variations are indicated, i.e., from 50 rpm
50 50 50
Speed 0 0 0
(rpm) -50
-50 -50
0.72 0.72 0.72
Flux 0.71 0.71 0.71
0.7 0.7
(Wb) 0.7 0.69 0.69
0.69
10 10
Torque 0 0 10
-10 -10 0
(Nm) -20 -20 -10
-20
-30 -30 -30
1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2 1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2 1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2
Time (Sec) Time (Sec) Time (Sec)
(a) (b) (c)
Fig. 7 Response of PMSM drive during speed step change from forward (50 rpm) to reverse
(−50 rpm) conditions a DTC, b PTC and c PCC
Speed (rpm) 1550 1550
Speed (rpm)
1500 1500
1450 1450
0.72 0.72
Torque (Nm) Flux (Wb)

0.71 0.71
0.7 0.7
0.69 0.69
14 14
12 12
10 10
8 8
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
Current (A)
Current (A)
5 5
0 0
-5 -5
3.8 3.82 3.84 3.86 3.88 3.9 3.92 3.94 3.96 3.98 4 3.8 3.82 3.84 3.86 3.88 3.9 3.92 3.94 3.96 3.98 4
Time (Sec) Time (Sec)
(a) (b)
1550
Speed (rpm)
1500
1450
0.72
0.71
0.7
0.69
14
12
10
8
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
Current (A)
-5
3.8 3.82 3.84 3.86 3.88 3.9 3.92 3.94 3.96 3.98 4
Time (Sec)
(c)
Fig. 8 Response of PMSM drive during loaded condition at 1500 rpm a DTC, b PTC and c PCC
(forward) to −50 rpm (reverse). The executed response of PMSM drive operated
with DTC (2 s to 2.023 s), PTC (2 s to 2.023 s) and PCC (2 s to 2.024 s) techniques
are presented in Fig. 7. From these, it is verified that all control techniques perform
almost similar transition response for PMSM drive. However, PCC-operated PMSM
drive has slight variation in dynamic response, i.e., 2 s to 2.024 s, due to indirect
control of torque via current space vector.
Under loaded condition (load torque as 12.5 Nm) at rated speed of 1500 rpm,
PMSM drive response for DTC, PTC and PCC techniques is presented in Fig. 8. From
these tests, it is evident that PTC and PCC techniques exhibit overall improved control
response when compared to DTC. The obtained flux and torque ripples (standard
deviation values) are presented in Table 3. From this, it is clearly observed that PTC
of PMSM drive stands top gaining less torque ripples, followed by PCC and DTC
at the end, whereas for current THD observations, PCC stands top with less THD,
followed by PTC and DTC at the end. Therefore, in comparison of PTC and PCC
techniques, direct involvement of current objective in CF of PCC results less current
THD than PTC. While in PTC, CF is structured with flux and torque objectives, which
is accountable for less torque ripples than PCC. Flux response is much influenced
Table 3 PMSM drive

Control Flux ripple Torque ripple Current THD
performance comparison
technique (Wb) (Nm) (%)
under various control
techniques DTC High High (1.207) High (15.22)
(0.0067)
PTC Low (0.0053) Lower (0.615) Low (7.41)
PCC Lower Low (0.681) Lower (6.70)
(0.0031)
by direct control of current component (ids ) which is possible in PCC resulting less
flux ripples than PTC.
The outcomes of this entire analysis are listed in Table 3, which differentiates
among control techniques by quantitative and qualitative data. Thus, PMSM drive
operation with various control techniques is studied.
4 Conclusion
In this paper, the concept and operation of different vector control techniques such
as DTC, FCS-MPC (PTC and PCC) for PMSM drive application are studied. PMSM
drive is operated with these control techniques in simulation platform. The control
responses for different operating conditions of PMSM drive are observed. The
presented qualitative and quantitative analysis clearly depicts operational differences
among them. In comparison of FCS-MPC with DTC, FCS-MPC techniques serves
overall better performance for PMSM drive.
References
1. Samir K, Rodriguez J, Wu B, Bernet S, Perez M (2012) Powering the future of industry:

high-power adjustable speed drive topologies. IEEE Ind Appl Mag 18(4):26–39
2. Behzad A, Rahrovi B (2010) Minimum-copper-loss control over full speed range of an IPMSM
drive for hybrid electric vehicle application. In: 2010 IEEE vehicle power and propulsion
conference, pp 1–6
3. Felix B (1972) The principle of field orientation as applied to the new TRANSVECTOR closed
loop control system for rotating field machines. Siemens Rev 34(5):217–220
4. Pradeep K, Dhundhara S, Makin R (2016) Performance analysis of PMSM drive based on
FOC technique with and without MRAS method. In: 2016 International Conference on Recent
Advances and Innovations in Engineering (ICRAIE), pp 1–6
5. Fatih K, Topaloglu I, Faruk Cakir M, Gurbuz R (2013) Comparative performance evaluation of
FOC and DTC controlled PMSM drives. In: 4th international conference on power engineering,
energy and electrical drives, pp 705–708
6. Isao T, Noguchi T (1986) A new quick-response and high-efficiency control strategy of an
induction motor. IEEE Trans Industry Appl IA-22(5):820–827
7. Mino-Aguilar G et al (2010) A direct torque control for a PMSM. In: 2010 20th international
conference on electronics communications and computers, pp 260–264
8. Dehong Z, Zhao J, Li Y (2016) Model-predictive control scheme of five-leg AC–DC–AC
converter-fed induction motor drive. IEEE Trans Industr Electron 63(7):4517–4526
9. Jose R, Kazmierkowski MP, Espinoza JR, Zanchetta P, Abu-Rub H, Young HA, Rojas CA
(2013) State of the art of finite control set model predictive control in power electronics. IEEE
Trans Industrial Informatics 9(2):1003–1016
10. Turker T, Buyukkeles U, Faruk Bakan A (2016) A robust predictive current controller for
PMSM drives. IEEE Trans Industrial Electronics 63(6):3906–3914
11. Eshwar K, Ravi Eswar KM, Vinay Kumar T (2020) An effective predictive torque control
scheme for PMSM drive without involvement of weighting factors. IEEE J Emerging Selected
Topics Power Electronics. https://doi.org/10.1109/JESTPE.2020.2989429
12. Eshwar K, Vinay Kumar T (2020) Effective predictive torque control scheme for four-level
open-end winding permanent magnet synchronous motor drive. Int Trans Electrical Energy
Syst 30(10):e12536
Design and Implementation of Discrete
Controller-Based Zeta Converter
for Solar Power Applications
R. Uthirasamy, V. Kumar Chinnaiyan, U. S. Ragupathy,

and S. Vishnu Kumar
Abstract In this paper, a closed-loop discrete controller is implemented for Zeta

converter modules. Discrete controller offers good dynamic response, fine voltage
regulation, and it can be programmable without external passive components.
Zeta converter is capable of producing non-pulsating output current and has good
adaptability. An analog proportional–integral–derivative (PID) and discrete PID
controllers are designed, and their performances are compared. Effectiveness of
the controllers is demonstrated and validated with experimental results. Achieved
result shows that the discrete controller provides good transient response. Further,
the performance of connecting dual snubber with Zeta converter has been analyzed.
Dual snubber is connected in the Zeta converter, and the conducted EMI results have
been observed. Simulation model of a solar module and Zeta converter to control
the voltage to the DC motor has been developed using PSPICE software. From the
experimental results, it is inferred that the dual snubber reduces the switching losses
and suppresses the dv/dt and di/dt effects.
Keywords Zeta converter · Discrete controller · Electromagnetic interference ·

Snubber circuit · Switching loss
1 Introduction
DC–DC converters play a major role for the integration of renewable energy
resources. The efficiency of DC–DC converter is estimated by its duty ratio and
R. Uthirasamy (B) · V. Kumar Chinnaiyan

Department of Electrical and Electronics Engineering, KPR Institute of Engineering and
Technology, Coimbatore, India
U. S. Ragupathy
Department of Electronics and Instrumentation Engineering, Kongu Engineering College, Erode,
India
S. Vishnu Kumar
Department of Electronics and Communication Engineering, Vel Tech Rangarajan Dr. Saguthala
R&D Institute of Science and Technology Chennai, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_7
62 R. Uthirasamy et al.
its losses. Zeta converter topology is proposed to interface the solar PV and drive
load system. The emerging renewable energy such as solar energy is the feasible
solution for today’s energy crisis. The output current/power of an individual cell or
an array of cells connected in series or in parallel depends on the working voltage
available at the terminals. The maximum power produced by the PV system varies
with intensity of solar irradiation and temperature. Thus, to raise the competence of
PV system, it is essential to operate the system with highest output power. Maximum
power point tracking (MPPT) has been implemented with a variety of techniques
and power conditioner. The subsystem of the photovoltaic installations is the solar
converter. The employment of Zeta converters in the charge controllers is ordinary in
a standalone photovoltaic system. Zeta converter is a DC–DC converter having two
inductors and two series capacitor. It is accomplished to operate in moreover step-
up or step-down manner. It performs a non-inverting buck-boost function similar
to that of a SEPIC converter. Zeta converter is one of the best choices to regulate
the solar power. Proposed Zeta converter is working in continuous conduction mode
(CCM) with a coupled inductor [1–5]. The major challenge in the field of DC to
DC converter is the control aspect. The control method necessitates the efficient
modeling and enabling systematic study of the converters. In general, the converters
are time invariant and nonlinear, and it comprises with the large passive components.
In common, analog controller design technique experiences higher complexity in
control, lesser flexibility to superior functions and system changes and have lack of
consistency. The proposed discrete controller provides lot of advantages than their
analog controller. It also offers in its dynamic response as firmness, fast response
and fewer peaks overshoot [6]. In this proposed work, high-degree DPWM signals
are produced by maintaining high system switching frequency. The main purpose of
control of discrete controller has to run the boost converter along with a duty ratio so
that the reference voltage and the output voltage of the DC component are equal. The
parameter support stability in spite of changes in the load or deviations in the input
voltage. The state space model builds the controller and makes simulation much
quicker than the other method [7–10]. The Zeta converter is represented by means
of state space method, and the simulation is carried out by MATLAB/Simulink soft-
ware. The block diagram of digitally controlled Zeta converter with entire setup is
revealed in Fig. 1.
The output voltage and the reference voltage are compared using comparator,
and the output is fed to compensator. Generally, compensators are designed using
flip-flops. Desired PWM pulses are generated in the comparator and fed to the zeta
converter from side to side the gate driver [11, 12]. Use of buck converter is general
in the charge controller of solar-powered DC motors. The converters that control the
input voltage of DC motor generate EMI which is conducted all the way through
the cables to photo voltaic system; from there on, it is radiated when the solar cells
act as antenna. Those unwanted radiations impede with the communiqué structure
over and above any receptive tools nearby. To stay away from such unnecessary
and unwarranted interference, the finest way is to assail the source of interference,
specifically the EMI produced by the Zeta converter switching strategy.
Design and Implementation of Discrete Controller-Based … 63
Fig. 1 Digitally controlled

block diagram of Zeta
converter
2 Zeta Converter
Zeta converter consists of inductors L 1 and L 2 and capacitor c and diode D. Its main
advantages is to maintain the output voltage polarity same as the input. As the Zeta
is a switch mode circuit, its performance is strongly reliant in the stability of the L
current and C voltage. The voltage of the Zeta converter is implicit to activate in the
continuous conduction mode (CCM) as shown in Fig. 2. There exist two modes of
operation using one switch mode, when it is turned off. It consists of MOSFET as
switch, two capacitors C 1 and C 2, diode D, two inductors L 1 and L 2 and load R.
2.1 ON State Condition of Zeta Converter (MODE 1)
While the switch S is turned on (ON-state), the diode D gets reversed biased, and it
is in off state as shown in Fig. 3. For the duration of this state, inductor L 1 and L 2
get energize, and the inductor current iL1 and iL2 increase linearly. The capacitor C 1
will discharge to V o , and it is associated in sequence with L 2 . So the voltage across
the inductor (L 2 ) is V s and diode is V s + V o . By applying Kirchoff’s voltage law,
voltage across L 1 will be,
Fig. 2 Circuit diagram of

Zeta converter
Fig. 3 Mode I operation of zeta converter
di L1 Vs
= (1)
dt L1

Vs
I L1 = Ton (2)
L1
And voltage across L 2 will be,
di L2 Vs VC1 VC2
= + − (3)
dt L2 L2 L2

VS VC1 Vo
I L1 = + − Ton (4)
L2 L2 L2
By applying Kirchoff’s current law, the rate of voltage through the capacitors will
be,
dVC1 i L2
=− (5)
dt LC1

i L2
VC1 = − Ton (6)
C1
dVC2 i L2 VC2
= − (7)
dt C2 R L C2

i L2 VC2
VC1 = − Ton (8)
C2 R L C2
2.2 OFF State Condition of Zeta Converter (MODE 2)
While the control switch S is turned OFF (OFF-state), the diode D gets high potential
at anode, and it is in ON state. At this condition, inductor L 1 and L 2 are in de-
energized. The voltage polarity of the conductor changes, the diode will get forward
biased and it will conduct. Energy in inductors is de-energized to capacitor C 1 and
load resistance, respectively. Consequently, the energy in the inductor current iL1
and iL2 is decreasing linearly. By applying Kirchoff’s voltage law, voltage across
inductor L 1 is expressed as,
di L1 Vc1
=− (9)
dt L1
Vc1
I L1 = Toff (10)
L1
Voltage across inductor L 2 is expressed as,
di L2 VC2
=− (11)
dt L2
VC2
I L2 = Toff (12)
L2
The proposed Zeta converter is pretend using MATLAB/Simulink and the output
voltage attained using discrete PID controller. The controller constraints are in under
considerations with steady-state error, settling time, peak time, rise time, ripple output
voltage and overshoot which is evaluated alongside its PI, PID controllers are premed-
itated for the identical Zeta converter. Steady-state error examined for load change
is much smaller than 1% and no undershoot or overshoot is clear. The presentation
necessities for the Zeta converter with Discrete PID controller are improved than
discrete PI, analog PID and analog PI controllers. The results thus accomplished
with discrete PID controller for Zeta converter is comparable and agreement with
the numerical computation. It is confirmed that the discrete system demonstrates an
enhanced outcome than the analog controllers. The input voltage is fixed as 10 V
until 0.03 s and then varied to 14, 18, 14 and 10 V throughout the time instance 0.03 s,
0.06 s, 0.09 s, and 0.12 s, respectively. The load resistance is also changed concur-
rently with an input voltage. The load resistance is fixed as 14 till 0.03 s, and then
varied to 18 , 22 , 18 , and 14 during the instance 0.03 s, 0.06 s, 0.09 s and
0.12 s, respectively. The reference voltage is also all together changed up to 0.075 s,
it is 7 V afterward it is 20 V. In the simulated answer, the performance of Buck action

up to 0.075 s and then carry out as a boost process in Zeta converter with the changes
in input voltage and load resistance. In dual operation, overshoot or undershoots are
not observed in the output voltage, and there is no mark for steady-state error, and
output ripple voltage is forever smaller than 3%. By changing the capacitance and
inductance values of the Zeta converter, the simulated outcomes are tabulated. The
system does not prove any steady-state error, undershoot or overshoots, and it settles
down at a speedy rate with a settling time of about 4 ms for all the values. Simulation
is also brought out by changing the load not limiting to R, it can extend to RL and
RLE load. The output voltage has not varied due to the variation in nature of the
load. Therefore, the above results confirmed that the designed discrete controller
gives tight output voltage regulation, better stability and robustness. The efficiency
against load resistance graph of the Zeta converter and the discrete PID controlled
Zeta converter is depicted in Fig. 4. The input voltage is 12 V, reference voltage is
21 V, and the changes load resistance is between 2 and 12 , then the efficiency
of the discrete PID controlled Zeta converter efficiency has been varied from 97.8
to 96.8%, and the efficiency of the uncontrolled Zeta converter is varied from 84 to
90%.
Figure 5 depicts the output voltage responses of the analog PID controlled and
discrete PID controlled Zeta converter. Fix the Zeta converter input voltage is 14 V
and the reference output voltage is 21 V, the attained analog PID controlled Zeta has
more ripple voltage, overshoot and steady-state error. It has more rise time and less
accurate settling time. The discrete PID controlled Zeta has low settling time and rise
time, and the peak overshoot is less than 1% with an insignificant steady-state error
with no ripple voltage. With this response, it is very well identify that the performance
parameters are outstanding in discrete PID controller than in analog PID controller.
The output voltage response of the discrete PID controlled Zeta converter for change
in load is view sequentially. The nature of load is changed as R, RL and RLE and
changing nature has not influenced the output voltage of the converter is clear. The
Fig. 4 Comparison of
efficiencies for Zeta and
discrete PID controlled Zeta
converter
Fig. 5 Comparison between

output voltage responses of
discrete PID controlled and
analog PID controlled Zeta
converter
load resistance is differed as 14 , 18 and 10 ; the converter will be capable

to produce the output voltages as 21 V, 20.998 V and 6.996 V for the reference
voltages of 21 V, 21 V and 7 V, respectively. Then the simulation result is taken for
the inductance of 5 mH and 10 mH, inserted to the resistance of 14 and 18 ,
the output voltage is acquired in the order of 7.001 V and 21.002 V, respectively, for
the fixed reference of 7 V and 21 V. Again, the output has been obtained using RLE
load with a resistance of 20 , inductance of 1 mH and a voltage source of 2 V, the
controller is capable to find the output voltage as 20.997 V for the specific reference
of 21 V. Similarly, an output voltage of 7.001 V is tracked for the fixed reference of
7 V, whose RLE values are 9 , 5 mH and 3 V.
4 Hardware Results and Discussion
In the hardware setup, the input voltage to the discrete PID controlled Zeta converter
is 12 V, reference voltage is fix as 21 V and the resistance of load is considered as
14 ; the obtained output voltage is 21.1 V as depicted in Fig. 6. It has smaller settling
time and rise time for the time period of 1.5 ms and has oscillation at the initial time
period, but it has settled shortly. No undershoot or overshoot is evident. The input
voltage is obtained at channel 2 and the output voltage is obtained at channel 3. In
the output voltage response, steady-state error is 0.1 V which is lesser than 1%.
In the same way, the input voltage is 18 V, for 21 V reference and the load resistance
is 14 ; the resultant voltage is illustrated in Fig. 7.
5 Conclusion
In this paper, the discrete PID controller for Zeta converter has been designed and
implemented. The discrete PID controlled Zeta converter has improved efficiency
Fig. 6 Output voltage response for 12 V input, R0 = 14 , and V ref = 21 V
Fig. 7 Output voltage acquired for 18 V input, R0 = 14 , and V ref = 21 V
than the conventional Zeta converter. The results attained with the proposed discrete
PID controller for DC–DC converter are in concurrence with the arithmetic compu-
tation and also better perform than the existing analog controllers. Simulation results
divulge that the converters not only illustrate the evidence of the steady state and
transient performance but also enhance the efficiency of the Zeta converter. The
numerical investigation, simulation results and the investigational response reveal
that the controller attains high output voltage regulation, good stability, superior
dynamic performances and more efficiency. This topology is well-suited with all
other DC–DC converter and also can be extended for any of the applications such as
PV, Wind, telecommunication applications and speed control of DC motor drives.
References
1. Maki A, Valkealahti S (2012) Power losses in long string and parallel connected short strings of
series-connected silicon-based photovoltaic modules due to partial shading conditions. IEEE
Trans Energy Convers 27(1):173–183
2. Walker GR, Sernia PC (2004) Cascaded dc-dc converter connection of photovoltaic modules.
IEEE Trans Power Electron 19(4):11301139
3. Walker GR, Pierce J (2006) Photo voltaic DC–DC module integrated converter for novel
cascade and bypass grid connection topologies-design and optimization. In: Proceeding of
power electronics specialist conference-PESC pp. 1–7, 18–22 (June 2006)
4. Kaushika N, Gautam N (2003) Energy yield simulations of interconnected PV arrays. IEEE
Trans Energy Convers 18(1):127–134
5. Lee J, Min B, Kim T, Yoo D, Yoo J (2009) High efficient interleaved input-series-output-
parallel-connected converter for photovoltaic power conditioning system. In: Proceedings of
energy conversion congress and exposition, pp 327–329
6. Chen L, Wu H, Xu P (2015) A high step-down non-isolated bus converter with partial power
conversion based on synchronous LLC resonant converter. In IEEE conferences applied power
electronics conference and exposition
7. Ahmad MdW, Anand S (2016) Power decoupling in PV system using partial power processing
converter. IEEE conference in compatibility, power electronics and power engineering (2016)
8. Zapata JW, Renaudineau H (2016) Partial power DC–DC converter for photovoltaic microin-
verters. IEEE Ind Electron Soc
9. Zhao J, Yeates K, Han Y (2013) Analysis of high efficiency converter processing partial
input/output power. IEEE control and modeling for power electronics
10. Agamy MS (2014) An efficient partial power processing converter for distributed PV
architectures. IEEE Trans Power Electron 29(2) (2014)
11. Agamy MS, Chi S, Elasser A, Harfman-Todorovic M, Jiang Y, Mueller F, Tao F (2013) A high
power density DC–DC converter for distributed PV architectures. IEEE J Photovolt 3(2):791–
798
12. Suntio T (2015) Comments on an efficient partial power processing converter for distributed
PV architectures. IEEE Trans Power Electron 30(4)
Structural Optimization of LMS
Adaptive Filter Using Multi-stage
Cascaded Configuration
S. Hannah Pauline, Samiappan Dhanalakshmi, and R. Kumar
Abstract In telecommunication systems, the most crucial challenge is to recover a

signal corrupted by noise. Adaptive filters are widely utilized for signal de-noising
applications with least mean square adaptive algorithm being the most desired. Even
though LMS algorithm is simple in structure and robust; it suffers from slow conver-
gence speed and has high mean square error (MSE) value. Therefore, in this paper, we
introduce a structural optimization of adaptive filters using conventional LMS adap-
tive algorithm. Here, a multi-stage cascaded LMS adaptive filter model is proposed
for adaptive noise cancellation (ANC) systems, and the novelty is that the number
of filter stages to be cascaded is adjusted automatically to provide optimum perfor-
mance. The simulation is carried out for a signal de-noising application and the
efficiency of the proposed filter model is tested with regards to MSE, SNR and ANR.
The advantages of the proposed filter model are proved in the results.
Keywords LMS · MSE · Signal de-noising · Adaptive noise cancellation · SNR
1 Introduction
Noise is one of the causes for signal quality loss, rendering it useless for further
processing. Signal de-noising is therefore applied in many areas, including telecom-
munications and biomedical applications. Before the signal is further processed for
advanced applications, the signal de-noising mechanism should be applied which is
S. Hannah Pauline · S. Dhanalakshmi (B) · R. Kumar

Department of Electronics and Communication Engineering, College of Engineering and
Technology, Faculty of Engineering and Technology, SRM Institute of Science and Technology,
SRM Nagar, Kattankulathur, Kanchipuram, Chennai, Tamil Nadu 603203, India
S. Hannah Pauline
R. Kumar
https://doi.org/10.1007/978-981-16-4943-1_8
72 S. Hannah Pauline et al.
Fig. 1 General block

representation of ANC
done by adaptive noise canceller (ANC). For ANC, adaptive filters are more desir-
able because, depending on the error signal, they can change their coefficients, and
no prior knowledge is needed. Adaptive filters [1] are used in numerous applications
including echo [2] and noise cancellation [3]. Fig. 1 depicts the fundamental ANC.
The primary input signal provided to the ANC is the noisy signal d i (n) defined as
di (n) = a(n) + n(n), (1)
where a(n) is the noise-free signal and n(n) is the added noise. a(n) and n(n) are
not time correlated to each other and the input to the filter x i (n) is a noise in time
correlated to n(n). The adaptive filter gives the replica of the noise signal n̂(n) as
yo (n) = w T (n)xi (n), (2)
where w(n) = [w0 , w1 ,…wM-1 ]T and xi (n) = [x 0 (n), x 1 (n − 1),…x M-1 (n-M + 1)]T
are weights of the filter and its input, respectively, M is the order of filter and the
error is computed as
eo (n) = di (n) − yo (n) = di (n) − n̂(n), (3)
such that the effect of noise is minimal. The efficiency of the adaptive filter
is improved by the using suitable algorithms like LMS and its variants. Due to
its feasible implementation and robustness, the LMS adaptive algorithm [4] is
commonly used. The LMS algorithm, however, suffers from low convergence speeds.
The filter structure can be adjusted as suggested by several researchers to achieve
optimum performance with regard to steady-state MSE and convergence speed.
Cascade configuration [5] of adaptive filter has demonstrated optimal performance
and is successfully used in ANC systems. The cascaded filter structure was first
proposed by [6] to boost the broadband noise signal and its tracking of multiple
sinusoids. A reduction of MSE for linear prediction application is achieved by the
FIR cascaded structure proposed by Prandoni and Vetterli [7]. Cascaded LMS adap-
tive filter models have been successfully used also for motion artefact deletion from
ECG [8] and EEG [9] signals. Compared to traditional LMS adaptive filters, the
2-stage [10] and 3-stage [11] cascaded LMS adaptive filter architectures used for
Structural Optimization of LMS Adaptive Filter … 73
ANC systems provide better efficiency with regard to SNR. For successful removal
of impulse noise from speech signals with reduced computational complexity, the
multi-stage cascaded adaptive filter structure proposed in [12] is used.
The above studies show that a cascaded adaptive filter model performs better
than conventional single stage adaptive filter in regards to convergence speed and
MSE. Therefore, in this paper, we have explored the efficacy of multi-stage cascaded
LMS filter model for signal de-noising application. Also, we have introduced auto-
matic control logic to adjust the number of stages to be cascaded to obtain optimal
performance in terms of MSE and convergence speed. The proposed multi-stage
cascaded LMS adaptive filter model is detailed in Sect. 2. Section 3 includes the
MATLAB simulation results, thus verifying the proposed method’s effectiveness,
and a conclusion with the future scope is included in Sect. 4.
2 Proposed Multi-stage Cascaded LMS Adaptive Filter

Model with Automatic Stage Control
Conventional LMS algorithm is preferred due to its simple structure and superior
stability. But its convergence speed is low and has high MSE value. Cascaded LMS
filter structures have proved to increase the speed of convergence and provides lower
MSE values. We present the multi-stage cascaded LMS adaptive filter structure using
a cascaded connection of the conventional LMS adaptive filter for signal de-noising
application. Also. we have implemented an automatic control logic which controls
the number of stages to be cascaded. Figure 2 represents the schematic diagram of
the proposed multi-stage cascaded LMS filter model for ANC system.
The primary input signal d L (n) to each stage, as depicted in the figure, is the error
eL−1 of the previous stage ANC. The reference residual noise signal from previous
stage x L−1 (n) − yL−1 (n) is given as the filter input x L (n) for the next stage ANC.
The primary input signal d 1 (n) to stage I ANC wherein has the clean signal a(n)
corrupted by additive noise n(n)
Fig. 2 Schematic diagram of proposed multi-stage cascaded LMS adaptive filter

d1 (n) = a(n) + n(n). (4)
The stage I filter is given a reference input signal n (n) correlated in time to the
noise n(n). a(n) and n(n) are assumed to be uncorrelated, and the noise signal from
d 1 (n) is eliminated using n (n).
x1 (n) = n (n). (5)
Stage I filter output is
y1 (n) = w1T (n)x1 (n) = wlT (n)n (n) = n̂(n), (6)
where w1 (n) = [w0 , w1 ,…wM −1 ]T and x1 (n) = [x 0 (n), x 1 (n − 1),…x M −1 (n-M +

T
1)] are weights of the filter and its input, respectively, at stage I, M is the filter order.
LMS algorithm is employed to adjust the filter weights as
w1 (n + 1) = w1 (n) + μe1 (n)n (n), (7)
μ is the step-size parameter and the error of stage I is
e1 (n) = d1 (n) − y1 (n) = a(n) + n(n) − n̂(n) = a(n) + n(n), (8)
where n(n) = n(n) − n̂(n) is the residual noise to be reduced. The error signal
e1 (n) of stage-I ANC is applied to stage-II ANC as its primary input signal d 2 (n)
d2 (n) = e1 (n) = d1 (n) − n̂(n) = a(n) + n(n), (9)
and the input x 2 (n) to the stage-II filter is the difference between the input signal
and output signal of stage-I filter x 1 (n) − y1 (n).
x2 (n) = x1 (n) − y1 (n) = n (n) − n̂(n) = n (n). (10)
The residual noise signal n̂(n) is
y2 (n) = w2T (n)x2 (n) = w2T (n)n (n) = n̂(n), (11)
where w2 (n) = [w0 , w1 , …wM −1 ]T and n (n) =

[n 0 (n), n1 (n − 1), . . . n (n − M + 1)] are weights of the filter and its
T
input, respectively, at stage II. The weight update equation of stage-II LMS filter is
w2 (n + 1) = w2 (n) + μe2 (n)n (n), (12)
The stage-II ANC error signal e2 (n) is
e2 (n) = d2 (n) − y2 (n) = a(n) + n(n) − n̂(n) = a(n) + δn(n) (13)

where δn(n) = n(n)−n̂(n) is the residual noise that is to be reduced. Further-

more, the error signal of the previous stage is the primary input signal for each stage.
The input signal to the Nth stage filter is the reference residual noise from the N − 1th
stage ANC. The number of stages required to give optimal performance is adjusted
automatically, and at the optimal Nth stage, the error from stage-(N-1) constitutes
the primary input signal of the Nth stage given by,
d N (n) = e N −1 (n) = a(n) + γ n(n), (14)
where γ is a minimal noise. The reference noise input to Nth stage filter is the
reference residual noise from the preceding (N–1)th stage ANC.
x N (n) = x N −1 (n) − y N −1 (n) = γ n (n). (15)
The stage-N adaptive filter output yN (n) gives the estimate of the remaining noise
γ n̂(n) at the N th stage.
n ) = w TN (n)γ n (n) = γ n̂(n),

y N (n) = w TN (n)x N ( (16)
where wN (n) = [w0 , w1 ,…wM −1 ]T and γ n(n) =

[γ n 0 (n), γ n 1 (n − 1), ...γ n M−1 (n − M + 1)]T are the coefficients of the filter
and its input, respectively, at stage N. Stage-N filter co-efficient update is
w N (n + 1) = wL (n) + μe N (n)x N (n). (17)
The error e N (n) of stage-N ANC is calculated by
e N (n) = d N (n) − y N (n) = a(n) + γ n(n) − γ n̂(n) ≈ a(n). (18)
γ n(n) − γ n̂(n) ≈ 0; therefore, e N (n) ≈ a(n). We conclude from the above

analysis that as the filter approaches its optimal stage the clean signal is estimated
with minimal noise. To minimize the noise and thus improve the steady-state MSE,
it is necessary to automatically change the number of stages to its optimal N = Nopt
stage. We employ an automatic selection of stages (refer Sect. 2.2) to attain optimal
performance.
2.1 Mean Square Error (MSE)
The error at the optimal stage is
e N (n) = d N (n) − y N (n) = a(n) + n(n) − [y1 (n) + y2 (n) + · · · + y N (n)]. (19)
At the optimal stage N of the ANC, y1 (n) + y2 (n) + · · · + y N (n) = ñ(n) (estimate
of n(n)) and the MSE is denoted as:

E |e N (n)|2 = E |a(n)|2 + E[|n(n) − ñ(n)2 ] − 2E[a(n)(n(n) − ñ(n))|]
(20)
= E |a(n)|2 + E[|n(n) − ñ(n)2 ] − 2E[a(n)n(n)|] + 2E[|a(n)ñ(n)|]
The following equation is obtained due to the uncorrelation between added noise
n(n) and the uncorrupted signal a(n).
2E [|a(n)n(n)|] = 0 (21)
Meanwhile, a(n) and n̂(n) are also uncorrelated; hence, the following is stated,
2 E[|a(n)ñ(n)|] = 0 (22)
Inserting Eqs. (21) and (22) in (20)

E |e N (n)|2 = E |a(n)|2 + E[|n(n) − ñ(n)|2 ] (23)
Also, we note that the replica of the noise-free signal a(n) is achieved as.
E[|n(n) − ñ(n)|2 ] is minimized. It means that at the optimal filter stage N,
y1 (n) + y2 (n) + · · · + y N (n) is as close to n(n) as possible, and hence E[|n(n) −
ñ(n)|2 ] is minimized (i.e.),
[y1 (n) + y2 (n) + · · · + y N (n)] ≈ n(n) (24)
The above analysis proves that the noise signal can be removed from the input
signal d 1 (n) by adjusting the number of stages of the filter, and the O(n) represents
the de-noised from the ANC.
O(n) = e N (n) ≈ a(n) (25)
Thus, we infer that the de-noised signal O(n) is obtained as the number of stages
in the cascaded filter structure approaches its optimal value.
2.2 Control Logic for Automatic Stage Selection
The MSE value reaches its minimal value only at the filter’s optimal stage as proved
in the above section. The Pearson cross-correlation function is computed between
the error of each stage L, eL (n) and the reference input noise n (n). This value is
used to compute the optimum filter stage. It is presumed that the noise signal n (n)
which is the reference noise is in time correlated to the added noise signal n(n) but
is not correlated to the uncorrupted signal a(n). The error output of each stage ANC
is the replica of the uncorrupted signal (i.e.) e L (n) = â(n); therefore, the correlation
between eL (n) and n (n) reduces as the filter reaches its optimal stage. The estimated
correlation coefficient between eL (n) and n (n) is defined as

ρeLn = Cov eL , n /σeL σn (26)
where ρ eL,n is the Pearson product–moment correlation coefficient Cov(eL,n ) is

the co-variance of variables eL and n , σ eL and σ n are the standard deviation of eL
and standard deviation of n . In the proposed method, eL(n) is the estimate of the
uncorrupted signal at each stage and n (n) is the reference noise used at stage-I. Since
we have assumed that the information signal a(n) and the added noise are in time
uncorrelated, the value of ρ eL,n should be low. The estimated correlation function
ρ eL,n is investigated at each stage and further adaptive filter stages are added until
the value of ρ eL,n reaches a minimal threshold value at the optimal cascaded filter
stage.
The proposed filter structure is applied to signal de-noising application for extracting
the uncorrupted signal. The primary input signal to stage-I ANC is d 1 (n) = a(n) +
n(n) = Asin(2πf 1 ) + Bsin(2πf 2 ) where f 1 = 10 Hz is the low-frequency information
signal, and f 2 = 100 Hz is the high frequency additive noise signal. The amplitudes
of A and B are 1 and 0.5, respectively. The reference input signal to stage-I filter
is a noise signal correlated to n(n) given by n (n) = B1 sin (2πf 2 ) where B1 has
an amplitude of 0.35. The filter length is taken as 4, and the initial filter weights
are assumed to be zero. The value of μ is taken as 0.01. The performance criteria
utilized to validate the proposed multi-stage (M-S) cascaded LMS filter model are
the MSE, SNR and average noise reduction (ANR) (in dB). Figure 3 presents the
MSE performance for the above specifications such that the input SNR = 13 dB. The
speed of convergence is significantly improved and the MSE is further minimized
by using the proposed multi-stage cascaded LMS adaptive filter.
The proposed filter model is tested by adjusting the input SNR to be 21 dB and the
MSE performance of the proposed multi-stage cascaded LMS adaptive filter model
as compared to the existing cascaded models is presented in Fig. 4. We observe that
for an input SNR of 21 dB, the proposed filter attains minimum MSE value and faster
convergence speed.
From Table 1, we infer that the proposed multi-stage cascaded LMS adaptive
filter framework in ANC systems outperforms the existing single stage and 2-stage
cascaded LMS adaptive filter models in terms of SNR and ANR (average noise
reduction). The proposed adaptive filter structure in ANC gives an output SNR value
of at least 3–4 dB higher than the existing cascaded adaptive filters and achieves
Fig. 3 Convergence of MSE with iterations for an input SNR = 13 dB
Fig. 4 Convergence of MSE with iterations for an input SNR = 21 dB

Table 1 Comparison of MSE, SNR and ANR for various adaptive filter models
Input SNR Filter structure MSE SNR (dB) ANR (dB)
13 dB LMS adaptive filter 0.0328 27.23 29.46
Existing 2-S cascaded LMS filter 0.0265 29.35 30.03
Proposed M-S cascaded LMS filter 0.0200 32.16 32.62
21 dB LMS adaptive filter 0.0329 27.19 28.35
Existing 2-S cascaded LMS filter 0.0244 30.17 30.83
Proposed M-S cascaded LMS filter 0.0154 34.80 35.18
higher average noise reduction levels as compared to the use of conventional LMS
adaptive filter and 2-stage LMS cascaded adaptive filter structure in ANC systems.
4 Conclusion and Future Scope
A multi-stage cascaded LMS adaptive filter model is presented in this paper based
on automatic control of stages to be cascaded. The proposed multi-stage(M-S)
cascaded LMS adaptive filter model used in ANC systems offers an improved solu-
tion to achieve higher convergence speed and a minimal MSE value by automat-
ically adjusting the number of filter stages to be cascaded. From the simulation
results, we infer that that the proposed variable stage filter gives good performance
as compared to the conventional LMS and the existing 2-stage cascaded LMS adap-
tive filter structure. Using such structures, cost-effective hardware ANC systems can
be implemented in future.
References
1. Sayed AH (2003) Fundamentals of adaptive filtering (1st ed). Wiley Interscience

2. Sankaran SG, Beex AA (1997). Acoustic echo and noise canceler improvements for hands free
telephones. IEEE Southeastcon’ 97, New York 148–150
3. Sambur M (1978) Adaptive noise canceling for speech signals. IEEE Trans Acoust Speech Sig
Proc 26(5):419–423
4. Haykin S, Widrow B (2003) Least-mean-square adaptive Filters, 1st edn. Wiley, Newyork
5. Poularikas AD (2014) Adaptive filtering: fundamentals of least mean squares with MATLAB® ,
1st edn. CRC Press, Taylor and Francis Group
6. Ahmed N, Hush D, Elliott G, Fogler R (1984) Detection of multiple sinusoids using an adap-
tive cascaded structure. In: IEEE International Conference on Acoustics, Speech, and Signal
Processing, USA pp 199–202
7. Prandoni P, Vetterli M (1998) An FIR cascade structure for adaptive linear prediction. IEEE
Trans Signal Process 46(9):2566–2571
8. Kim H, Kim S, VanHelleputte N, Berset T, Geng D, Romero I, Yazicioglu RF (2012) Motion
artifact removal using cascade adaptive filtering for ambulatory ECG monitoring system. In:
IEEE biomedical circuits and systems conference, Taiwan, pp 160–163
9. Garcés Correa A, Laciar E, Patiño HD, Valentinuzzi ME (2007) Artifact removal from EEG
signals using adaptive filters in cascade. In: 16th argentine bioengineering congress and the
5th conference of clinical engineering, Journal of Physics, IOP Publishing
10. Dixit S, Nagaria D (2017) Design and analysis of cascaded LMS adaptive filters for noise
cancellation. Circ Syst Signal Proc 36(2):1–25
11. Maurya AK (2018) Cascade-cascade least mean square (LMS) adaptive noise cancellation.
Circ Syst Sig Proc 37(9):3785–3826
12. Awad A (2019) Impulse noise reduction in speech signal through multi-stage technique. Eng
Sci Technol Int J 22:629–636
THD Optimization with Low Switching
Frequency Control for 15-Level Reduced
Switch Asymmetric Multilevel Inverter
Gireesh Kumar Devineni, Aman Ganesh, Neerudi Bhoopal,

and D. S. N. M. Rao
Abstract The switching modulation techniques of high switching frequencies are

not recommended in medium voltage drive applications with a power of megawatts
because of high power losses and low converter performance. Fundamental strate-
gies like selective harmonic elimination (SHE) are a popular alternative. Since the
transcendental equations formulated by SHE are simply nonlinear in nature, it has
proven to be an important challenge for the scientists to achieve a viable solution
on the desired modulation index. This article presents the comparison analysis of
the GA and PSO approach with good initial guess, and its hybrid optimization of
PSO-GA is used for solution of the SHE equation set to various modulation index
values. The %THD generated at various modulation index values is also consistent
with the harmonic standards of IEEE 519-1992.
Keywords GA · PSO · PSO-GA · Hybrid Algorithms · THD · SHEPWM
1 Introduction
Multilevel inverter technology has grown in popularity in recent decades due to its
advantages of producing increased levels of output voltages using less power switch
requirements for medium voltage and high-power applications. Multilevel inverters
are currently available in three topologies: DC MLI, CHB MLI, and FCMLI. H-bridge
MLI is typically favored for various industrial uses among all these topologies [1–3].
Low-frequency switching and high-frequency switching controls are two types of
G. K. Devineni · A. Ganesh (B)

School of Electronics and Electrical Engineering, Lovely Professional University, Phagwara,
Punjab, India
G. K. Devineni · N. Bhoopal
Department of Electrical and Electronics Engineering, B V Raju Institute of Technology,
Narsapur, Telangana, India
D. S. N. M. Rao
Department of Electrical and Electronics Engineering, Gokaraju Rangaraju Institute of
Engineering and Technology, Hyderabad, India
https://doi.org/10.1007/978-981-16-4943-1_9
82 G. K. Devineni et al.
modulation techniques used to control multilevel inverters [4]. The following effects
can be caused by high frequency modulation in inverters, such as carrier-based pulse
width modulation schemes.
• Reduced system efficiency due to increased thermal losses.
• Motor bearing failure and insulation damage in case of drive control due to high
dv/dt.
• Effect of electromagnetic interference due to the power frequency of sidebands
in the order 10–30 kHz.
• Ripples and harmonics in voltage and current waveforms.
However, low-frequency switching modulation methods such as SHE have
key features, like the less thermal loss, control on harmonic o/p voltage, and a
better harmonic profile. To generate an optimal solution at a specific modulation
index for low percent of THD to satisfy the harmonic standards of IEEE 519-1992,
SHE methodology is suffering from the cumbersome nature to deal with the SHE
equations. This problem has received considerable attention in recent years, and
many computational algorithms are being used to address SHE equations to achieve
improved solutions with little time and effort in computation. The viable solutions
for the nonlinear SHE equations are sought using stochastic optimization techniques
such as genetic algorithms, species-based modified optimization techniques, and
firefly algorithms. Recently, Manasrah et al. [5] have suggested the new PSO-GA
algorithm with the combined merits of both GA and PSO algorithms, which incor-
porates the efficient, scalable, and compact coding of GA and PSO algorithms to
address the problem of workflow forecasting. This article compares the feasibility of
the PSO-GA algorithm with that of conventional GA and PSO algorithms in solving a
set of SHE equations that are developed in the control of single-phase switched-diode
15-level inverter [6].
2 Problem formulation for SHEPWM
The selective harmonic elimination technique is often termed the low-frequency

switching method according to harmonic elimination theory. It aims to find the suit-
able switching angles of the converter, to eliminate the selected lower harmonics,
so that THD can be mitigated on the load voltage and current. SHE has numerous
advantages, including good control on harmonics, lower thermal losses, and triple
harmonics that are multiple harmonics of the third order. The SHE methodology may
also be referred to as the coded PWM method and was typically used in application
areas including improvements to the power quality, effective medium voltage control,
HVDC, and distribution systems [7].
High-frequency switching methods like carrier-dependent PWM and SVPWM
generate bands across the carrier frequencies, resulting in a high THD percentage.
In addition to that, the system with a switched frequency of 500 Hz results in signif-
icant thermal loss if multilevel inverters are run at an output level of more than
THD Optimization with Low Switching Frequency … 83
1 MW. For this reason, one of the common strategies at MW rated power is the low-
frequency switching technique such as selective harmonic elimination. In the other
hand, for decades, scholars are forced to have potential alternatives to the transcen-
dental, nonlinear SHE set. The increase in voltage levels also increases the amount
of variables in SHE formulations, increasing the computing complexity [8]. Among
the most common topology is switched-diode asymmetric 15-level inverters for solar
PV applications. For analysis, thus, 3-source, 7-switch 15-level single-phase inverter
is considered in this article [9]. Figure 1 shows the proposed asymmetric 15-level
inverter circuit which has two units named primary circuit which is responsible for
level generation, and other is auxiliary circuit which is used for polarity reversal.
Using Fourier series analysis of 15-level output of the inverter in Fig. 1, the output
voltage equation is represented by
∞
4Vdc
V (ωt) = [cos(kα1 ) + cos(kα2 ) + · · · + cos(kαn )] sin(kωt) (1)
k=1,3,5,....
kπ
where n = number of switching angles.

The range of switching angles between 0° and 90° (0◦ < α < 90◦ ) is taken
as constrain in the SHE technique. Dividing the square wave output into Fourier
Level Generation Polarity Generation

Fig. 1 Proposed 15-level multilevel inverter with reduced number of switches
components of sine wave in compliance with the Fourier series, there are even no
harmonics due to its symmetry. The triplen harmonics were omitted by varying the
modulation index in multiples of 3. In terms of switching angles, it is necessary to
show the fundamental output voltage (V 1 ) as
4Vdc
[cos(kα1 ) + cos(kα2 ) + · · · + cos(kαn )] = V1 (2)
π
The maximum fundamental voltage V 1max can be written as follows:
4Vdc
V1max = (3)
π
Therefore, the modulation index M i can be represented as
π V1
Mi = (0 ≤ Mi ≤ 1) (4)
4Vdc
The switched-diode 15-level inverter consists of 7 degrees of freedom. One of

those seven is generated from fundamental voltages, and the other six degrees are
used to eliminate other odd harmonics of order: 5th to 19th. In multilevel inverters,
modulation processes may be categorized by switching frequency. The key purpose
is to improve maximum fundamental voltage by reducing harmonic content of lower
order. Combining Eqs. (2) and (4), the transcendental equations can be written as
⎫
cos(α1 ) + cos(α2 ) + cos(α3 ) + cos(α4 ) + cos(α5 ) + cos(α6 ) + cos(α7 ) = Mi ⎪
⎪
⎪
⎪
⎪
cos(5α1 ) + cos(5α2 ) + cos(5α3 ) + cos(5α4 ) + cos(5α5 ) + cos(5α6 ) + cos(5α7 ) = 0 ⎪
⎪
⎪
⎪
⎪
⎬
cos(7α1 ) + cos(7α2 ) + cos(7α3 ) + cos(7α4 ) + cos(7α5 ) + cos(7α6 ) + cos(7α7 ) = 0
(5)
cos(11α1 ) + cos(11α2 ) + cos(11α3 ) + cos(11α4 ) + cos(11α5 ) + cos(11α6 ) + cos(11α7 ) = 0⎪⎪
⎪
⎪
⎪
⎪
cos(17α1 ) + cos(17α2 ) + cos(17α3 ) + cos(17α4 ) + cos(17α5 ) + cos(17α6 ) + cos(17α7 ) = 0⎪⎪
⎪
⎪
⎭
cos(19α1 ) + cos(19α2 ) + cos(19α3 ) + cos(19α4 ) + cos(19α5 ) + cos(19α6 ) + cos(19α7 ) = 0
Fitness can be achieved by adding the intended objective to the penalty factor. The
consistency of the solution for the initial target is the fitness value. The principle goal
here is to obtain a set of switching angles to achieve the optimal value of the magnitude
of the fundamental harmonic. The fitness function is determined accordingly for each
solution.
3 Control Algorithms
The switching angles of proposed inverter are optimized by nature-inspired opti-

mization algorithms. The following algorithms of GA, PSO, and hybrid GA-PSO
algorithms were used in this research for solving the nonlinear equations generated
from SHEPWM as given in Eq. (5).
3.1 Genetic Algorithm
Genetic algorithms (GAs), based on biological evolution, are computerized search

methods. Genetic algorithms are based on the population size of pre-selected candi-
dates. The implementation of genetic algorithms takes place following phases for
selective harmonic elimination problem [10, 11].
1. Initialization: Initial population of any candidate solution in the search space
can be developed by the following equation, close to the middle of each switching
boundary. For each solution set, the number of populations was calculated as 20.
⎡ ⎧ ⎫⎤
⎨ αiUj − αiLj ⎬
αiIPj = αi j + ⎣αiLj ± rand j ⎦ (6)
⎩ 2 ⎭
where αiIPj represents initial population matrix, and αiLj represents initial guess of
solutions from α1 to α7 .
2. Evaluation: The fitness values of the candidate solutions are determined using
the pre-formulated objective function as seen in the equation until the population for
each switching angle is initialized or the offspring population is set.
1
FOBJ (α) = (7)
1 + F(α)
where
F(α) = mod f 1 (α) + mod f 2 (α) + mod f 3 (α) + · · ·
3. Selection: The selection to produce offspring was made by assigning more

instances of these solutions with higher fitness values to the candidate’s solutions
and applying a survival mechanism. This was accomplished by calculating the fitness
of each population and by comparing it with a random number of (0, 1) that is
determined by the collection of roulette wheels.
4. Crossover or Recombination: Two or three parental solutions are summarized
to create new solutions, which can be theoretically enhanced (e.g., offspring). Any
solution is converted to an 11-bit binary, pairs have been randomly selected from
parental crossover solutions, and a random number ‘r’ has been generated for each
pair in a random version of (1, 1) to align ‘r’ with a specified probability crossover
(here 0, 6) to establish whether or not the pair is to be perforated, over point was
found. A single-point crossover was used here.
5. Mutation: Although two or more chromosomes are recombined, localized yet
random mutations change the solution. Here too, for every descendant a random
number ‘r’ is created, in which ‘r’ contrasts with the pre-defined probability of
mutation (here 0, 1). If the mutation was deemed suitable, a random number was
established between (1.10) to determine the mutation rate, complementing this single
bit.
6. Replacement: To replace the parents of the second generation, the first
generation’s parental population was selected, recombined, and mutated.
7. Termination: If the termination criterion is not reached, then repeat steps 2 to
6.
3.2 Particle Swarm Optimization
Kennedy and Eberhart suggested PSO in 1995, and it is defining Swarms’ sociological
behavior [12]. The phases in PSO algorithm for solving SHEPWM problem are given
below.
Step 1: Initialize the parameters of particle vectors X i , V i , Pbest , Gbest , and inertia
weight of the particle C 0 . Choose number of generations as 100, size of population
as 40, cognitive parameter C 1 as 0.5, and social parameter C 2 as 1.25.
Step 2: Test the conditions for 0 < (C1 + C2 ) < 2 and (C1 + C2 )/2 < C0 < 1.
The system would then be guaranteed to converge to a stable equilibrium position if
the two conditions were met. If false, go to Step 1.
Step 3: The new position and velocity vectors of particles were determined using
the following equation.

vli (t + 1) = w(t).vli (t) + Cind .rand1, ( pli − xli (t) + Csoc .rand2. G i − xli (t) (8)
Then, the new position is defined as
xli (t + 1) = xli (t) + vli (t + 1) (9)
Step 4: Evaluate the objective function of the particles using

4
V ∗ − V1 N
1 Vh 2
OF = min 100 ∗ 1 ∗ + 50 ∗ k (10)
αk V1 h V1
K =2 k
Step 5: Check for the constraint on fitness function as

π
α1 < α2 < α3 < α4 < α5 < α6 < α7 < (11)
2
Step 6: Check for P(xi) < P(Pi), if not then i = i + 1 go to step 3.
Step 7: Update the particle’s best local position if the best local position is better
than before. Thus Pi = Xi replaces the local best position.
Step 8: Pg = min (P neighbor).
Step 9: Terminate the process if the optimal switching angles achieved.
3.3 PSO-GA Hybrid Algorithm
A hybrid approach is proposed which combines genetic algorithm (GA) with PSO.
This hybrid algorithm combines GA and PSO principles and produces individuals
not only through the GA crossover and mutations but also through PSO processes
in the new generation. It can solve the problem of local minima of the PSO and has
greater search accuracy.
The algorithm of GA_PSO with combined accuracy is given below:
Step 1: Initialization of variables for GA.
Step 2: Initialization of variables for PSO.
Step 3: Arrange the PSO group, and determine the optimal GA solution coefficient
for each condition.
Step 4: Impose GA’s optimal location condition, and perform GA mutations and
crossover.
Step 5: Select the best solution and stop when the best individual in GA or PSO
has met the termination criteria.
Step 6: If the PSO’s condition is met (target value or iteration number), the
reproduction process is stopped. If not, it will go step 3.
Step 7: Perform a hybrid method if generations could be precisely segregated by
the iterative items N. Select P individuals arbitrarily according to their fitness from
each sub-system.
Figure 2 presents the comparison between convergence characteristics of GA,
PSO, and GA-PSO hybrid algorithms. The convergence characteristics show the
number of iterations required for a given algorithm to evaluate the fitness function.
Compared to GA and PSO, the hybrid PSO-GA algorithm converges at a faster rate
for the 65 iterations.
Fig. 2 Objective function versus number of iterations

4 Results and Discussions
Using MATLAB Simulink, the suggested asymmetrical structure is developed

with GA, PSO, and combined PSO-GA hybrid algorithms, as well as the SHE PWM
technique. The proposed asymmetrical structure’s input DC sources are 37 V, and the
measured peak voltage (Vpeak) is 259 V. The desired inverter’s switching frequency
is 50 Hz, the harmonic frequency is 1 kHz, and the Nyquist frequency with total
harmonics is 5 kHz. The RL load is considered with R = 26.8 and L = 9.9 mH.
For a modulation index of 0.9, the switching angles are obtained. The proposed asym-
metric multilevel inverter’s switching pulses are shown in Fig. 3 for both primary
circuit (level generation) and auxiliary circuit (polarity generation). The proposed
inverter can generate 15 levels of output voltage with 7 steps in each half cycle, each
having a voltage magnitude of 37 V.
Figure 4(a) represents load voltage waveform, and Fig. 4(b) represents the load
current waveform of 15-level output. The THDs of voltage and current harmonic
waveforms were evaluated (Figs. 5 and 6) for GA, PSO, and PSO-GA hybrid algo-
rithms. The PSO-GA algorithm has given a less voltage THD of 5.34% and current
THD of 2.66% compared to GA and PSO algorithms. The comparative analysis of
Fig. 3 Gating pulses for the primary and auxiliary circuit
Fig. 4 a Load voltage and b Load current waveforms of proposed inverter

Fig. 5 Voltage harmonic distortion a GA b PSO and current harmonic distortion c GA d PSO
Fig. 6 a Voltage harmonic distortion with PSO-GA b Current harmonic distortion with PSO-GA
these three algorithms for % voltage and current harmonic distortion is given in
Fig. 7.
Fig. 7 Comparison of THD

for GA, PSO, and their
hybrid PSO-GA algorithm
5 Conclusions
This work focused primarily on comparative analysis of GA, PSO, and their hybrid
PSO-GA algorithms for solving nonlinear transcendental SHE equations, which were
built in the single-phase asymmetric inverter to achieve the appropriate switching
angles. PSO-GA is a robust and efficient algorithm that has solved SHE equation
set from 0.5 to 1.0 modulation index values with less computational effort. The
comparative analysis is done for the total harmonic distortion in an asymmetric 15-
level inverter for different modulation indexes using SHEPWM control. The THD
estimated at the inverter output using PSO-GA algorithm is 5.34% without using any
filter at the inverter output, which is comparatively less than the individual GA and
PSO algorithms. Hence, this research concludes that the PSO-GA hybrid algorithm
will give optimal switching angles for the 15-level asymmetric multilevel inverter
using (SHEPWM) low-frequency switching control.
References
1. Rodríguez J, Lai JS, Peng FZ (2002) Multilevel inverters: a survey of topologies, controls, and
applications. IEEE Trans Ind Electron. https://doi.org/10.1109/TIE.2002.801052
2. McGrath BP, Holmes DG (2002) Multicarrier PWM strategies for multilevel inverters. IEEE
Trans Ind Electron. https://doi.org/10.1109/TIE.2002.801073
3. Mekhilef S, Abdul Kadir MN (2011) Novel vector control method for three-stage hybrid
cascaded multilevel inverter. IEEE Trans Ind Electron https://doi.org/10.1109/TIE.2010.204
9716
4. Mittal N, Singh B, Singh SP, Dixit R, Kumar D Multilevel inverters: a literature survey on
topologies and control strategies. https://doi.org/10.1109/ICPCES.2012.6508041
5. Manasrah M, Ali HB (2018) Workflow scheduling using hybrid GA-PSO algorithm in cloud
computing. Wirel Commun Mob Comput 2018. https://doi.org/10.1155/2018/1934784
6. Bharath VS, Gopinath M (2014) Tumbling total harmonic distortion in multi inverter with the
aid of hybrid (GA-PSO). J Theor Appl Inf Technol 62(1):131–138
7. Devineni GK, Ganesh A (2020) Problem formulations, solving strategies, implementation
methods & applications of selective harmonic elimination for multilevel converters. J Eur des
Syst Autom 53(6):939–952. https://doi.org/10.18280/jesa.530620
8. Manohar VJ, Trinad M, Ramana KV (2016) Comparative analysis of NR and TBLO algorithms
in control of cascaded MLI at low switching frequency. Procedia Comput Sci 85(Cms):976–
986. https://doi.org/10.1016/j.procs.2016.05.290
9. Kumar DG, Ganesh A, Naga Malleswara Rao DS (2021) Design and analysis of a novel
cascaded 15-level asymmetric inverter using PSO and whale algorithms. In: 2021 international
conference on sustainable energy and future electric transportation (SEFET), Hyderabad, India,
2021, pp 1–6. https://doi.org/10.1109/SeFet48154.2021.9375752
10. El-Naggar K, Abdelhamid TH (2008) Selective harmonic elimination of new family of multi-
level inverters using genetic algorithms. Energy Convers Manag 49(1):89–95. https://doi.org/
10.1016/j.enconman.2007.05.014
11. Anjali Krishna R, Padmasuresh L, Muthukumar P (2018) Genetic algorithm based 15 level
multilevel inverter with SHE PWM. Int J Eng Technol 7(2.33) Special Issue 33:893–897
12. Rasheed M, Omar R, Sulaiman M, Halim WA (2019) Particle swarm optimisation (PSO)
algorithm with reduced numberof switches in multilevel inverter (MLI). Indones J Electr Eng
Comput Sci 14(3):1114–1124. https://doi.org/10.11591/ijeecs.v14.i3.pp1114-1124
Signal Conditioning Circuits for GMR
Sensor in Biomedical Applications
G. Anand, T. Thyagarajan, B. Aashique Roshan, L. Rajeshwar,

and R. Shyam Balaji
Abstract Superparamagnetic nanoparticles offer a wide range of applications in the

emerging field of electronic devices. Recent developments have identified them as
components for a new type of magnetoresistance sensor. A novel measuring method
is used where increased field sensitivity is bought at the value of an inherent device
noise. The proposed GMR sensor has many advantages, yet its output is feeble due
to noise, and hence, there is a requirement for designing a signal conditioning circuit
to improve the signal strength. The objective is to analyze the low concentration
(sample) with a better signal conditioning circuit for the GMR sensor.
Keywords Superparamagnetic nanoparticles · Magnetoresistance sensor · Signal

conditioning circuit
1 Introduction
Nanotechnology and microtechnology have rapid progress and influence over the
scientific developments which improve the standard of life [1]. Superparamagnetic
nanoparticles have a good vary of applications in a trend of electronic devices [2].
Recent developments have identified them as components for a replacement style of
magnetoresistance sensor. Resistive sensors are an inevitable part of many instrumen-
tation systems and play an essential role in the measurement of physical quantities
like pressure, force, temperature, etc., [3]. Resistive sensors range from the poten-
tiometer, strain gauge (traditional types) to anisotropic magnetic resistance (AMR)
and giant magnetoresistance (GMR) sensor (conventional) [4]. To improve sensi-
tivity, accuracy, compatibility, and reduced size with electronic systems, magnetic
field measurement has been developed [5]. GMR sensors have taken a most important
role in the field of magnetic sensing due to their high signal strength, quick response
under a lower magnetic field, small size, high sensitivity, frequency response, reduced
power consumption, and low cost [6, 7]. These merits of GMR sensors are coun-
terbalanced with a demerit that makes it hard to assess the accurate reading at the
G. Anand (B) · T. Thyagarajan · B. Aashique Roshan · L. Rajeshwar · R. Shyam Balaji

Department of Instrumentation, Anna University, MIT Campus, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_10
94 G. Anand et al.
output. The output obtained will be feeble [8]. To overcome this demerit, there is a
need to design an advanced signal conditioning circuit.
A basic instrumentation system consists of a sensor, signal conditioning unit, and
process unit. A non-electrical signal is converted to an electrical signal to the signal
conditioning unit by the sensing element [9]. The signal conditioning unit converts
the electrical signal to the anticipated voltage level. Accurate signal conditioning
is a key to any industrial data acquisition or industrial control application. Many
sensors do not provide a clean and clear signal due to various reasons such as elec-
trical or RF noise or wire loss. Such signals require a signal conditioning unit before
being captured and processed by a data acquisition device [10]. Signal conditioning
includes various processes, but a number of the foremost common are amplification,
isolation, filtering, coupling, and linearization. To this, the main aim of this paper is
to design a signal conditioning unit for a GMR sensor by comparing and analyzing
a couple of filters, AC/DC coupling, and amplifiers. LTspice® simulation tool is
employed for circuit analysis and implementation. LTspice® is a high-performance
SPICE simulation software, schematic capture, and waveform viewer with enhance-
ments and models for easing the simulation of analog circuits. LTSpice is a flexible,
accurate, and free circuit simulator available for Windows and Mac [12]. It is easy-
to-use circuit designing software and may be applied for analysis for sort of circuit.
It is less time-consuming, and all circuits can be executed accurately.
2 Methodology
To design a signal conditioning unit, the schematic arrangement and selection of

active and passive elements of a circuit play a foremost role. It influences the overall
functioning of the system. The success of the designed measurement system depends
on the design and performance of the signal conditioning unit [11]. The selection of
filters and amplifiers was decided by the electrical output characteristics. The filters
selected for comparison are the band-stop filter, band-reject filter, and notch filter.
The instrumentation amplifiers used for comparison are OP07, AD822, and INA118.
Qualitative and quantitative analyses were performed, and therefore, the preferred
amplifier and filter were selected.
2.1 Comparison of Instrumentation Amplifiers
OP07 has an exceptionally low input offset voltage, and high open-loop gain makes
it beneficial for high-gain instrumentation applications. The minimal input voltage
range of ±13 V was attained with a high CMRR of 106 dB and high accuracy due
to its high input impedance.
Figure 1 describes the OP07 amplifier circuit implemented using the LTspice®
simulation tool.
Signal Conditioning Circuits for GMR … 95
Fig. 1 OP07 amplifier circuit implemented
The AD822 is a low-power FET input op-amp, dual precision which will func-
tion from a single supply of 5 V till 30 V. It has a single-supply capacity with an
input voltage range extending below the negative rail, allowing the AD822 to adapt
input signals below ground. Output voltage swing extends to within 10 mV of each
rail, providing the maximum output dynamic range. Figure 2 illustrates the AD822
amplifier circuit implemented using the LTspice® simulation tool.
The INA118 is an instrumentation amplifier for general purpose with excellent
accuracy and low power consumption. It is a three op-amp design, flexible, and small
that makes this device an excellent choice for a wide range of applications. Current-
feedback input provides wide bandwidth, even at high gain. Internal input protection
could withstand till ±40 V devoid of loss. This amplifier has been utilized to have
Fig. 2 AD822 amplifier circuit implemented

96 G. Anand et al.
Fig. 3 INA118 amplifier circuit implemented
low offset voltage, high common-mode voltage, and drift voltage. Figure 3 explains
the INA118 amplifier circuit implemented using the LTspice® simulation tool.
2.2 Comparison of Filters
Band-stop filter (BSF) is a frequency selective circuit that operates exactly in reverse
to the band-pass filter. The band-stop filter is called a band-reject filter which passes
all frequencies apart from those within a specified stopband. If the stopband is highly
attenuated and very narrow for few hertz, the band-stop filter is known as a notch
filter. Figure 4 refers to the filter implementation using the LTspice® simulation tool.
A band-rejection filter allows to pass most of the frequencies unaltered, but it
mitigates a specific range to extremely low levels. It operates exactly in reverse to
the band-pass filter. The word band rejection refers to the deliberate removal of a
known frequency in signal processing. Figure 5 refers to the filter implementation
using the LTspice® simulation tool.
Notch filters are similar to the band-pass filter that has a combination of low-
and high-pass filters in the circuit design. The distinction is that it features a twin-T
parallel resistance-capacitance (RC) network as in Fig. 6 circuit to acquire a deep
notch. A professionally designed notch filter can produce an attenuation of more than
−60 dB at the notch frequency. It eliminates a frequency of a narrow band and allows
frequencies above and below of that band. This filter is also known as a band-reject
filter as it eliminates frequencies. Figure 6 refers to the filter implementation using
the LTspice® simulation tool.
Fig. 4 Band-stop filter implementation
Fig. 5 Band-reject filter circuit implementation
2.3 Analysis of the Instrumentation Amplifiers
Quantitative Analysis: The parameter considered to compare the above circuits is

power dissipation.
Power dissipation is a measure of the rate at which energy is dissipated, or lost,
from an electrical system. The internal energy increases, causing its temperature to
98 G. Anand et al.
Fig. 6 Notch filter implementation
rise above the ambient (surrounding) temperature. This causes energy to dissipate
away from the conductor into the surroundings, through the process of heat transfer.
Qualitative Analysis: The desired instrumentation amplifier is selected based on
the power dissipation which provides the high output gain amplification.
2.4 Analysis of the Filters
Quantitative Analysis: The parameters considered to compare the above circuits are
signal-to-noise ratio (SNR) and common-mode rejection ratio (CMRR).
Qualitative Analysis: The total harmonic distortion (THD) is the parameter used
in this analysis. Harmonic distortion is a common form of distortion in applications
where the peaks of the output signal get “clipped.” As listed in Table 2, a filter with
a lower percentage of THD is preferable.
2.5 AC/DC Coupling
The alternating coupling circuit in Fig. 7 permits only AC signals to pass through a
connection, and it removes the DC offset by a DC blocking capacitor. Normalizing
the signal to a mean of zero, the DC component is rejected by the AC coupling.
Fig. 7 AC coupling circuit implementation
The direct coupling circuit in Fig. 8 permits both AC and DC signals to pass
through a connection. The DC element is a 0 Hz signal which acts as an offset about
which the AC component of the signal fluctuates.
Fig. 8 DC coupling circuit implementation

100 G. Anand et al.
3.1 Comparison and Analysis of Instrumentation Amplifiers
The output of the OP07 and AD822, which give less gain amplification, is shown
in Figs. 9 and 10. From Fig. 11 there is a high gain amplification in INA118 which
Fig. 9 OP07 amplification circuit output
Fig. 10 AD822 amplification circuit output

Fig. 11 INA118 amplification circuit output
serves the purpose of amplifying the feeble signal which is required. Hence, the INA
118 is selected as the preferred Instrumentation Amplifier for future processes.
Thus, by analyzing the above three circuits’ power dissipation from Table 1, it is
found out that INA 118 instrumentation amplifier satisfies both quantitatively and
qualitatively.
Fig. 12 Band-stop filter output

102 G. Anand et al.
Fig. 13 Band-reject filter output
Fig. 14 Notch filter output
Table 1 Power dissipation

Instrumentation amplifier Power dissipation (mW)
OP07 227
AD822 162
INA118 90
Table 2 Total harmonic

For 50 Hz SNR (in dB) CMRR (in dB)
distortion
Band-stop filter 19.5 −63
Band-reject filter 42 −65
Notch filter 6 −67.5
3.2 Comparison and Analysis of Filters
The output of the BSF and BRF which allows or completely do not reject the noise,
are shown in Figs. 12 and 13. From Fig. 14 there is a sharp curve in the notch filter
output which rejects the noise at that frequency that is exactly required. Hence, the
notch filter is selected for future processes. Thus, from analyzing the above Table 2
it is found that the Notch filter satisfies both Quantitatively and Qualitatively.
3.3 AC/DC Coupling
From Figs. 15 and 16, AC/DC coupling is used to eliminate the DC offset voltage,
and hence, sensor output always generates the actual output without the necessity
for any biasing.
Based on the studies carried out for various analog sections, the final signal condi-
tioning circuit has been formulated and simulated as shown in Fig. 17. The output
of the simulation in Fig. 18 provides a better result for the GMR sensor compared
with any signal conditioning circuits.
Fig. 15 AC coupling output

104 G. Anand et al.
Fig. 16 DC coupling output
Fig. 17 Signal conditioning circuit implementation
4 Conclusion
The output from GMR sensors is very feeble which might be affected by the noise.
Hence, there is a need for a signal conditioning circuit for better output. Based on
various simulation studies for noise filtering, AC and DC coupling and amplification
circuits for biomedical applications have been carried out, and the desired signal
conditioning circuit is finalized to overcome the demerit of the GMR sensor. Now,
signal conditioning circuit is ideally implemented in a software simulation platform.
Further, it will be implemented in real time.
Fig. 18 Signal conditioning circuit output
References
1. S. Cardoso et al (2017) Challenges and trends in magnetic sensor integration with microfluidics
for biomedical applications. J Phys D Appl Phys 50(21). https://doi.org/10.1088/1361-6463/
aa66ec.
2. Park J (2015) Superparamagnetic nanoparticle quantification using a giant magnetoresistive
sensor and permanent magnets. J Magn Magn Mater 389:56–60. https://doi.org/10.1016/j.
jmmm.2015.04.049
3. Murzin D et al (2020) Ultrasensitive magnetic field sensors for biomedical applications. Sensors
(Switzerland) 20(6):1–32. https://doi.org/10.3390/s20061569
4. Chugh VK, Kalyan K, Anoop CS, Patra A, Negi S (2017) Analysis of a GMR-based plethys-
mograph transducer and its utility for real-time blood pressure measurement. Proc Annu Int
Conf IEEE Eng Med Biol Soc EMBS pp 1704–1707. https://doi.org/10.1109/EMBC.2017.803
7170
5. Bernieri GB, Ferrigno L, Laracca M (2013) Improving performance of gmr sensors. IEEE Sens
J 13(11):4513–4521. https://doi.org/10.1109/JSEN.2013.2271275
6. Daughton JM (2000) GMR and SDT sensor applications. IEEE Trans Magn 36(5):2773–2778.
https://doi.org/10.1109/20.908586.
7. Smith CH, Schneider RW (1998) Magnetic field sensing utilizing GMR materials. Sens Rev
18(4):230–236. https://doi.org/10.1108/02602289810240592
8. Zhang D, Pan Z, Zhou H, Zhang W (2016) Magnetic sensor based on giant magneto-impedance
effect using the self-regulating technology on the bias magnetic field. Sens Actuators, A Phys
249:225–230. https://doi.org/10.1016/j.sna.2016.09.005
9. Bhaskarrao NK, Anoop CS, Dutta PK (2017) A simple signal conditioner for tunneling
magneto-resistance based angle sensor. In: 2016 IEEE annual India conference (INDICON),
pp 1–6, 2017. https://doi.org/10.1109/INDICON.2016.7839065
10. Elangovan K, Anoop CS (2020) A digital signal-conditioner for resistive sensors and its utility
for linearizing GMR-based magnetometer. 2020 IEEE Sens Appl Symp SAS 2020 - Proc,
https://doi.org/10.1109/SAS48726.2020.9220031
11. Juárez-Aguirre R et al (2013) Digital signal processing by virtual instrumentation of a MEMS
magnetic field sensor for biomedical applications. Sensors (Switzerland) 13(11):15068–15084.
https://doi.org/10.3390/s131115068
106 G. Anand et al.
12. Aminudin RDH, Iryanti M (2019) The characterization of giant magnetoresistance sensor for
prototype of bridge deflection measurement. J Phys Conf Ser 1280(2). https://doi.org/10.1088/
1742-6596/1280/2/022065
Modeling and Simulation of SPVGC
System Using Conventional Controller
P. Chandra Babu , B. Venkata Prasanth, and P. Sujatha
Abstract This article addresses the system connected to the solar PV grid by the
DC to DC converter and inverter (VSC) for the supply of electricity to the power
grid, addressing the major device elements, which include the first one 20-kW solar
array, second one grid side inverter and third one boost converter, mathematical
modeling and THD change by irradiation effect. This paper begins with a PV system
model, simulation circuit and discussion. This paper has a concise overview of all
part included in this system and considers environmental temperature and solar radi-
ation practical data at BVRIT campus. It also addressed a common algorithm for
MPP monitoring. The results of simulations demonstrate how the variations in solar
radiations will influence the production of any PV module and shows THD of the
photovoltaic-grid connected.
Keywords DC–DC Converter · THD · P&O · SPV · VSC inverter
1 Introduction
Photovoltaic (PV) is the name of a method to convert solar energy into direct elec-
trical energy using semi-conductive materials which exhibit the photovoltaic effect, a
physical, photochemical and electrochemical phenomenon commonly studied. Solar
panels consisting of multiples of solar cells can be used for supplying effective solar
P. Chandra Babu (B)

Department of Electrical and Electronics Engineering, JNTUA College of Engineering,
Anantapuram, India
Department of Electrical and Electronics Engineering, B V Raju Institute of Technology,
Narsapur, Telangana, India
B. V. Prasanth
Department of Electrical and Electronics Engineering, QIS College of Engineering & Technology,
Ongole, A.P, India
P. Sujatha
Department of Electrical and Electronics Engineering, JNTUA College of Engineering,
Anantapuram, A.P., India
https://doi.org/10.1007/978-981-16-4943-1_11
108 P. Chandra Babu et al.
power in a photovoltaic system. The first stage includes the photoelectric effect from
which a second electrochemical mechanism takes place involving the ionization of
crystallized atoms in a series, producing an electric current.
Solar panels have variety of solar cells with different materials are used in PV
power generation. Like silicone-based monocrystalline, polycrystalline, amorphous,
copper indium sulfides and cadmium telluride products currently used in photovoltaic
content. Solar cable copper links modules (cable modules), arrays (cable arrays) and
subfields. The development of different solar cells has progressed rapidly in recent
decades due to the rising demands for green energy [1, 2].
The lower the cost of solar energy, the more favorable it is relative to conventional
energy, and the more attractive it is to electricity and energy users all over the world.
In California, solar power is currently cheaper than most other peak generators,
including those operating on low-cost natural gas, at prices far below USD 100/MWh
($0.10/kWh). Lower energy costs also boost demand for energy consumption, as
the cost of renewable energy is very attractive relative to retail electricity prices. PV
prices dropped considerably by 2011 and will continue to fall. The solar cell’s average
retail price under Solar buzz has fallen from $3.50 per watt to $2.43 per watt during
2011[3]. The “Best Class” price was expected to decline to $0.50/Watt at the end of
2012 and was anticipated to decline to $0.36/Watt before 2017. Many locations PV
utilization increased and is usually defined as photovoltaic cost at or below the retail
energy prices. Further, the conversion efficiency from solar to electricity is very low,
with a figure of only 9–17%. So, similarly MPPT is a key part of a solar pv grid-based
system that ensures that the maximum available energy is always extracted from the
PV panel at all times and driven to the AC grid, an ideal infinite power sink [1, 2,
4–12]. The literature presents various MPPT algorithms, including P&O and Hill
Climbing (H&C) [4], grid-related PV-connected power plants which operate parallel
to conventional electrical sources and are primarily used as grid power supplies
[4, 13].
The simplest grid connection system is implemented in MATLAB/Simulink, this
model is similar to all practical residence use model having a PV array, and a converter
unit connected at point of connection. The 3-phase grid-connected PV system with a
widely used perturb and observe technique [4, 5] for maximum power point tracking
(MPPT) used to evaluate model performance under variable irradiation condition.
2 System Description of PV Grid-Connected System
Figure 1 displays the components used in the proposed model of 20 kW PV arrays

connected to a 25 kV grid. Perturbation and observation technology is being introduce
for MPP. The major devices are (1) The utilities grid. (2) PV arrays with a power
of 20 kW. (3) DC–DC boost converters (magenta block) which are used to raise the
PV output to 500 V DC. (4) 3-phase VSC inverter (orange blocks) which is used
to transform the DC voltage supplied by arrays (500 V) to an AC voltage of 260 V.
Modeling and Simulation of SPVGC … 109
Fig. 1 PV system with grid connection
(5) A 100 kVA three-phase coupling transformer. 260/25 kV. (6) Capacitor banks
10 kVar used to damp out the harmonics generated by the VSC.
3 Modeling of PV Array and MPPT
3.1 PV Array Modeling
The solar cell is the solar panel’s building block. The mathematical model is devel-
oped and derived to better understand the PV module. The performance characteris-
tics of the PV module are highly dependent on three important factors, namely solar
radiation, cell temperature and PV module output voltage.
The mathematical modeling of the solar panel (module) is evaluated in order to
obtain the performance characteristics of the PV module. The parameters of voltage,
current and power output vary according to radiation, temperature and load current.
The impact of these three variations is taken into account in the simulation process
such that temperature and solar shifts arise. In view of PV cell output power (P) and
voltage, the PV cell performance is calculated under solar (S) radiation and temp.
(T). Relevant equations can be obtained on the basis of Kirchhoff’s first law. The
performance characteristics of PV cell models can be described in these equations.
3.1.1 Ideal Solar Cell Model
Analogous circuit in illustration 2 can be interpreted as an ideal solar cell. A solar

cell equation shown in Eq. 1 defines the IV characteristic.
I = Iph − Io (ekTqv − 1) (1)
3.1.2 Practical Solar Cell Model
The I–V properties of a true solar cell differ from the properties of the ideal solar
cell. The functional solar cell can have a set of parallel resistors (circuit see Fig. 2).
I = Iph − Id − Ish (2)
The current flow creates voltage drop in the elements of the identical circuit.
Voltage is obtained between the diode and resistor is
Vj = V + IRc (3)
The photon generated current can be expressed by
Iph = NpIg (4)
Ig = NpIg (5)
Icc Gi
IG = + I (Tt − Tr ) (6)
1000
The current transferred more to the diodes, after the series resistance is ignored.

qV 0
I D = N p Icat exe −1 (7)
Nc AEkT
Fig. 2 PV cell equivalent

circuit
In the above-mentioned equation, the saturation rate is the Icat, A is the ideal
component, k is the Boltzmann constant, N c the series number of cells, q is the
electron load, V 0 is the PV module output voltage, N p is the number of cells.
Reverse saturation current (RSC) is provided

T3 q Eg 1 1
Isat = Ior exp − (8)
Tr kT Tr T
The current is delivered by the shunt resistor
Vo
Ish = (9)
Ns Rsh
The complete output current from the PV module is, substitute Eqs. 7, 8 and 9 in
2

q Vo
I = N p Ig − N p Isat exp − 1 − Ish (10)
Ns AkT
The cell diode saturation current depends on the cell temperature and is described
by

T3 q Eg 1 1
Isat = Ior exp − (11)
Tr kT Tr T
The fill factor of solar cell is calculated as a division of the maximum power that
is converted into electrical power and described as
VMP IMP
Fill Factor = FF = (12)
VOC ICC
Solar cell efficiency is calculated as a fraction of converted power to input power,
see Eqs. 13 and 14
Pmax = Voc Isc FF (13)
Voc ∗ Isc FF
η= (14)
Pin
In the above equation, V oc , ISC, FF are the voltage of the opening circuit, short-
circuit current and fill factor, η is the efficiency. Manufacturer specifications for the
module «1Soltech 1STH-215-P» are listed in Table 1, the 20 k-Watts grid-connected
PV array composed of 10 parallel strings and each string having 10 series connected
modules, total 100 modules connected, each one produce 213.15 W, total power is
21.315 KW, i.e., (10 * 10 * 213.15 W = 21.315 kW).
Table 1 PV module
Selected parameters Values
configuration details
Parallel String 10
Series connected modules per string 10
Module 1Soltech 1STH-215-P
Max. Power 213.15 W
Voc 36.3 V
Isc 7.84 A
Vmp 29 V
Imp 7.35 A
Ncell 60
3.2 Maximum Power Point Tracking (MPPT)
PV module efficiency is an issue that limit the large utilization of PV systems so

a DC–DC is installed into any MPPT system using an MPP controller. It can run
the PV systems for a certain set of conditions at the maximum power point and
optimize the array performance. This is achieved by the MPPT by continuously
regulating the voltage or current irrespective of the attached load attached. As a result,
several MPP tracking (MPPT), the most common method used in the present analysis,
was developed and implemented, namely perturb and observe [3, 13–15]. MPPT
techniques vary in complexity, needed sensors, convergence speed, cost, efficiency
range, hardware implementation, popularity, and in other respects [3, 15].
The ability to track the maximum power point (MPP) for the uniform solar irradi-
ation has proved traditional MPPT technique. However, traditional techniques have
struggled to detect the true MPP under rapidly changing environments and partly
shaded conditions [12]. The P&O and H&C are widely used and popular because
they are easy to apply [3, 14] in this paper, P&O is considered for implementing
MPPT in PV system.
In general, the following properties should be given by a good MPPT algorithm,
which also succeeds in partial shading conditions [11]. (1) Monitoring MPP easily
to achieve high performance, (2) Simple to use with a low device load, (3) The need
for fewer and less costly sensors (removing boost conversion sensors reduces cost
significantly) (4) Impose minimum disruption at grid side. [11]
3.2.1 P&O MPPT
The output voltage of the PV array is interrupted by a small rise in the P&O algorithm
resulting in a power shift (Del P). When the Del P is positive, the voltage adjusts
to the MPP. It was in the same direction. If the negativity of Del P is reached, the
device is not running at the optimum location, so it is necessary to reduce the size of
the device to return the operating point to the MPP [13].
3.3 Boost Converter
The diagram shows the simple boost converter in Fig. 3. It is used if the output voltage
is greater than the input voltage. The inductor current flows through the diode with
V x = V o, while the transistor is ON V x = V in . It is assumed that the current is still
flowing for this analysis (continuous conduction). The voltage around the inductive
system is shown in Fig. 4, and the average voltage must be zero for the constant
average current.
Vin ton + (Vin − V0 )toff = 0 (15)
You can change this to the following
Vo T 1
= = (16)
Vin toff (1 − D)
Fig. 3 Boost converter circuit
Fig. 4 Voltage and current

waveforms
and the power balance provides for a lossless circuit is
IO
= (1 − D) (17)
Iin
The voltage output must be greater than the input voltage, as the duty ratio “D” is
0 to 1. A negative sign means that the voltage is inverted. Input voltage for a converter
is V in = 290 V, then output voltage V o = 290/(1–0.42) = 500 V.
3.4 VSC Inverter
Direct power (DC) and voltage are provided by PV modules. However, alternating
current (AC) and voltage (AC) are required in order to feed the electricity to the grid.
Inverter is the unit for converting DC to AC. They may also be responsible for main-
taining the PV array operating point on the MPP. It is also done with MPP monitoring
algorithms for controlling purposes. The VSC converts 500 V DC voltages to 260 V
AC and retains the unit power factor. The control mechanism of VSC consists of 2
control loops, which are (a) external & (b) internal control loops, in both cases PI
controller was implemented.
a. A loop regulates DC link voltage to ±250 V. (Read color)
b. B loop provides current components (Yellow color) by regulates grid currents,
see Fig. 5 control circuit.
Fig. 5 control circuit of VSC

4 Simulation Results
4.1 Simulink Model of PV Cell
In the present model, 60 cells are connected in series, each cell voltage is 0.605 V, a
simple PV module circuit as shown in Fig. 2. To compare the operational behavior of
PV cell module, the simulation is performed with different solar irradiations like 250,
500, 850 W/m2 and 1000 W/m2 with constant temperature 25 ◦ C, the corresponding
MPP changes from 5.342 kw, 10.8 kw, 18.22 kw and 21.32 kw, respectively, see
Fig. 6. For different irradiation conditions, the performance characteristics (current
vs voltage curve) of PV array are shown in Figs. 6a. The MPP is changes from
19.54 kw, 20.86 kw to 21.32 kw when temp changes from 45 ◦ C, 35 ◦ C and 25 ◦ C,
respectively, with constant 1000 W/m2 . As it can be seen from current vs voltage
curves, the PV cell current is highly dependent on the value of solar irradiation.
The output power is reduced when irradiation is reduces from 1000 to 250 W/m2 .
Figure 1 shows the simple circuit diagram of PV grid-connected system, through
dc–dc converter and inverter.
In this paper, standard test condition and practical test conditions are considered
for performance evaluation. In the standard test condition, solar irradiance is taken as
1000 W/m2, and temperature is taken as 25 ◦C as shown in Fig. 7a. In the practical test
condition, solar irradiance and temperature one year from September 2018 to August
2019 is logged at B V Raju Institute of Technology, Narsapur, Medak, Telangana
as shown in Fig. 7b. Month wise mean values of solar irradiance and temperature
are calculated as shown in Fig. 7c and with this data performance of proposed grid-
connected PV system is evaluated.
Array type: 1Soltech 1STH-215-P; Array type: 1Soltech 1STH-215-P;

10 series modules; 10 parallel strings 10 series modules; 10 parallel strings
100
100
Current (A)
Current (A)
1 kW/m 2
50 0.85 kW/m 2
o o
45 C 25 C 50
o 0.5 kW/m 2
30 C 0.25 kW/m 2
0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Voltage (V) Voltage (V)
4 4 X: 290
10 10
4 Y: 2.132e+04
2.5
Power (W)
Power (W)
o
25 C 2 1 kW/m 2
2 o
o
1.5 0.85 kW/m
2
C
45 C 30
2
1 0.5 kW/m
2
0 0.25 kW/m
0 50 100 150 200 250 300 350 400 100 150 200 250 300 350
Voltage (V) Voltage (V)
(a) (b)
Fig. 6 I–V and P–V curve at a different temperatures 25, 30 and 45 degrees at 1000 irradiation
b different irradiation 1000, 850, 500 and 250 W/m2 at 25◦
(a) STC condition (b) PTC Condition

Ir (W/m^2)
1000
X: 0.7577
500 Y: 1179
0 0.2 0.4 0.6 0.8 1 1.2
Time (seconds)
Temp (deg. C)
30
20 X: 0.7577
10 Y: 29.84
0
0 0.2 0.4 0.6 0.8 1 1.2
Time (seconds)
(c) Simplified Temperature and Irradiation
Fig. 7 Simulation diagram of irradiation and temperature in different condition
4.2 STC Condition
In the standard test conditions, PV voltage (V) and PV current (A) are show in Fig. 8a,
b, respectively, actual and reference DC link voltages are compared and plotted in
Fig. 8c. Since the solar irradiance and temperature are constant at all instances, PV
output also constant at all the instances. The PV output power and grid power are
compared and plotted in Fig. 9. Power difference of 2 kW is recorded between output
power and grid power and considered as converter losses. Grid current and voltage
are plotted in Fig. 10a, b, Fig. 11 represents inverter line to line voltage.
4.3 Dynamic Condition
In the dynamic test conditions, PV voltage (V) and PV current (A) are not changed,
and actual and reference DC link voltages are compared and plotted in Fig. 12. Since
the solar irradiance and temperature are variable at all instances, PV output also
varies at all the instances. The PV output power and grid power are compared and
Fig. 8 a Voltage and b current generated by PV c DC Link voltage comparing with reference
voltage
Fig. 9 Powers generated by the PV and grid power

Fig. 10 Current and voltage injected into the grid a grid voltage b current
Fig. 11 DC–AC inverter line to line voltage

Fig. 12 PV output characteristics DC link voltage
Fig. 13 Power flow at grid side and source side
plotted in Fig. 13. Power difference at 0.76 s is 2.6 kW is recorded between output
power and grid power and considered as converter losses.
The THD analysis in the both cases are analyzed, in the STC conditions, the grid
current THD is reported as 9.67%, similarly in PTC grid current THD is reported as
15.77%. Hence, significant amount of THD is varied from STC to PTC conditions
see Fig. 14. Nevertheless, control topologies need to be enhanced to overcome this.
5 MPPT Performance Key Points
5.1 Tracking
The GMPP Access: Since solar photovoltaic systems do not even obtain uniform
sunshine for a short period of time between very near areas, partial shading is also
possible due to inevitable conditions [12, 16]. These situations can lead to multiple
Fig. 14 THD analyses of 50 Hz grid current wave form a THD at STC. b THD at dynamic
conditions
LMPPs, which are based on the IV–PV feature, affecting the MPPT tracking perfor-
mance. Traditional MPPT algorithms are ineffective for monitoring and shading
GMP PCs, while intelligent MPPT for optimization are designed to track GMPP
over several LMPPs.
5.2 Convergence Speed
A strong MPPT algorithm needs to converge at the appropriate current and voltage
with good speed and accuracy, regardless of the constant or dramatic change in
solar irradiance. Reasonably, the GMPP takes more time than the hybrid solution
to adopt a traditional maximum power point. In addition, the GMPP incorporates
hybrid algorithms with limited or reduced oscillations [12, 16].
5.3 Sensitivity
An effective MPPT algorithm must be sufficiently adaptive to handle any conditions

and changes in the atmosphere. It needs to be able to respond quickly and track the
GMPP in a particular PV system [17].
5.4 Design Complexity
One of the important factors is the selection of an adaptable MPPT which is retained
due to the difficulty of its development for a specific PV device [12, 16]. How
successful the algorithm searches for the real GMPP in the life of various LMPPs
depends on how complex the MPPT technique is. In addition, the photovoltaic system
does not harvest the entire solar energy.
6 Conclusion
Due to the abundance of sunlight, SPV is the best and important source of electricity
in the renewable energy system. But it has some disadvantages, such as weather
interference, poor efficiency. For this function, a power electronics interfaced MPPT
is used to obtain maximum power from a photovoltaic system. In this paper, we have
discussed and analyzed the modeling of PV array, most recent technique, i.e., P&O
based MPPT and grid-connected system output power, I–V, P–V characteristics, PV
power, THD at STC and practical condition and required performance key points
for any MPPT are discussed. THD verified with real PV input data, it changed from
9.67% to 15.77% due to source side disbalances caused due to change in irradiation
and temperature.
References
1. Technical Report, “Growth of electricity sector in India from 1947–2012”, Central Electricity
Authority (CEA) of India
2. Doolla S Grid connection issues for distributed generation—review & standards. Department
of Energy Science and Engineering, Indian Institute of Technology Bombay
3. Subudhi B, Pradhan R (2013) A comparative study on maximum power point tracking
techniques for photovoltaic power systems. IEEE Trans Sustaine Energ 4(1)
4. Sera D et al (2013) On the perturb-and-observe and incremental conductance MPPT methods
for PV systems. IEEE J Photovoltaics 3(3):1070–1078
5. Zope PH et al (2010) Performance and simulation analysis of single-phase grid connected PV
system based on source inverter. In: International conference on power electronics, drives and
energy system
6. Kasera J et al Design of grid connected photovoltaic system employing incremental conduc-
tance MPPT algorithm. J Electr Eng
7. Sujatha P et al (2020) Implementation of anfis-mptc for 20 kwp spv power generation and
comparison with flmppt under dissimilar conditions. Int J Amb Energy 41(1):1–11. https://doi.
org/10.1080/01430750.2019.1707117
8. Ram JP Rajaseka N, Miyatake, M (2017) Design and overview of maximum power point
tracking techniques in wind and solar photovoltaic systems: a review. Renew Sustainable Energ
Rev 73:1138–1159. https://doi.org/10.1016/j.rser.2017.02.009
9. Adly M, Strunz K (2018) Irradiance-adaptive PV module integrated converter for high efciency
and power quality in standalone and DC microgrid applications. IEEE Trans Industr Electron
65(1):436–446. https://doi.org/10.1109/TIE.2017.2723860
10. Technical Report, “Monthly all india installed generation capacity report, June 2014”, Central
Electricity Authority (CEA) of India
11. https://energyeducation.ca/encyclopedia/Photovoltaic_cell.
12. Ghasemi MA, Foroushani HM, Parniani M (2016) Partial shading detection and smooth
maximum power point tracking of PV arrays under PSC. IEEE Trans Power Electron
31(9):6281–6292
13. “Photovoltaic Cell Conversion Efficiency,” U.S. Department of Energy, Retrieved, 19 May
2012
14. Kamarzamana NA, Tanb CW (2014) A comprehensive review of maximum power point
tracking algorithms for photovoltaic systems. Renew Sustain Energ Rev 37:585–598
15. Onat N (2010) Recent developments in maximum power point tracking technologies for
photovoltaic systems. Int J Photoenergy 1–10
16. Ali A et al (2020) Investigation of MPPT techniques under uniform and non-uniform solar irra-
diation condition–A Retrospection. IEEE Access 8:127368–127392. https://doi.org/10.1109/
ACCESS.2020.3007710
17. Venkata Prasant B (July 2019) Performance evaluation of MPPT in a PVG with conventional
and intelligent controllers under STC and DTC. Int J Recent Technol Eng ISSN: 2277-3878,
8(2):4984–4991. https://doi.org/10.35940/ijrte.b1075.078219
Study of the Operating Parameters
of a Two-Bed Adsorption System:
Comparison Between
the Silica–Gel–Water and Zeolite–Water
Pairs
Mouhamadou Lamine Cisse and Biram Dieng
Abstract The aim of this work is to model firstly the different components of a two-
bed adsorption cooling system with continuous cold production fed by hot water from
biogas or thermal sensors, then in a second step, to the numerical resolution in order
to determine the evolution of the various temperatures at the level of the adsorber,
the desorber, and to evaluate the COPs. For the validation of the model, the zeolite–
water couple was compared with the literature. The results of the analysis show
that the COP of zeolite–water is 0.38, lower than that of silica–gel–water which is
0.62. The evaporating temperature reached is between 0 and 5 °C and the cooling
temperature between 8.01 and 8.55 °C. This temperature range is acceptable as it
is intended for dairy products. The variation in the desorbed fraction is between
1.36 and 1.47 kg/kg s (decrease in the desorbed fraction), during adsorption, it is
between −1.1 and −1.05 kg/kg s (increase in the adsorbed fraction). The desorption
temperatures reached are 56.8 °C for silica–gel and 51.25 °C for zeolite–water.
Keywords Simulink · Zeolite · Water · Dormand–Prince · Temperature ·

Desorption
1 Introduction
The demand for energy, especially in refrigeration, is considerable [1].

Adsorption systems have received a lot of attention in recent years because of
their environmental friendliness. This study is interested in the production of cold by
adsorption with biogas and thermal sensors as energy source. It is a two-bed system.
It is used for the preservation of dairy products with a temperature range of 4–8 °C.
This type of system constitutes a promising application for solar cooling in general
M. L. Cisse (B) · B. Dieng

Research Team in Renewable Energies, Materials and Lasers, Department of Physics UFR
SATIC, University Alioune Diop of Bambey Senegal (UADB), Campus, Diourbel BP 30,
Bambey, Senegal
B. Dieng
https://doi.org/10.1007/978-981-16-4943-1_12
124 M. L. Cisse and B. Dieng
Fig. 1 Study system
and for the adsorption system in particular. The coefficient of performance will thus
be the subject of a determination, as will the other operating parameters (Fig. 1).
2 System Description
The heat is supplied by hot water produced either from the thermal sensor, more
precisely the vacuum sensor, or from biogas. The hot water enters the desorber
(E2) and transmits its heat to the adsorbent. The latter begins the desorption of the
refrigerant contained within it as soon as it reaches the desorption temperature. Then
the desorbed fluid being in the form of vapor will condense in the condenser with
rejection of heat. Finally, when it reaches saturation pressure, the fluid is passed to
the evaporator.
During this time, the adsorber (E1) cooled by fans begins an adsorption phase
of the evaporated fluid at the level of the evaporator allowing continuous cold
production. The two elements work alternately.
The system will be controlled by valves.
Study of the Operating Parameters of a Two-Bed Adsorption … 125
3 Mathematical Model
3.1 Adsorbent Bed
Many studies have focused on mathematical modeling, especially for a cylindrical

generator, which is the typical configuration, both for the arts and the basic struc-
ture of other complex generators [2, 3]. Because adsorption refrigeration is essen-
tially a phenomenon involving heat and mass transfer processes, the modeling work
is primarily concerned with adsorption performance, heat transfer in the adsor-
bent bed, and the transfer process coupled mass. Among the different adsorption
performance models, non-equilibrium adsorption models (nonlinear equations) have
become increasingly dominant [4, 5], while equilibrium adsorption models have
received less attention [6, 7].
3.2 Energy Balance
The mathematical model of the physical problem stated below is presented as follows.
Energy balance is at the level of the adsorber and desorber.
The heat transfer can be described as follows where δ = 0 if the concerned
compartment operates as a desorber and δ = 1 if the concerned compartment operates
as an adsorber.
If ϕ = 1, the adsorber is connected to the evaporator, and if ϕ = 0, the adsorber
is connected to the desorber.
dTa dw
(m ad ∗ Cad + m a ∗ Ca + m a ∗ w ∗ C pr ) ∗ = m a ∗ Hads ∗
dt dt
dw
+ δm a ∗ C pr,v ∗ ϕ(Tev − Ta ) + (1 − ϕ) Tr,v − Ta ∗
dt
+ ṁ f,ad ∗ C p f T j,in − T j,out (1)
Condenser
dTcd dwdes
m r,cd ∗ C pr + m cd ∗ Ccd ∗ = −m a ∗ ∗ Lv
dt dt
dwdes
− m a ∗ cpr,v ∗ Tgn − Tcd
dt
+ ṁ f,cd ∗ cp f (T j,in − T j,out ) (2)
Evaporator
dTev dwads
m ev ∗ m r,ev ∗ C p,r = −m a ∗ ∗ Lv
dt dt
Table 1 Parameters
Symbol Value Unit
ma 50 kg
Mf, ad 1.6 m3 /h
mr, ev 10 kg
Ca 0.924 Kj/kg K
Cpr, v 1.85 kJ/kg K
Cad 0.905 Kj/kg/K
Hads 2800 KJ
Lv 2500 KJ
UAcd 6090 W/K-
Mf, ev 2 m3 /h
Mf, cd 3,7 m3 /h
Tev, in 9 °C
Tcd, in 30 °C
Tg, in 70 °C
UAev 3420 W/K
Ccd, Cev 0.386 Kj/kg/K
mad 64.04 kg
Mcd 24.28 kg
mev 12.45 kg
dwdes
− ma ∗ ∗ Cpr ∗ (Tcd − Tev ) + ṁ f,ev ∗ T j,in − T j,out (kW) (3)
dt
Calculation of COP.
Q ev
COP = (4)
Q des
3.3 Numerical Solution
The numerical method used is that of Dormand–Prince. This method is nothing other
than that of EULER of order 45. The primary data used in the numerical calculations
are given in Table 1.
4 Simulations
This curve Fig. 2 shows that the desorption temperature varies between 49 and
56.8 °C. It reaches a maximum value equal to 56.8 °C. The adsorption temperature
has an initial value of 55 °C, then it decreases to 49.2 °C considered as the end
of adsorption value. After these two phases, preheating (from 49.2 to 50.8 °C) and
precooling are observed.
An increase in the variation dw/dt of the desorbed fraction is observed Fig. 3.
This shows that the desorbed fraction decreases over time. The minimum value of
the variation is 1.36 kg/kg s, and the maximum value of the variation is evaluated
at 1.47 kg/kg s. This evolution is justified because the desorption being progressive,
the fraction decreases over time. It stops at the end of the desorption.
A negative growth is observed in the variation dw/dt of the adsorbed fraction
Fig. 4. This shows that the adsorbed fraction increases over time. The minimum
value of the variation is −1.1 kg/kg s, and the maximum value of the variation is
evaluated at −1.05 kg/kg s.
The simulation shows a constant condensing temperature equal to 30 °C. Figure 5.
Fig. 2 Temperature evolution at the adsorber and desorber as a function of time for the silica–gel
water
Fig. 3 Evolution of the variation of the desorbed fraction over time for silica–gel water
This curve Fig. 6 shows that the desorption temperature varies between 49 and
51.25 °C. The adsorption temperature, for its part, has an initial value of 52 °C, then
it decreases to 50.9 °C considered as the end of adsorption value. After these two
phases, preheating and precooling are observed. The comparison with Fig. 7 shows
the same pattern, in particular with regard to the temperatures of SE1 (adsorption)
and SE2 (desorption). This model in Fig. 7 has already been validated [8].
The evolution of the COP shows a maximum value of 0.38 at the end of the cycle.
This is compared with Fig. 9. We see an increasing evolution like that observed for
a two-bed system using the same couple with a COP of 0.56 Fig. 9 [8].
An evaporation temperature is observed which varies between 0 and 5 °C Fig. 10.
It is equal to 5 °C over 6.5 h then reaches 0 °C for the rest of the cycle. This value
obtained is relatively low and favors a cooling temperature below 10 °C.
It can be seen that the cooling temperature Fig. 11 varies between 8.01 and 8.55 °C.
This relatively low value is suitable for cooling dairy products.
The curve shows an increase in COP over time Fig. 12. It reaches a maximum
value of 0.38 for zeolite–water and 0.62 for silica–gel–water. The latter is close to 0.6
in [9]. This value is low in the case of adsorption systems, compared to absorption
systems.
Fig. 4 Evolution of the variation of the adsorbed fraction over time for silica–gel water
5 Conclusion
This modeling and simulation work made it possible to determine the evolution of
desorption, condensation, and evaporation temperatures in the case of an adsorption
system with two adsorption beds. This after inserting the input parameters of our
model, these results were subsequently compared with that of other studies to measure
validity. A comparison between the couples was made.
Fig. 5 Evolution of the condenser temperature for silica-gel-water

Fig. 6 Evolution of the adsorber and desorber temperatures for zeolite–water
Fig. 7 Profiles of the different temperatures of 13X/cacl2–water–zeolite [8]

Fig. 8 Variation COP for zeolite–water over time
Fig. 9 Evolution of the COP of the Zeolite 13X/ CaCl2 –water pair [8]
Fig. 10 Variation of the evaporator temperature for silica–gel over time

Fig.11 Variation of the cooling temperature for silica–gel water over time
Fig. 12 Variation of silica–gel–water and zeolite–water COP as a function of time
References
1. Clito FA (1961) Afonso, recent advances in building air conditioning systems. Appl Therm Eng
26(2006):1971
2. Zhao HZ, Zhang M, Liu ZY, Liu YL, Ma XD (2008) Mechanical and experimental study on
freeze proof solar powered adsorption cooling tube using active carbon/ methanol working pair.
Energy Convers Manage 49:2434–2438
3. Mers AA, Azzabakh A, Mimet A, Kalkha HE (2006) Optimal design study of cylindrical finned
reactor for solar adsorption cooling machine working with activated carbon-ammonia pair. Appl
Therm Eng 26:1866–1875
4. Leong KC, Liu Y (2004) Numerical modeling of combined heat and mass transfer in the
adsorbent bed of a zeolite/water cooling system. Appl Therm Eng 24:2359–2374
5. Yang PZ (2009) Heat and mass transfer in adsorbent bed with consideration of non-equilibrium
adsorption. Appl Therm Eng 29:3198–3203
6. Zhang XJ, Liu HX, Wang RZ, Shi F (2002) Numerical simulation of heat transfer in regenerator
of solid adsorption refrigeration system. Renew Energy 26:599–610
7. Hu J, Exell RHB (1994) Simulation and sensitivity analysis of an intermittent solar powered
charcoal/methanol refrigerator. Renew Energy 4:133–149
8. Alizera et al (2014) Performance evaluation of Zeolite 13X/CaCl2 two-bed adsorption
Refrigeration system. Int J Thermal Sci 80:76–82
9. Alam KCA et al (2013) Solar adsorption cooling: a case study on the climatic condition of
dhaka. J. Comput. 8(5)
Investigation of Hybrid Fault Ride
Through Control Strategies
in Grid-Connected Wind Energy
Conversion Systems
J. Preetha Roselyn, C. R. Raghavendran, and D. Devaraj
Abstract The significant issues which are associated with grid code during grid
fault are the fault ride through (FRT) capability of wind farms. This paper presents a
detailed investigation of different hybrid fault ride through schemes in DFIG-based
wind system under both symmetrical and unsymmetrical fault conditions. This work
analyzed the performance of hybrid FRT combinations like crowbar resistance with
RL circuit, DC chopper associated with crowbar resistance, DC chopper connected
with series dynamic resistor, voltage control scheme with crowbar resistance are
developed and studied under symmetrical and unsymmetrical fault conditions. The
simulation results are obtained using MATLAB Simulink version 2019b, and the
hardware real-time simulation is implemented in OPAL-RT real-time simulator.
The proposed hybrid FRT schemes provide promising results improving the system
parameters under different fault scenarios in grid-connected wind system.
Keywords Doubly fed induction generator · Fault ride through · Grid code
compliance · Grid faults · DC link voltage control
1 Introduction
Renewable energies play a vital role in modern power grid of which wind energy has
the fastest deployment in many countries. Various operational guidelines have been
developed for the integration of distributed generation by several countries according
to different grid codes [1–3]. Doubly fed induction generators (DFIGs) are the most
widely used wind energy conversion system due to its special features like variable
speed and real-reactive power control [3–6]. The control of DC link voltage and real
J. Preetha Roselyn (B) · C. R. Raghavendran

Department of Electrical and Electronics, SRM Institute of Science and Technology,
Kattankulathur, Chennai, India
D. Devaraj
Department of Electrical and Electronics, Kalasalingam Academy of Research and Education,
Srivilliputhur, India
https://doi.org/10.1007/978-981-16-4943-1_13
138 J. Preetha Roselyn et al.
power transfer is carried out using the grid side power converter, whereas the rotor-
side converter is used for reactive power control in the grid during normal operating
conditions [7–9]. During fault conditions, the DFIG-based wind system is usually
disconnected from the grid to protect the rotor-side power converter and DC link
capacitor [10, 11]. But according to grid code requirements, the distributed genera-
tion should be connected to the grid during fault conditions till a certain extent and
hence support the grid. The suitable fault ride through (FRT) schemes are essential
for DGs to support the DG-connected grid system by regulating the real and reactive
power and also to protect the power electronic components in the system. Many
conventional FRT methods, namely crowbar protection circuit, dynamic voltage
restorer (DVR), superconducting fault current limiters (SFCL), series damping resis-
tors (SDRs) are mentioned in literature. In this paper, the different conventional FRT
schemes such as crowbar resistance, DC chopper, series dynamic resistor and voltage
control scheme and hybrid schemes such as crowbar resistance with RL, crowbar
resistance with series dynamic resistor and series dynamic resistor with DC chopper
and voltage control with crowbar protection scheme are exhaustively investigated
under symmetrical fault conditions for all grid parameters like voltage at the point of
common coupling, DC link voltage, stator current, rotor current, post fault transient
current, real and reactive power injection. The paper is organized as follows: The
paper is organized as follows: FRT grid codes are detailed in Sect. 2. The details of
the hybrid FRT methodologies are explained in Sect. 3. The results and discussions
of simulation and hardware are detailed in Sect. 4. The conclusion and future scope
are represented in Sect. 5.
2 FRT Grid Requirement
The grid code requirement during fault conditions is provided in Fig. 1. The curve
represents that the wind turbine should stay connected to the grid during the abnormal
condition and supports grid stability. Table 1 represents the fault clearing time, the
nominal system voltage, and the duration of the fault clearing. The V f specifies that
15% of nominal system voltage should be maintained during the fault condition, and
the V pf signifies the minimum voltage under normal operation of the wind system.
Fig. 1 Grid code FRT curve

in various constrains
Investigation of Hybrid Fault Ride Through … 139
Table 1 Indian grid code

Nominal system voltage Clearing time V pf (kV) V f (kV)
requirement
(kV) (ms)
400 100 360 60
220 160 200 33
132 160 120 19.8
110 160 96.25 16.5
66 300 60 9.9
3 Framework of Hybrid FRT Schemes
3.1 Proposed Hybrid FRT Strategies
To get the best characteristics of each FRT scheme, the conventional FRT strategies
are combined in a proper way for better FRT capability of the system. The proposed
hybrid FRT schemes developed and implemented in this work are crowbar resistance
with RL circuit, DC chopper with crowbar resistance, DC chopper with the series
dynamic resistor and voltage control scheme with crowbar resistance.
3.1.1 Hybrid Crowbar Resistance with RL Circuit
The hybrid crowbar protection circuit is designed with thyristor bypass resistors as
shown in Fig. 2. This combination helps in diminishing the rotor current and helps in
suppressing the DC link voltage. When the value of the terminal voltage decreases
below the threshold value of the comparator, the FRT scheme will be activated and
the rotor controller is turned off. When the terminal voltage exceeds the threshold
value, the crowbar will be deactivated and the rotor controller will be activated.
3.1.2 Hybrid DC Chopper with Crowbar Resistance
Crowbar resistance is incorporated with DC chopper and is shown in Fig. 3. This

hybrid controller protects the DC link capacitor from overvoltage during grid fault
conditions. The combination of dc chopper and crowbar resistance helps in reducing
DC link voltage to the permissible value, and the crowbar resistance helps in main-
taining the rotor currents. This combination thus helps in controlling DC link voltage,
rotor currents and active power.
Fig. 2 Hybrid crowbar resistance with RL circuit FRT strategy
Fig. 3 Hybrid DC chopper with RL circuit FRT strategy
3.1.3 Hybrid Voltage Control Scheme and Crowbar Resistance
The combination of the voltage control scheme with crowbar resistance as shown
in Fig. 4 helps in reducing the oscillation of the real power and rotor current. The
crowbar resistance is adequate in damping the oscillations in the rotor current, and
the voltage control scheme helps in regulating the PCC voltage during grid faults.
3.1.4 Hybrid DC Chopper with SDR
In this combination, when the DC link voltage exceeds the limit, the chopper will be
turned on. Additionally, with this hybrid DC chopper as shown in Fig. 5, SDR can
restrict the rotor overvoltage and also limit the high rotor current. The rotor current
can decrease the charging current to the DC link capacitor and hence reduces the DC
link overvoltage.
Fig. 4 Hybrid voltage control scheme with Crowbar Resistance FRT strategy
Fig. 5 Hybrid DC chopper with SDR FRT strategy
The performance of the hybrid FRT strategies is tested in a grid-connected 9 MW,

DFIG-based wind system in MATLAB 2019b version. The proposed FRT strategy
is validated in OPAL-RT HIL platform. The DFIG is operated at an average wind
speed of 12 m/s. The transient behaviors of the DFIG-based grid-connected system
are studied under different fault conditions such as symmetrical and unsymmetrical
faults, and the results under symmetrical faults are provided in this section.
4.1 Simulation Results
The hybrid FRT strategies, namely (a) combined crowbar resistance with RL circuit,
(b) DC chopper with crowbar resistance, (c) DC chopper interlinked with SDR and
(d) voltage control scheme with crowbar resistance are developed and implemented
in test system as shown in Fig. 6. The different parameters are DC link voltage, real
and reactive power, stator and rotor currents, the voltage at PCC is measured and
analyzed under symmetrical fault condition.
4.1.1 Performance of Hybrid FRT Schemes Under Symmetrical Faults
Under this scenario, a three-phase short-circuit fault is simulated which produces low-
voltage fault behaviors in DFIG-WECS. The system parameters under symmetrical
fault are discussed as follows:
(i) DC link Voltage
Figure 7 shows the DC link voltage which rises instantaneously because of the
rotor inrush currents injected from the RSC into the DC link capacitor. The hybrid
FRT controllers are more effective in suppressing the DC link voltage than conven-
tional FRT scheme. Various hybrid schemes are introduced to recognize which
hybrid FRT scheme is more effective in reducing the DC link voltage. The DC link
is suppressed to 1155 V by using DC chopper with series dynamic resistor, which
is very less when compared to other combinations. DC chopper with crowbar
resistance is the next effective method in maintaining the DC link voltage within
the permissible limit.
(ii) Voltage at Point of Common Coupling
Figure 8 shows illustrates the dip in voltage, at the point of common coupling
during the short-circuit fault. By introducing the hybrid FRT scheme, the dip in
the voltage is reduced better than conventional FRT schemes and the DC chopper
Fig. 6 Single line diagram of 9 MW grid-connected wind system

Fig. 7 DC link voltage of the DFIG-WECS under symmetrical fault: a crowbar resistance with
RL, b DC chopper with CBR, c DC chopper with SDR, d voltage control scheme with CBR
Fig. 8 Voltage at PCC of the DFIG-WECS under symmetrical fault: a crowbar resistance with RL,
b DC chopper with CBR, c DC chopper with SDR, d voltage control scheme with CBR
Fig. 9 Real power of the DFIG-WECS under symmetrical fault: a crowbar resistance with RL,
b DC chopper with CBR, c DC chopper with SDR, d voltage control scheme with CBR
with SDR scheme helps in reducing the voltage dip to 0.2 p.u with a recovery
time of 3.17 s and hence the transient behavior of the system.
(iii) Real Power
Figure 9 illustrates the real power of the system when the system is subjected to
symmetrical fault condition. When the voltage control scheme is introduced in the
system, the oscillation is reduced and reaches 0.8 p.u during the fault clearing time.
When the hybrid voltage control scheme with crowbar resistance is introduced in
the system, the real power reaches to 0.5p.u and oscillation is also less.
(iv) Reactive Power
Figure 10 illustrates the reactive power injected into the system when it is subjected
to symmetrical fault condition. When the DC chopper FRT scheme is introduced
to the system, the reactive power injection levels reach the value of 0.1p.u, and
when hybrid FRT scheme like DC chopper combined with SDR is introduced under
the same condition, it can be seen that the reactive power injection is increased to
0.5p.u. This hybrid combination is verified as the potential hybrid FRT scheme when
compared to the other FRT schemes in case of reactive power support.
The performance comparison of different hybrid FRT schemes developed for the
test system is shown under symmetrical fault conditions. From the figures, it is clear
that the voltage control scheme with crowbar resistance and DC chopper with SDR
schemes is good in reducing the post fault transient current.
Fig. 10 Reactive power of the DFIG-WECS under symmetrical fault: a crowbar resistance with
RL, b DC chopper with CBR, c DC chopper with SDR, d voltage control scheme with CBR
4.2 Real-Time Simulation Results
The hardware simulator is composed of OPAL-RT interconnected with the PC in

which the wind system of maximum power of 9 MW is generated at the wind speed
of 12 m/s. The power coefficient C p is maintained to be 0.323. The performances
of the proposed configuration under real-time environment are evaluated under the
fault conditions and are discussed in following subsections. In the proposed hybrid
controller, DC chopper with SDR which performs better in FRT capability in the
simulated environment is tested in the hardware setup. A hardware simulator was
designed and built on OPAL-RT RCP/HIL[Version-OP4510] interconnected with
the RT lab simulator software installed PC, and the waveforms are observed using
the oscilloscope shown in Figs. 11 and 12. The performance of the proposed FRT
controller under real-time environment is evaluated under the unsymmetrical LLG
fault condition. In OPAL-RT platform, the SimPower Systems model is used with
subsystem blocks, wind system as the Master Subsystem (SM) and the FRT controller
in the Slave Subsystem (SS).
Under this scenario, unsymmetrical LLG fault is simulated which produces
low-voltage fault behaviors in DFIG-WECS. It has been analyzed that the hybrid
controllers are more effective in suppressing the DC link voltage (Fig. 13) than
conventional schemes. DC chopper with SDR hybrid FRT scheme is more effective
in suppressing the DC link voltage. The DC link is suppressed to 1150 V by using DC
chopper combined with series dynamic resistor, and hence selected as the suitable
hybrid FRT scheme in suppressing the DC link voltage. Figure 14 shows the real
Fig. 11 Hardware setup for FRT controller
Fig. 12 Hardware configuration
power absorbed by the system when it is subjected to unsymmetrical fault condition.

When the DC chopper with SDR is introduced in the system at that condition, the real
power reaches to 0.5 p.u and oscillation is less when compared to the conventional
FRT scheme. Figure 15 illustrates the voltage dip in the point of common coupling
during the unsymmetrical fault condition. The rotor current increases which lead to
heavy dip in the PCC point. By selecting DC chopper with SDR scheme, the voltage
level is reduced only to 0.7 p.u and recovery with 3.16 s. This hybrid scheme is the
best one when compared to other FRT schemes by reducing the voltage sag at the
PCC point (Table 2).
Fig. 13 DC link voltage
Fig. 14 Real power
Fig. 15 Voltage at PCC

Table 2 Experimental results of hybrid FRT controllers under unsymmetrical fault condition
Voltage at PCC DC Link Voltage Real Power
Dips to 0.7 pu and recovered at Peak value of 1150 V Oscillations reached 0.5pu during
3.16 s fault clearing
5 Conclusion
The hybrid FRT schemes are studied under different system parameters such as oscil-
lations in the active and reactive power, rotor and stator currents, DC link voltage,
etc. which are successively improved. In addition, different combinations of hybrid
FRT schemes are designed and analyzed on different system parameters which will
improve the consistency of both the wind farm and the power system. From the
comprehensive analysis on different hybrid FRT schemes, it is understood that DC
chopper with SDR schemes perform well in improving the system conditions and
supporting the grid under symmetrical faults. In future, new control schemes incor-
porated with intelligent control techniques like artificial neural network and fuzzy
logic could be implemented for the wind turbines to enhance the stability of the
system.
References
1. Zeng Z, Yang H, Zhao R, Cheng C (2013) Topologies and control strategies of multi- func-
tional grid-connected inverters for power quality enhancement: a comprehensive review. Renew
Sustain Energy Rev 24(5):223–270
2. Lund H, Munster E (2006) Integrated energy systems and local energy markets. Energy Policy
34(10):1152–1160
3. Tazil M, Kumar V, Bansal RC, Kong S, Dong ZY, Freitas W, Mathur HD (2010) Three-phase
doubly fed induction generators: an overview. IET Electr Power Appl 4(2):75–89
4. Asuhaini AB, Zin BM, Pesaran MHA, Khairuddin AB (2013) An overview on doubly fed
induction generators control and contributions to wind based electricity generation. Renew
Sustain Energy Rev 27:692–708
5. Mwasilu F, Justo JJ, Ro KS, Jung JW (2012) Improvement of dynamic performance of Doubly
Fed Induction Generator-based wind turbine power system under an unbalanced grid voltage
condition. IET Renew Power Gener 6(6):424–434
6. Eriksen PB, Ackermann T, Abildgaard H, Smith P, Winter W, Rodriguez G (2005) System
operation with high wind penetration. IEEE Power Energ Mag 3(6):65–74
7. Pannel G, Atkinson DJ, Zahawi B (2010) Minimum-threshold crowbar for a fault-ride- through
grid-code-compliant DFIG wind turbine. IEEE Trans Energy Conserv 25(3):750–759
8. Noureldeem O (2012) Behavior of DFIG wind turbines with crowbar protection under short
circuit. Int J Electr Comput Sci 12(3):32–37
9. Jadhav HT, Roy R (2013) A comprehensive review on the grid interaction of doubly fed
induction generator. Int J Electr Power Energy Syst 49:8–18
10. Morren J, De Haan SW (2005) ride through of wind turbines with doubly-fed induction
generator during a voltage dip. IEEE Trans Energy Convers 20(2):435–441
11. Lopez J, Gubia E, Olea E, Ruiz J, Morroyo L (2009) Ride through of wind turbines with
doubly fed induction generator under symmetrical voltage dips. IEEE Trans Industr Electron
56(10):4246–4254
Performance Analysis of Single-Axis
Solar Tracker Using IoT Technique
S. Usha, A. Geetha, T. M. Thamizh Thentral, C. Subramani, R. Ramya,

and C. S. Boopathi
Abstract To empower the abstraction of maximum power from the solar panel using
MPPT with P&O algorithm, this proposed work enlightens the practice of single-axis
solar tracker exploitation by Arduino Uno. The introduction of IoT into the project
helps in remote monitoring of the entire system and also the analysis with the neat
graphical representation of the data over a course of time. The significant charac-
teristic for monitoring, supervising and performances estimations of integration of
Internet of Things (IoT) technology which is an efficient wireless communication
technology is considered which disables the difficulties faced in physical monitoring
of solar photovoltaic (PV) and getting the values of voltage and current.
Keywords Arduino · Node MCU · Illumination · Power generation · Internet of

Things · Wireless communication
1 Introduction
In this contemporary world of globalization, the growth of mankind leads to

increasing demand of energy consumption which affects the conventional sources
of electricity [1, 2]. Keeping in mind, man has discovered several techniques of
harnessing electrical energy from renewable resources. The renewable resources are
naturally replenished. The most suitable of them is solar energy which is free and
provides us with light and heat, being pollution-free. It is an eco-friendly supply
[3–5].
Despite of solar being a promising source of energy, there is a need of improving
method to harness the energy. With being variable weather condition, a limitation
affecting solar module, different maximum power point tracking methods are used
[6]. The easiest to implement method is perturb and observe (P&O) which yields
highly efficient output. MPPT is used in the extraction of the maximum. DC-DC
converter act as an interference between the PV module and the load. By varying the
S. Usha (B) · A. Geetha · T. M. T. Thentral · C. Subramani · R. Ramya · C. S. Boopathi

Tamil Nadu, Kattankulathur 603203, India
https://doi.org/10.1007/978-981-16-4943-1_14
152 S. Usha et al.
duty cycle, the load impedance is varied to match the peak point of power with the
source [7–10].
Instead of a fix system, a single-axis solar tracker has been implemented in the
project using Arduino, light detecting resistors (LDR) and a motor. Solar panel is
assembled with connecting a motor to track the maximum sunlight which will shine
on the panel directly at any time of the year and day [11, 12].
System characteristic is being monitored and optimized using perturb and
observed technique and by the use of boost converter. The source impedance is
adjusted to match the load impedance by varying the duty cycle. The purpose of this
project is to also develop a solar tracker based on the intensity of light, to increase
the output voltage, current and power [13–16].
2 Description of Block Diagram
Figure 1 tells about the basic of the entire project. It consists of Arduino for all
the calculation of the voltages, for integrating with the IoT and also helps with the
automatic solar tracking. The energy received from the solar panel goes directly
into the DC-DC boost converter in which the duty cycle changes according to the
varying input voltage. It is done using the MPPT algorithm based on the perturb &
observe (P&O) method. It is also called hill climb method which adjusts the duty
cycle and then by comparing, it moves towards maximum power point. The system
also consists of IoT platform which uses things peak platform for connecting the
Fig. 1 System architecture

Performance Analysis of Single-Axis Solar Tracker … 153
Fig. 2 Boost converter
system to internet. All the readings for the voltage and current are uploaded onto the
platform which then can be viewed from anywhere just by accessing the platform.
The tracking is facilitated by using two LDRs in combination with motor which asses
the current light situation in the area and accordingly control the panel orientation.
By using the solar tracking, the overall power generation can be increased by almost
10% to 25%, hence, lowering per unit cost.
3 DC-DC Boost Converter
A boost converter, whose output voltage is higher than input voltage. It steps up the
voltage and consists of two semiconductors with one energy storing element like
capacitor or inductor or the combination of the mentioned elements (Figs. 2 and 3).
4 P&O MPPT Algorithm Analysis and Implementation
In the P&O algorithm, small increments are added until the maximum power point
is reached, and this is how PV array is perturb; initially, the power is calculated by
values of V1 and I1 simultaneously calculating V2 and I2 for P2. If P2 is positive,
154 S. Usha et al.
Fig. 3 DC-DC boost

converter
Fig. 4 MPPT tracking curve
then it goes in the same direction; once P2 reads negative, the output power is flowing
back towards the maximum power point by adding a negative increment. At the time
of maximum power point, the system oscillates between point. The controller tracks
operating point and brings the voltage of the solar module to operate around MPP.
DC module output is located with DC-DC converter. The circuit model also sends
real-time data to a free IoT platform for analytical approach (Figs. 4 and 5).
5 Analysis and Simulation
The simulation for the solar and MPPT module is done using MATLAB Simulink
2013. The PV cell, i.e. independent source supplies the proposed circuit; the input
is given by the output, and PV terminals provide duty cycle for the boost converter.
The PV panel used is 12v, and the output voltage is boosted up to 27 V. The output
voltage by varying the duty cycle k whose value is 0.67 which is calculated by using
Fig. 5 P&O algorithm
the formula Vo = Vs./1-k the pulse width of the pulse generator is varied, and various
voltage ranges at the output can be obtained.
For entire duration of the simulation, the cell temperature for photovoltaic cell is
maintained at constant 250 C with a mean of 4000 lumens (Fig. 6 and 7).
Once the power from the solar panel is injected into boost converter, the dsPIC
controller starts functioning. The dsPIC works by varying the duty cycle based on
the input voltage sensed by the MPPT, therefore, by which the output voltage gets
boosted to maximum power point (Figs. 8, 9 and 10; Table 1).
6 Solar Tracker
We have used LDRs which help in controlling the rotation of the servo motor based
on the light intensity of the area. The circuit consists of two LDRs, servomotor,
motor driver, Arduino UNO and PV panel; the flowchart for the mechanism is given
(Fig. 11).
As seen in the diagrams, the solar panel will get orientated towards the sunlight
as we know the sun rises in the east and sets in west, so we can easily place the
156 S. Usha et al.
Fig. 6 Simulink circuit
Fig. 7 Simulink MPPT circuit
prototype to track the sunlight all day out producing maximum outputs (Figs. 12, 13,
14 and 15).
7 Conclusion
This project deals with the entire process of solar generation in which the production
can be increased by using algorithms. The modified IoT into the project helps in
remote monitoring of the entire system, and also the analysis with the neat graphical
representation of the data over a course of time was analysed. The P&O algorithm
Fig. 8 PV panel voltage
Fig. 9 Output voltage
varies the duty cycle of the boost converter based on the voltage and current calcula-
tion and keeps the voltage production around the maximum point so as to maximize
the power generation. The tracking system used also helps by increasing the overall
power generation of the system and helps in reducing the overall unit generation cost
in a large-scale project.
158 S. Usha et al.
Fig. 10 Output current
Table 1 Output parameters

MPPT Output voltage Output current Output power
method
Perturb & 23.78 V 9.7A 232.37 W
observe
Fig. 11 Flowchart of
horizontal-axis solar tracker
160 S. Usha et al.
Fig. 12 Side view (panel horizontal)
Fig. 13 Panel titled 45 degrees in NW direction

Fig. 14 Panel titled 45 degrees in NE direction

162 S. Usha et al.
Fig. 15 Hardware setup of proposed method
References
1. Suruhanjaya Tenaga (Energy Commission) Malaysia Energy Statistic Handbook, (2015)

2. Y.-H. Chang and C.-Y. Chang, “A Maximum Power Point Tracking of PV System by
Scaling Fuzzy Control,” presented at International Multiconference of Engineers and Computer
Scientists, Hong Kong, (2010)
3. S. Chin, J. Gadson, and K. Nordstrom, “Maximum Power Point Tracker,” Tufts University
Department of Electrical Engineering and Computer Science, (2003), pp. 1–66
4. Faranda R, Leva S (2008) Energy Comparison of MPPT techniques for PV Systems. WSES
Transaction on Power Systems 3:446–455
5. T. P. Nguyen, “Solar Panel Maximum Power Point Tracker,” in Department of Computer
Science & Electrical Engineering: University of Queensland, (2001), pp. 64
6. Ayushi Nitin Ingole., “Arduino based Solar Tracking System”,. International Conference on
Science and Technology for Sustainable Development, Kuala Lumpur, Malaysia, (2016) May
24 – 26
7. Gagari Deb, Arijit Bardhan Roy. Use of Solar Tracking System for Extracting Solar Energy.
International Journal of Computer and Electrical Engineering, vol. 4, no. 1, (2009), 42–46
8. Esram T, Chapman PL et al (2007) Comparison of photovoltaic array maximum power point
tracking techniques. IEEE Transactions on Energy Conversion EC 22(2):439–445
9. D. Hohm and M. Ropp, “Comparative study of maximum power point tracking algorithms
using an experimental, programmable, maximum power point tracking test bed,” in Proc. 28th
IEEE Photovoltaic specialists conference. IEEE, (2000), pp. 1699–1702
10. Femia N, Granozio D, Petrone G, Spagnuolo G, Vitelli M (2006) Optimized one-cycle control
in photovoltaic grid connected applications. IEEE Trans Aerosp Electron Syst 42(3):954–972
11. A. N. A. Ali, M. H. Saied, M. Mostafa, and T. Abdel-Moneim, “A survey of maximum ppt
techniques of pv systems,” in IEEE Energy tech, (2012)
12. C. Hua and C. Shen, “Comparative study of peak power tracking techniques for solar storage
system,” in Proc. APEC’98, vol. 2. IEEE, (1998), pp. 679–685
13. K. Irisawa, T. Saito, I. Takano, and Y. Sawada, “Maximum power point tracking control of
photovoltaic generation system under non-uniform insolation by means of monitoring cells,”
in Proc. IEEE 28th Photovolt. Spec. Conf., (2000), pp. 1707–1710
14. S. Yuvarajan and S. Xu, “Photovoltaic power converter with a simple maximum power point
tracker,” in Proc. Int. Symp. Circuits Syst., vol. 3, (2003), pp. III-399–III-402
15. Rai AK, Kaushika ND, Singh B, Agarwal N (2011) Simulation model of ANN based maximum
power point tracking controller for solar PV system. Sol Energy Mater Sol Cells 95(2):773–778
16. S. K. Kollimalla and M.K. Mishra, “Novel adaptive P&O MPPT algorithm for photovoltaic
system considering sudden changes in weather condition” in Proc. Int. Conf. Clean Elect.
Power, (2013), pp. 653–658
Wireless Power Transfer for Electric
Vehicles with Better Ground Clearance
A. Geetha, S. Usha, T. M. Thamizh Thentral, C. Subramani,

J. Santha Kumar, and C. S. Boopathi
Abstract In modern era, in electric vehicles charger regularization, the following

phase to make the charging procedure more convenient is to eradicate the usage of
wired cable sandwiched by linking the electric vehicles and charger to accomplish
wireless charging of electric vehicles, and a wireless power transfer (WPT) system
ought to be depicted with respective ground clearance of electric vehicle. It is an
innovation technology which can be applied for all electric vehicles (EVs) as it helps
to get rid of user involvement. The crucial impediment for acquiring wireless charging
is ground clearance which downgrades the power transfer efficiency. The theory of
WPT for different ground clearance is elucidated, and the corresponding cordless
charger device is analysed. The portrayed cordless charger device has capacity to
distribute the power of about 45v utmost ground clearance of 20 cm. The battery
designed is 4.5KWh and the super capacitor (SC) of 3.8KWh which is sufficient to
charge electric vehicle.
Keywords Electric vehicle · Wireless power transfer · Ground clearance · Power

converter · Vehicle charging
1 Introduction
The trade market of electric vehicles (EVs) has gradually begun to grow. To revive
the vehicle within a short span, high-power charging devices or charging stations are
required by the existing conductive charging method. Additional inconvenience is
A. Geetha · S. Usha (B) · T. M. Thamizh Thentral · C. Subramani · C. S. Boopathi

Kattankulathur 603203, Tamil Nadu, India
C. S. Boopathi
J. Santha Kumar
Department of Mechanical Engineering, SRM Institute of Science and Technology,
https://doi.org/10.1007/978-981-16-4943-1_15
166 A. Geetha et al.
caused by incompatible plugs receptacles between different EV models. Charging

infrastructure can be shared by different EV models as for the wireless charging
technologies [1, 2].
The deployment of 85 million autonomous vehicles is expected by 2035; hence,
establishment of wireless charging is needed, to incorporate in such vehicles [2].
WPT makes the system simplified, automatic, secure, economical and more efficient,
whereas using of cables makes system so expensive. For instance, at the time of
placing EVs for little while in parking lot, wireless charger systems enable automated
charging. It eradicates the need for pilot interruption in course of charging. Main
hindrance for acquiring wireless charging are coil misalignment and large air gap.
To improve misalignment tolerance, various coil structures are used. The radial coil
is pre-owned based on its unsophisticated erection [3–6]. Using miscellaneous coils,
like double D-type [8, 9] and solenoid type, [10] leads to increased air gap between
transmitter and receiver coils.
The reduction of coil misalignment before WPT’s commencement improves
transfer efficiency. In [11], the wireless charger was demonstrated by using conveyors
and servo motors which helps in positioning the receiver to the transmitter. Air gap
between the charging pad and the EV underbody is another limitation in WPT. In
[12], a wireless charger was developed for ground clearance of 20 cm, which removes
transporting data through amplitude modulation of the coil current.
As a result of numerous stances of vehicles and ground clearance, magnetic
coupling is varied. This succumbs the power efficiency. To specify this significance,
using of high reactive power and resonant elements is strongly recommended in both
transmitter and receiver side. By using power conversion technique, the contactless
coupler and the power electronics system, the battery is charged [13].
This system was designed with and supercapacitor and a lithium-ion battery. This
cordless charger comprises of rectifier and LC filter. This algorithm used here is
constant voltage and constant current, which compensates the transient state and to
achieve more stability. This wireless charger with lithium-ion battery of 4.5 KWh
and supercapacitor of 3.8KWh was analysed with efficiency of 86%.
2 Proposed Work
In recent trend, wireless power transfer is initializing to seek more and more attention.
The block diagram comprises of grid, rectifier, inverter, LC filter, linear transformer,
supercapacitor and battery as shown in Fig. 1. The low-frequency AC supply is
supplied by grid and is rectified and inverted to high-frequency AC supply. In order
to evict the harmonics, LC filter is brought into use. The high-frequency AC supply
is provided to transmitter coil, and an emf is induced.
Due to the principle of Faraday’s law of electromagnetic induction, the induced
emf energizes the receiver coil. On account of magnetic field as shown in Fig. 2, the
variations in the coupling factor k which are calculated by
Wireless Power Transfer for Electric Vehicles … 167
+ + AC
Battery _ Receiver
_V
C DC
Energy
Transfer
Grid C
AC DC Vi
440v, Transmitter
DC AC L
50 HZ
Fig. 1 Schematic layout of the proposed system
Fig. 2 Two magnetically

coupled coils with good
coupling
I1 Ø I2
V1 V2
M
K = (1)
L1 L2
It is led by variations in the ground clearance, where the mutual inductance is

M, and the transmitter and receiver coil inductances are L1 and L2, respectively. The
received AC supply is rectified and supplied to the battery through the supercapacitor.
In order to preserve the battery from catastrophic damage eruption by abrupt vehicle
moving while charging, supercapacitor is applied.
Fig. 3 Simulink diagram of the proposed system
3 Circuit Analysis
In the context of wireless transmission system, efficiency of whole system is depen-

dent on ground clearance of electric vehicles. Since in case of wireless transmission,
the transmitter is present in the atmosphere, which terrifically reduces the coupling
factor. Now here, it is earnestly indispensable to examine the coupling coefficient (k)
by adjusting the number of turns and distances between the transmitter and receiver
coil. Figure 3 depicts the simulation diagram of the proposed system. It comprises
of rectifier, inverter, LC filter, supercapacior and load. The proposed system tries to
optimize the ground clearance during wireless charging of the electric vehicle.
The vital part in the determination of coupling coefficient is ground clearance, but
for realistic motive, the number of turns will not be changing automatically as is the
case of other parameters. So, in this content, we have concentrated on the dependence
of coupling coefficient on the number of turns and distance between the coils. In the
event of change in the coupling coefficient, efficiency of the system appropriately
varies. In the designing of system parameters, coupling coefficient is useful.
The reference voltage (constant value) and gate pulse were given by PWM technique.
The PWM’s advantage is low power loss in the switching devices.
There is an absence of current when the switch is in zeroth condition, and the
power is transferred to the load when the switch is in the first condition, and also
there is zero voltage drop between the switches. The input and output currents and
the pulses offered to the power switches are as shown in Fig. 4.
Fig. 4 PWM pulses, input current and output current
Table 1 provides the details of the components used and the corresponding values
used for the system. A detailed information about the transmitter and receiver as well
as the sub-system is provided with their scope of usage (Fig. 5).
Table 1 Details of the components used

Stage Component Symbol Value Unit Scope
Transmitter MOSFET FET1-FET4 1 m Inverter
Capacitor Cr 160 nF LC filter
Inductor Lr 37.5 µH
Regulator
Primary winding 1 52 V Transmitter
0.01
75 µH
Receiver Receiver winding 2 6 V Receiver
0.01
1.36 µH
Diode D1-D4 1 m Rectifier
Capacitor C0 100 µH Supercapacitor
Resistor R0 3 Load
Sub-system Diode D5-D8 1 m Rectifier
Inductor L1-L4 152 µH Filter
Diode D9-D12 1 m Rectifier
Capacitor Cf 47 µH RC filter
Resistor Rf 150
Fig. 5 Input and output voltages of the proposed system
An input supply voltage of 440 V is provided to the proposed system as shown in

figure. The power converter designed will offer an output voltage of 45 V is obtained.
Hence, a suitable voltage for the wireless charging of the vehicle is obtained with
good efficiency.
5 Conclusion
In this material, an authentic method is portrayed to handle different ground clearance

in wireless power transfer systems. The measurements used for the above parame-
ters show reliable results for proposed theme. In the prototype, zero visible heating
of the used relays was notified. Several simulations were performed. Regardless of
previous studies, we inspected with some real-world charger placement limitations
in the surveillance area such as walls and ground clearance. The above delivered
system is economical and highly beneficial because it uses inverters, rectifiers and
filters to handle a different coupling factor. Separate control system is enrolled, where
the transmitter coil is automatically tapped for transferring the power to the battery
based on the ground clearance within 200 mm. Eventually, this paper contributes to
the typical hindrance of varying ground clearance. The wireless charger was proce-
durally verified for the battery at 4.5 kWh and supercapacitor at 3.8KWh with a peak
efficiency of 86%.
References
1. Geetha A, Subramani C, Thamizh Thentral TM (2018) An efficient wireless power transfer

using class E2 converter for electric vehicle. J Adv Res Dyn Control Syst 10:1–7
2. Qiu C, Chau KT, Ching TW, Liu C (2014) Overview of wireless charging technologies for
electric vehicles. J Asian Electr Veh 12(1):1679–1680
3. Wu Y, Zhang C, Liu B (2018) Research on the quick charging method of wireless power
transfer system for supercapacitor. In: 2018 13th IEEE conference on industrial electronics
and applications (ICIEA), pp 582–587
4. Sample AP, Meyer DT, Smith JR (2011) Analysis, experimental results, and range adaptation
of magnetically coupled resonators for wireless power transfer. IEEE Trans Industr Electron
58(2):544–554
5. Low ZN, Chinga RA, Tseng R, Lin J (2009) Design and test of a high-power high-
efficiency loosely coupled planar wireless power transfer system. IEEE Trans Industr Electron
56(5):1801–1812
6. Kim J, Kim J, Kong S, Kim H, Suh IS, Suh NP, Cho DH, Kim J, Ahn S (2013) Coil design
and shielding methods for a magnetic resonant wireless power transfer system. Proc IEEE
101(6):1332–1342
7. Miller JM, Onar OC, Chinthavali M (2015) Primary-side power flow control of wireless power
transfer for electric vehicle charging. IEEE J Emerg Selected Topics Power Electron 3(1):147–
162
8. Budhia M, Boys JT, Covic GA, Huang CY (2013) Development of a single-sided flux magnetic
coupler for electric vehicle ipt charging systems. IEEE Trans Ind Electron 60(1):318–328
9. Covic GA, Boys JT (2013) Modern trends in inductive power transfer for transportation
applications. IEEE J Emerging Selected Topics in Power Electron 1(1):28–41
10. Budhia M, Covic G, Boys J (2010) A new ipt magnetic coupler for electric vehicle charging
systems. In: IECON 2010—36th annual conference on IEEE industrial electronics society, pp
2487–2492
11. Elliott GAJ, Boys JT, Covic GA (2006) A design methodology for flat pick-up icpt systems.
In: 2006 1ST IEEE conference on industrial electronics and applications, pp 1–7
12. Omori H, Inoue M, Kimura N, Morizane T, Nakaoka M (2017) A new large-gap wireless ev
charger with a power superimposition communication. In: 2017 19th International conference
on electrical drives and power electronics (EDPE), pp 65–69
13. Schneider J, O’Hare J (2017) Alignment, verification, and optimization of high-power wireless
power charging systems. US Patent 9637014B2
Five-Level PUC Inverter-Based Shunt
Active Power Filter for Harmonic
Elimination
T. M. Thamizh Thentral, K. Vijayakumar, S. Usha, A. Geetha,

and C. S. Boopathi
Abstract The five-level packed U-cell (PUC) inverter topology is designed as an

active power filter to reduce the harmonics present in the nonlinear system. The
packed U-cell is a newly emerged multilevel inverter that can generate different
levels by utilizing capacitors at the output side. It is called packed U-cell since it has a
capacitor and a couple of power switches arranged in U shape. The energy conversion
takes place using a limited number of power semiconductor devices and capacitors,
thus reducing the cost of production. The model of a five-level PUC inverter is
designed, and the simulated waveforms along with their THD are discussed.
Keywords PUC5 inverter · Total harmonic distortion · Shunt active power filter ·
Harmonics
1 Introduction
Distribution network is one of the most important and complex parts of the power
system. Transmitting power from the distribution transformer to the power point has
challenging problems regarding reactive power and unbalance. Using more number of
switches and DC sources in power electronic converters by generating more voltage
levels with lower harmonics and smaller size of components is a field of research
still going on [1]. In many traditional multilevel converters topology such as flying
capacitors, neutral point clamped converts or classical cascaded H bridges which
have lesser total harmonic distortion (THD) in current and voltage mostly have many
switches required for the operation which results in complexity of circuit, increasing
the cost decreasing the reliability [2].
Most problems arise due to the presence of power electronic converters as these
converters generate harmonics in the source current and increase relative currents.
The main concerns with power converters are maintenance of unity power factor
throughout the operation and a relatively low total harmonic distortions of the input
T. M. Thamizh Thentral (B) · K. Vijayakumar · S. Usha · A. Geetha · C. S. Boopathi

https://doi.org/10.1007/978-981-16-4943-1_16
174 T. M. Thamizh Thentral et al.
AC waveforms [3]. By producing more levels of voltage, the multilevel converters are
able to decrease the current and voltage harmonics while keeping a low switching
frequency. To overcome various drawbacks of multilevel inverter, the packed U-
cell inverter was designed that offers the advantage of less number of switches,
thereby reducing switching losses and production cost. A five-level voltage wave-
form is obtained at the output by applying various voltage balancing techniques for
the capacitor. By making use of redundant switching states, the capacitance can be
reduced, and hence, small size capacitors can be used [4].
In this paper, the active power filter role in harmonic distortion is discussed in
Sect. 2. The PUC5 configuration and switching states are discussed in Sects. 3and
4, respectively. The modeling of a single-phase inverter with feedback is simulated
with the help of MATLAB/Simulink as discussed in Sect. 5.
2 Active Power Filters for Harmonic Elimination
Active power filters are now being utilized to compensate for both current and voltage
harmonics (depending on connection—shunt or series), power factor correction and
power quality improvement in power system networks. The active power filters are
made up of power converters and passive devices that have been designed to elim-
inate harmonic content and improve power quality [5]. They started to gain more
popularity recently due to their ability to compensate for harmonics demanded by
various nonlinear loads. Current harmonics caused due to distortion by nonlinear
loads results in decrease of power quality due to high THD. In order to solve these
problems, active filters are introduced in the grid to reduce harmonics and improve
the power factor [6].
Active filters are widely classified into three types, namely shunt, series and
hybrid. The shunt active filter purpose is to cancel load harmonics present in the
current fed to the supply. The advantage of using shunt apart from reactive power
compensation is that it can transfer the compensation current with small amounts
of active fundamental current so as to compensate for the losses in the systems [7].
The shunt active power filter is commonly composed of a voltage-source converter
(VSC) or a current-source converter with DC link capacitors mainly [8].
APFs can be controlled by two methods, i.e., direct mode or indirect mode. In
direct mode, the reference current is taken as the extracted harmonic components
of the load current. The harmonic extraction can be done by filters, power theories,
neural networks, fast individual harmonic extraction technique, Park transformation,
wavelet transformation, etc. [9]. In indirect mode, the AC input current is in phase
with the standard voltage without having to use any harmonic extraction component.
This mode has simpler system structure and better harmonic performances, and
hence, the one controller that it used for generating a signal for reference is modulated
and sent to switches with required pules. The PUC converter here acts as an active
power filter connected in shunt, hence bringing about harmonic elimination in the
load current [10].
Five-Level PUC Inverter-Based Shunt Active Power … 175
3 PUC5 Configuration Design
The PUC5 consists of packed U-cells in which each cell has an arrangement of two
switches and one capacitor.
The number of levels can be obtained by the relation:
2(N +1) − 1 = Number of levels
where N = 1,2,3.
The number of capacitors is obtained by the relation:
N = 2(N c+1) − 1
where N is the number of voltage levels and Nc is the number of capacitors required.
From the equations, it is understood that the topology not only uses a single DC
source, but also reduces the number of switches to obtain the required voltage levels.
4 PUC5 Configuration Design
Figure 1 shows the single-phase PUC5 inverter topology. The different possible
switching states are listed in Table 1. The on state of the switches is represented with
“*” in the table, and off state is mentioned as “-”. Although the PUC7 inverter has
a number of components as compared to other converters, switching frequency is
high and regulation of capacitor voltage becomes complicated. Hence, it is easier to
go with five-level converters, where the capacitor voltage is regulated at half of dc
supply value, thus eliminating current and voltage sensors.
From the following eight switching states, eight different paths of current flow
are obtained through the system. For generating five levels, six switches and two
capacitors are arranged in a U-cell in such a way, that in each half cycle, a total of
three switches conduct and a single switch in each leg.
The different levels in output waveform are obtained by using unequal DC sources.
Assuming that the capacitor voltages V1 = 2V2 = 2E, the PUC inverter will operate
as a five-level inverter with the output voltage as 0, ±E, ±2E. To obtain a voltage
level of say +2E, the switches S1, S5 and S6 must be on. Similarly, to get voltage level
−2E, the switches S3, S2 and S4 must be triggered. To obtain five levels, there will
be a total of eight different possible switching states. The capacitor can be charged
in one half cycle and discharged in next or vice versa. In order to fix the voltage of
the capacitor constant, the capacitor here is charged at each positive half cycle and
discharged in the negative half cycle.
To produce the desired voltage level, the source voltage is fixed at twice the input
DC source E, so that the capacitor can be charged up to E. To produce equivalent
Fig. 1 Per-phase PUC5 converter topology
Table 1 Switching states of PUC inverter

States S1 S2 S3 Output voltage Vt
1 * – – V1 2E
2 * – * V1–V2 E
3 * * – V2 E
4 * * * 0 0
5 – – – 0 0
6 - – * – V2 – E
7 – * – V2–V1 – E
8 – * * – V1 – 2E
Fig. 2 Block diagram representation of three-phase PUC5 inverter with nonlinear load
charging and discharging times in one period, here in state 2, the capacitor charges
from source while it is discharged in the next negative half cycle. The charging of
capacitor and discharging time depend on the load value only. This in turn also affects
the size of capacitor as larger loads will need small-sized capacitors in DC link and
thus also reducing the cost. The block diagram of the proposed three-phase PUC5
inverter-based nonlinear system proposed in this paper is shown in Fig. 2.
5 Experiment Results
The five-level PUC converter in shunt to the AC source system shown in Fig. 3
has been built in MATLAB/Simulink. The PI controller is used for generating the
waveform which is obtained from the difference of the reference and carrier signal
and injected between the load and AC source. The test parameters are as shown in
Table 2. The source currents and voltage waveforms and also output voltage and
current waveforms are shown in Fig. 4. The load, filter-side voltage and current are
also shown. The THD levels before and after introducing the converter are illustrated
in Fig. 5.
Fig. 3 Three-phase five-level PUC5 inverter with nonlinear load
Table 2 PUC5 inverter

Parameter Parameter value
parameter values
AC grid voltage 110 V
AC grid frequency 50 Hz
Filter-side inductor (Lf) 2.3 mH
Load-side inductor (Ll) 30 mH
Load-side resistance (Rl) 30
Load-side capacitance 600 µF
The positive and negative sides are equal and opposite to each other and together
form the upper and lower switches which in total forms an eight-switch combina-
tion. The vector values are generated from the respective voltage outputs from the
eight combinations which are zero in two cases and positive or negative in six of
them. A hexagon with an equal angle of sixty degree is formed with the nonzero
vectors. Therefore, space vector modulation technique generates a voltage in accor-
dance with the vectors calculated and generated due to the eight combinations. The
implementation of space vector modulation is found to be very adaptable to changing
supply configuration. Since the injection point is independent of load configuration
in this closed loop system, space vector modulation provides high accuracy in this
implementation. Eight possible combinations of on and off patterns may be achieved.
Fig. 4 Input and output current and voltage waveform
The on and off states of the lower switches are the inverted states of the upper ones
(Fig. 6).
It is clear that the, since the converter is introduced in shunt, there is significant
compensation by the APF in the current harmonics at the load side. Initially, before
the converter was introduced, the THD factor was found to be 23.25%. After the
introduction of the PUC5 converter, the THD level has dropped down to 2.03%. On
observing the waveforms for current on the load side, the harmonics are reduced
considerably. Since there is feedback given, according to changes in the load side,
then the compensation current will be injected to the load side at the point of common
coupling. The filter-side current which is responsible for the compensation harmonics
is shown in Fig. 5.
Fig. 5 a Source current % THD before filter, b source current % THD before filter
6 Conclusion
In this paper, the new technology converter, PUC5 was designed integrated with a
proportional and integral controller providing a feedback in a closed loop system and
studied. The PUC5 converter is connected in the AC grid in parallel with constant load
in order to reduce the current harmonics. The harmonics in the current are observed
to be reduced from 23.25 to 2.03% with the injection of current waveform which is
a negation of the reference current signal waveform and the carrier waveform. This
result is obtained when the load is operated at a value of 100%, and it varies with
the intake of load current by a variable nonlinear load if connected at the load side.
The PUC5 converter, therefore, would be highly advantageous as it is very easy to be
retrofitted in any kind of existing load configuration without the need to be modified
and executing the elimination of harmonics in AC power grid systems.
Fig. 6 Load and filter current and voltage
References
1. Vahedi H, Labbé PA, Al-Haddad K (2016) Sensor-Less five-level packed U-Cell (PUC5)
inverter operating in stand-alone and grid-connected modes. IEEE Trans Ind Informatics 12(1)
2. Tariq M, Meraj M, Azeem A, Maswood AI, Iqbal A, Chokkalingam B (2018) Evaluation
of level-shifted and phase-shifted PWM schemes for seven level single-phase packed U cell
inverter. PSS Trans Power Electron Appl 3(3)
3. Abarzadeh M, Vahedi H, Al-Haddad K (2019) Fast sensor-less voltage balancing and capacitor
size reduction in PUC5 converter using novel modulation method. IEEE Trans Ind Inf 15(8)
4. Vahedi H, Al-Haddad K () PUC5 inverter—a promising topology for single-phase and three-
phase applications. In: Conference: IECON 2016—42nd annual conference of the IEEE
industrial electronics society, October 2016
5. Wu JC, Jou HL, Feng YT, Hsu WP, Huang MS, Hou WJ (2017) Novel ciruit topolgy for three
phase active power filter. IEEE Trans Power Delivery 22(1)
6. Kocaeli University, Technical Education Faculty, Electrical Education Department (2009)
Operation of shunt active power filter under unbalanced and distorted load conditions
7. Jacob A, Abraham BT, Prakash N, Philip R (2014) A review of active power filters in power
system applications. Int J Adv Res Electr Electron Instrum Eng 3(6)
8. de Almeida Carlos GA, Jacobina CB, Me´llo GPRA, dos Santos EC Jr (2017) Shunt active
power filter based on cascaded transformers coupled with three-phase bridge converters. IEEE
Trans Ind Appl 53(5)
9. Busawon K, Putrus GA, Ran L (2005) Fast-individual-harmonic-extraction technique. Proc
Gener Transm Distrib 152(4)
10. Vahedi H, Shojaei AA, Chandra A, Al-Haddad (2016) Five-Level reduced-switch-count boost
PFC rectifier with multicarrier PWM. IEEE Trans Ind Appl 52(5)
Performance of 100-kW Rooftop PV
Plant in Library Building—A Case Study
in SRMIST
J. Divya Navamani, Tanmay Padhi, Aditi Kumari, and A. Lavanya
Abstract In this paper, performance analysis of 100-kW rooftop PV plant in library

building of SRM Institute of Science and Technology, Kattankulathur, Chennai, is
carried out. The site description and the plant capacity are mentioned as the details
obtained by the department, and calculation was performed on the same data. The
performance of the system over last three academic years has been compared. Further-
more, the analysis is extended by comparing this 100-kW PV plant with the similar
PV plant in the institute. The configuration of the PV plant taken for the compar-
ison is the same 100-kW with similar structure. The findings from the study and
comparison are added finally with the causes of the shortfall in the performance of
the system. Finally, the inferences are concluded which will guide the researchers to
find the solutions for the cause and effect observed in this 100-kW PV plant.
Keywords PV · Energy yield · Performance assessment
1 Introduction
As the modern world focuses more and more on renewable energy sources, solar
energy has become one of the most used energy forms across the world. Many
researches are going on to enhance its use in everyday life and how uses of this
energy can be further more expanded. It has been witnessed that a solar thermal
collector is used to convert energy into heat and simultaneously panels are used
J. Divya Navamani (B) · T. Padhi · A. Kumari · A. Lavanya

T. Padhi
A. Kumari
A. Lavanya
https://doi.org/10.1007/978-981-16-4943-1_17
184 J. Divya Navamani et al.
for generating electricity [1]. Since PV modules are made up of silicon cell with
lesser efficiency, it is necessary to work on the efficiency of the PV module to attract
many investors [2]. The system, however, needs to be sustainable and reliable, and
it totally depends on the optimization which can meet the demands of load [3].
The performance is also dependent on the environment, location and plant types,
so these things are important during installation for any system [4]. Recently, many
initiatives have been taken by the Government to install the panels over the roof in
many of the offices and institutions which can help to tackle the crisis [5]. Also, it
can reduce the effect of greenhouse gas produced through fossil fuel. The PV energy
is comparatively less expensive than any other renewable source, and it has been
found to be more profitable in rural areas [6]. Another factor which creates impact
on the performance is shading effect on the system, and it can influence the output
power and output energy of the PV system connected [7].
Various studies are performed in this literature on performance analysis of PV.
This study has been done on the PV system present on rooftop of library building
of SRM Institute of Science and Technology, Kattankulathur, India. There are 20
strings, and each string has 20 panels connected to it. The details of the plant and
the size have been given in the below part. Also, in the later part of this literature,
comparative study has been done. Performance for three PV plants was taken, and
energy yield and other parameters were calculated.
2 Description of the Plant
The PV power plant was installed in 2014 on rooftop of library building of SRM
Institute of Science and Technology, Kattankulathur-603,203. The grid-connected
plant consists of 20 strings each of which has 20 panels connected in series. The
overall installed capacity is 100-kW. The PV strings are connected to nine inverters
out of which four strings are connected to Inverter 1 and remaining eight inverters
are each connected with two strings. The rating of each panel is 250 W, and it has
400 panels which are put together to get 100-kW rating. The schematic diagram of
the modules connected and the sizing of inverter is given in Fig. 1a.
The photographs of the mono and polycrystalline solar PV module installed in
the plant are depicted in Fig. 1b, c. The nine inverters which are connected to the
strings of the PV module are illustrated in Fig. 1 d, e. The rating of each inverter
with their model/make is also presented in Table 1.
Initially, we started the study by analysing all the influencing parameters of the
energy yield from the solar PV plant. Solar irradiation data obtained from the PVsyst
7.1 are analysed from January to December, and the observations are plotted in
Fig. 2a, b.
Performance of 100-kW Rooftop PV Plant … 185
20
20 Inverter 1
Total 20 strings (20 panels in each string)
20 SOLIVIA 20 EU4
Number of panel = 400
20
20 Inverter 2
20 SOLIVIA 11 EU4 Grid
14 Strings connected to 6 inverters(2 strings for

each inverter)
20
Inverter 9
20 Delta RPI M10A
(a)
(b) (c)
(d) (e)
Fig. 1 (a) Schematic diagram of 100-kW plant, (b) and (c) solar PV module connected (d) and e
Inverter connected to the strings
Table 1 Details of the

Name Type Size (kWp)
inverters in the plant
Inverter 1 SOLIVIA 20 EU4 20
Inverter 8 RPI M10A 10
Inverter 9 RPI M10A 10
Fig. 2 a Solar irradiation in the location, b wind speed and relative humidity
3 Performance Analysis of the Plant
This section compares the performance of 100-kW PV plant in the library building
for the year 2018–2020. Figure 3a, b presents the graphical representation of the
plant for all the year from January to December. From this comparative study, it is
observed that the performance of the plant in the year 2018 is poor compared to next
two consecutive years. The average performance ratio of this plant for the year 2018
is 41% whereas for the year 2019 and 2020 is 70% and 64%, respectively.
4 Comparative Analysis
In the comparative study, the first comparison is carried out for the 100-kW PV plant
for the year 2018–2020. Secondly, the same plant is compared with the similar plant
for the year 2020.
A. Comparative study on the PV plant in Library Building:
Fig. 3 a Final yield for last

three years (2018–2020),
b comparison of
performance ratio for three
years (2018–2020)
Table. 2 Energy yield and energy produced by 100-kW solar system for the year 2018–2020
Date Total Yield in Date Total Yield in Date Total Yield in
Wh Wh Wh
01-01-2018 3,822,359 01-01-2019 11,002,279 01-01-2020 12,855,238
01-02-2018 3,972,767 01-02-2019 11,008,333 01-02-2020 10,793,741
01-03-2018 13,041,601 01-03-2019 13,673,836 01-03-2020 12,922,897
01-04-2018 13,458,374 01-04-2019 13,527,115 01-04-2020 8,044,220
01-05-2018 9,293,124 01-05-2019 12,933,989 01-05-2020 10,422,097
01-06-2018 4,328,229 01-06-2019 11,228,847 01-06-2020 10,218,172
01-07-2018 4,491,192 01-07-2019 11,133,617 01-07-2020 9,713,176
01-08-2018 4,222,856 01-08-2019 12,275,602 01-08-2020 9,690,844
01-09-2018 4,896,087 01-09-2019 9,164,699 01-09-2020 9,149,403
01-10-2018 6,605,451 01-10-2019 7,277,069 01-10-2020 9,897,554
01-11-2018 3,891,626 01-11-2019 10,623,177 01-11-2020 8,427,585
01-12-2018 8,989,861 01-12-2019 9,508,014 01-12-2020 7,583,870
This section discusses about the comparison of performance of library building

for last three (2018, 2019 and 2020) years. The 100-kW PV plant in library building
was installed in the year 2014. Table 2 presents the total energy yield in Wh for three
years. The graphical representation of this comparison is illustrated in Fig. 4.
In this section, three 100-kW PV plants are compared, and their performances were
studied. Table 3 depicts the energy yield for the year 2020 along with kWh/kWp of
all the PV plant in the institute. All the three PV plants are installed in the same
latitude and longitude with 500 m distance between them. The solar irradiation on
all the three plants is more or less the same when we observed the NREL data from
PVsyst 7.1 software. To deliberately observe the energy yield output of the plant
from January to December 2020, a graphical representation is presented in Fig. 5.
From this figure, it is noted that the performance of the three plants is quite closer
Fig. 4 Comparison of energy yield for three years (2018–2020)
Table 3 Energy yield and energy produced by solar array in the year 2020
Date Library Library Dental Dental Medical Medical
Building Building College College College College
(Wh) (kWh/kWp) (Wh) (kWh/kWp) (Wh) (kWh/kWp)
Jan-20 12,855,238 128.55 11,358,200 113.58 12,974,100 129.74
Feb-20 10,793,741 107.94 12,370,000 123.7 12,669,400 126.69
Mar-20 12,922,897 129.23 13,624,300 136.24 13,668,900 136.69
Apr-20 8,044,220 80.44 4,098,000 40.98 12,191,800 121.92
May-20 10,422,097 104.22 11,406,900 114.07 10,723,600 107.24
Jun-20 10,218,172 102.18 11,395,400 113.95 9,058,000 90.58
Jul-20 9,713,176 97.13 11,188,500 111.89 9,580,700 95.81
Aug-20 9,690,844 96.91 10,890,100 108.9 10,710,900 107.11
Sep-20 9,149,403 91.49 7,340,900 73.41 10,058,900 100.59
Oct-20 9,897,554 98.98 10,434,400 104.34 10,719,100 107.19
Nov-20 8,427,585 84.28 9,232,800 92.33 9,951,800 99.52
Dec-20 7,583,870 75.84 8,127,300 81.27 4,684,300 46.84
Fig. 5 Comparison of energy yield in Wh for three 100-kW PV plants (2020)
in the month of January, May, July, October and November. It is also observed that
the performance of dental college is poor compared to other two plants due to the
obstacles and other shading related issues by nearby building. The altitude of library
and medical college is also noted to be high compared to the dental college.
5 Inference from the Study
We made a complete analysis on the PV plant installed in library building of SRMIST.

It has 400 PV panels. This plant is installed in the year 2014. Initially, it is installed
with 400 numbers of monocrystalline PV modules with the rating of 250 W which
is conveyed to the total of 100-kW.
Figure 6a, b presents the picture of the library building and satellite image of the
building with solar panel where the 100-kW plant is installed. Figure 6c illustrates
the changes made in the solar PV module in the year 2017–18. It was reported that the
speed of Vardah cyclone was 155 km/hour and it causes $1 billion worth destruction
in Tamil Nadu. The hurricane strength uprooted many tress and poles in the state. In
SRMIST, it uprooted the installed PV array from the rooftop of the building, and it
falls on the other PV modules nearby. This damages around 45-kW of the 100-kW
PV system.
6 Conclusion
This paper elaborates the study carried out on the 100-kW solar PV plant in library
building in SRM Institute of Science and Technology. The plant has 400 number
of PV panel with 250 W rating, and it has nine inverters with different ratings. The
performance of this plant is studied, and major findings from this study are as follows.
(a) (b)
(c)
Fig. 6 a Library building in SRMIST, b satellite image of library building with 100-kW PV plant,
c changes made in the PV panels after cyclone in the year 2016
• Initially, the plant consists of monocrystalline PV panel, and it was installed in

the year July 2014.
• The plant was exposed to two cyclones which were occurred in the year 2016 and
2020.
• Due to Vardah cyclone, around 40–45 kW of PV plant was damaged completely.
After a year, the damaged PV module is replaced with polycrystalline PV panel.
• After the replacement, there was several issues in the plant, and the performance
of the plant in the year 2018 is 41%.
It is observed that the plant performance degrades after it was subjected to the
cyclone. The other comparison reveals that the performance of 100-kW plant in
dental college is quite low compared to 100-kW plant in library and medical college
in SRMIST.
References
1. Tarigan E (2020) Hybrid PV-T solar collector using amorphous type of solar cells for solar
dryer. In: 2020 International seminar on intelligent technology and its applications (ISITIA),
Surabaya, Indonesia, pp 352–356. https://doi.org/10.1109/ISITIA49792.2020.9163789
2. Ashwini K, Raj A, Gupta M (2016) Performance assessment and orientation optimization of
100-kWp grid connected solar PV system in Indian scenario. In: 2016 International conference
on recent advances and innovations in engineering (ICRAIE), Jaipur, pp 1–7. https://doi.org/10.
1109/ICRAIE.2016.7939505.
3. Husain NS, Zainal NA, Mahinder Singh BS, Mohamed NM, Mohd Nor N (2011) Integrated
PV based solar insolation measurement and performance monitoring system. In: 2011 IEEE
colloquium on humanities, science and engineering, Penang, Malaysia, pp 710–715https://doi.
org/10.1109/CHUSER.2011.6163827
4. Zou X, Jiang F, Liu H (2016) Performance analysis of a rooftop PV plant and a desert PV plant. In:
2016 Chinese control and decision conference (CCDC), Yinchuan, China, pp 6173–6176https://
doi.org/10.1109/CCDC.2016.7532107
5. Sharma R, Goel S (2017) Performance analysis of a 11.2 kWp roof top grid-connected PV
system in Eastern India. Energy Rep 3:76–84. ISSN 2352-4847. https://doi.org/10.1016/j.egyr.
2017.05.001.
6. Maammeur H, Hamidat A, Loukarfi L, Missoum M, Abdeladim K, Nacer T (2017) Perfor-
mance investigation of grid-connected PV systems for family farms: case study of North-West
of Algeria. Renew Sustain Energy Rev 78:1208–1220. ISSN 1364-0321. https://doi.org/10.1016/
j.rser.2017.05.004.
7. Dabou R et al (2017) Impact of partial shading and PV array power on the performance of grid
connected PV station. In: 2017 18th International conference on sciences and techniques of
automatic control and computer engineering (STA), Monastir, Tunisia, pp 476–481. https://doi.
org/10.1109/STA.2017.8314901
Design and Implementation
of a Low-Cost Mini Heliostat Solar
Tracking System in West Africa
Marie Pascaline Sarr, Ababacar Thiam, Biram Dieng,

and El Hadji Ibrahima Cisse
Abstract This paper proposes the design and implementation of a low-cost auto-
matic mini heliostat solar tracker system. The tracking system is designed as open
loop control based on astronomical equations. The experimental device consists of
a 0.3 × 0.3 m2 reflector, 3-mm thick with an Arduino card and two stepper motors.
The prototype tracking system has been tested during three days. The influence of the
different parameters like number of the day, time, heliostat position and references of
the tracking axis was analysed. The results obtained show some improvement with
RMSE of 0.050 and 0.085, respectively, for the altitude and azimuth. This system
can be used for small applications in rural areas
Keywords Mini heliostat · Solar tracking · Open loop control · Astronomical

equations · Experimental
1 Introduction
The lack of access to reliable and affordable energy electricity remains a key
constraint to achieving economic and social development of the West African. Access
to electricity in this region is at 52% with a great inequality between rural and urban
areas [1]. Demand in rural zones is too low to be satisfied and considered like losses by
the national electricity companies. With the significant solar potential of this region,
5.5 kWh/m2 /day [2], the concentrating solar power (CSP) technologies constitute an
excellent alternative to satisfy energy needs in electricity, hot water, cold, cooking,
M. P. Sarr (B) · A. Thiam

Efficacité et Systèmes Energétiques, Alioune Diop University of Bambey, Bambey, Sénégal
A. Thiam · E. H. I. Cisse
Laboratoire Eau, Energie, Environnement et Procédés Industriels, Ecole Supérieure
Polytechnique, Dakar, Sénégal
B. Dieng
Energies Renouvelables, Matériaux et Laser, Alioune Diop University of Bambey, Bambey,
Sénégal
https://doi.org/10.1007/978-981-16-4943-1_18
194 M. P. Sarr et al.
drying of products [3–5]. Among CSP technologies, solar tower power (STP) is
identified as the most efficient way for capturing and converting solar energy into
heat and electricity [6, 7]. However, the high cost of these plants due to their large
size makes this technology non-existent in West Africa [8]. The apparition of micro
solar tower power generates growing interest because of their high flexibility and
low-investment costs [9]. The small size heliostats are easier to be manufactured,
installed and cleaned. Moreover, the mini heliostats can be designed with local low-
cost materials creating job opportunities for local people. However, the only major
challenge remains the mastery of heliostats tracking in West Africa [10]. Accurate
tracking is essential for a better efficiency of STP. Nevertheless, improving precision
of heliostat tracking increases the complexity and cost of tracking. To reduce the
cost of heliostat tracking, error correction algorithms are often incorporated in the
tracking program in order to obtain and maintain good accuracy. Several studies have
carried on the different error sources and their correction strategy.
Baheti and Scott [11] have developed an error correction method to compensate
mechanical errors related to the heliostats movement. This model included base tilt,
azimuth and elevation bias offsets, and non-orthogonal drive axes. A closed-loop
solar sensor was used to periodically point the normal vector of the heliostat directly
to the sun. The tracking errors of the heliostat were obtained by comparing the
commanded position of the heliostat at each interval with the actual solar position as
calculated by the solar algorithm. These data were used to estimate the error model
coefficients and resulted in final RMS error reductions of 10:1 and 5:1 for azimuth and
elevation tracking, respectively. Stone and Jones [12] studied the geometric errors that
reduced the tracking accuracy of the heliostat at solar two. Three dominant sources of
error with their effect on heliostat tracking were described. These are the tilt of the axis
of rotation in azimuth, the encoder reference position error and the non-orthogonality
errors. The beam characterization system (BCS) is used to measure heliostat tracking
errors and improve heliostat tracking accuracy by adjusting the encoder reference
position errors. The strategy presented has been able to minimize but not correct these
sources of error. Khalsa et al. [13] studied the effect of non-random physical errors
(baseplate tilt about the East–West axis, baseplate tilt about the North–South axis,
azimuth reference bias, elevation reference bias, azimuth linear error, elevation linear
error, drive axis non-orthogonality and aiming error) on heliostat tracking. The data
was obtained by directing a single-test heliostat at a large white target on the tower.
The offset between the center of the reflected beam and the center of the target was
determined using optical image analysis to calculate tracking errors. This technique
was effective in correcting errors for a single-summer day. However, an accuracy
reduction in winter was observed. Berenguel et al. [14] have described a method
for automatic correction of angular offset with heliostat reflected image processing
techniques. A CCD camera is placed in front of the tower just before the solar field in
order to photograph the images reflected by the heliostats. For each reflected image,
its centroid is calculated and compared to the center of the receiver. The distance
between these two points constitute the error to be rectified. A maximum distance
of 7.4 cm has been obtained. Malan and Gauché [15] performed a method of esti-
mating and correcting tracking errors similar to [13] by adding errors related to very
Design and Implementation of a Low-Cost Mini Heliostat … 195
small installation and manufacturing tolerances. Using image processing techniques,

the tracking error offsets of each individual heliostat are determined periodically to
estimate the coefficients of the error model. Prototype experiments indicate a daily
open loop tracking error of the RMS normal vector of less than 1 mrad. Tracking
errors may also be caused by disagreements between the coordinate systems used by
solar engineers, civil engineers and surveyors, during the process of surveying a CSP
site as discussed by Gross and Balz [16]. To avoid this, transformations between the
different geodetic, Cartesian and Universal Transverse Mercator coordinate systems
have to be performed. Sun et al. [17] presented a strategy for correcting heliostat
tracking errors based on beam characterization system. The Hartley-Meyer algo-
rithm, which is a Gauss–Newton algorithm modified to solve nonlinear least squares
equations, is used to determine the six angular tracking parameters of a specific
heliostat. A reduction in tracking errors is noted from 3.14 to 0.73 mrad. However,
after six months of operation, the tracking accuracy decreases.
The purpose of this paper is to identify and study the different sources of errors in
the tracking of an experimental setup of mini heliostat in West Africa. An open loop
tracking system with some adjustments is proposed. Errors due to axis reference and
mirror tilt have been determined and taken into account after several tests based on
repeatability. The influence of the day and the heliostat position on the evolution of
the beam reflected has been studied.
2 Materials and Methods
The optical performance of the tracking system is evaluated using a mini heliostat
with a reflective surface of 0.3 × 0.3 m2 . An algorithm based on astronomical equa-
tions is used to determinate the sun and mini heliostat position according to the
geographical coordinates of the site, the day and the tower position. A command is
sent to the engines every 5 min. Errors are estimated by comparing between centroid
of beam reflected and target center so that this error signal can be used for adjustment
purposes. After several days of testing, systematic errors due to reference and sensor
tilt problems are corrected.
The heliostat tracking is done in open loop with the astronomical equations. Firstly,
solar vector S is determined with geographic coordinates and number of the day. The
sun position is identified by the altitude and azimuth angles obtained from Eqs. (1–4)
[18] as illustrated in Fig. 1:
−
→ −
→ −
→ −
→
S = Sz i + Se j + Sn k (1)
where
Sz = sin α (2)
Fig. 1 Coordinates of the sun, heliostat and beam reflected on the tower
Se = cos α sin A (3)
Sn = cos α cos A (4)
The tower is placed to south of the mini heliostat and is the origin. It is defined
by (5):
−
→ −
→ −
→ −
→
T = Tz i + Te j + Tn k (5)
where
Tz = sin λ (6)
Te = cos λ sin φ (7)
Tn = cos λ cos φ (8)
After, the mirror position is determined according to the law of Snell Descartes
knowing the position of the tower
−
→ − →
−
→ S + T
H = (9)
2 cos θ
This vector can be defined by the altitude αh and azimuth Ah angles of the mirror
sin α + sin λ
sin αh = (10)
2 cos θ
cos α sin A + cos λ cos φ
sin Ah = (11)
2 cos θ
where represents the incidence angle
3 Description of Experimental Setup
The setup on which the experimentation performed is shown in Fig. 2. The experi-
mental tests were conducted at Dakar, Senegal with latitude: 14.7°N and longitude:
17.4°W. The reflected surface is a mirror of 0.3 × 0.3 m2 with a thickness of 3 mm.
The mechanism of each axis is driven by a stepper motor of 400 steps per rotation via
a 10:1 transmission system to achieve a final output axis resolution of 4000:1. The
required motor torque has been calculated for an efficiency factor of 0.45 N.m. Limit
switches are installed at each axis of rotation to prevent mechanical movements from
exceeding the predefined limits.
To evaluate tracking performance, graduations of 5 cm per unit were made on the
target. The center of the target represented by the white point has coordinates (0, 0,
H). On the mirror surface, diagonals were made to detect the mirror center point for
evaluating easily the distance between target point and reflected beam (Fig. 3).
Every five minutes, a command is sent to the stepper motors to evaluate the
tracking errors (Fig. 4) using the method of Chiesi et al. [19]:

dA
ε A = a tan (12)
d
Fig. 2 Components of system

Fig. 3 Identification of heliostat and target centers
Fig. 4 Illustration of heliostat tracking error

dE
ε E = a tan (13)
d

d
ε = a tan (14)
D

where d = d A 2 + dE 2
dA azimuth error in cm.
dE altitude error in cm.
D distance between heliostat tower centers obtained by the laser distance meter
(m).
Table 1 Setup parameters of

Parameters
experiment
Height of heliostat (m) 1.10
Height of tower (m) 4.02
Heliostat surface (m2 ) 0.3 × 0.3
Distance between heliostat and tower’s center D (m) 4.56
Distance heliostat—tower (m) 3.50
Parameters of the experiment are presented in Table 1
An evaluation of the accuracy of the program is made by comparing the program

with the online calculator [20] as shown in Figs. 5, 6 and 7 for three different days.
The results obtained gave a mean absolute error (MAE) of 0.83, 0.43, and 0.32 for
the days 03/11, 10/11 and 14/11/2020. Tracking errors vary from day to day and can
therefore affect the optical performance of the tracking system.
The altitude and azimuth errors obtained experimentally for 03/11/2020 are
reported in Fig. 8. The target center was not reached with significant spills at the
end of the day. RMSE of 0.168 and 0.201 were obtained for the altitude and azimuth,
respectively. After many calibrations, an inconsistency in the references was noted
and taken into account at the tracking program. The absolute error was taken into
account at the algorithm level to try to reduce it. A second test was performed on the
day of 10/11/2020. Figure 9 shows the results obtained. The task was maintained on
the target center for a certain period of time. RMSE of 0.0486 and 0.113 were found
for the elevation and azimuth, respectively.
Fig. 5 Solar path of 03/11/20

Fig. 8 Tracking error obtained for 03/11/2020

Figure 10 illustrates the results obtained for the day of 14/11/2020. The spot was
held at the target for approximately 2 h. The RMSE obtained are 0.050 and 0.085
for elevation and azimuth, respectively.
The results obtained during these days show a reduction in errors. However, a
slight increase for the RMSE of the altitude of 14/11/2020. In addition, during the
experiments an evolution of the task was noted in one direction only, down and to the
right during the day as shown in the pictures in Fig. 11. An appropriate correction
can therefore easily be made. Analysis of the results of the experiments showed that
the azimuth axis recorded many more errors than the elevation axis. This may be due
to the fact that the wind acts more on the azimuth displacement than altitude.
A change of reflector shape is noted during the day on the receiver. Indeed, the
shape is normal until solar noon and then begins to distort afterward. This variation
of the shape depends on the heliostat position in the field and the time. During the
experiments, an error was noted on the conversion of the heliostat angles into steps.

Fig. 11 Optical images of the heliostat reflection on the target 10/11/2020 11h35. a, 12h30 b, 2h30
PM c, 14/11/20 à 11h35 d, 12h30 e, 2h30 PM f
In fact, one step is equal to 0.09°. If the decimal part is inferior at the one step, there
is not considered in the conversion. For a significant reduction, it is necessary to take
into account in the tracking algorithm the main sources of errors such as mechanical
design errors, leap seconds and computation time errors, approximations made during
the conversion of heliostat angles into steps, transmission systems.
5 Conclusion
In this paper, an experimental study of the optical performance of a mini heliostat

was carried out. The results obtained show a maximum deviation of the reflected
beam of 0.15 m and 0.2 m, RMSE of 0.050 and 0.085 for altitude and azimuth.
Analysis of the results obtained showed that parameters such as daylight, references
and mechanical design errors influence the mini heliostat tracking. The evolution of
the reflected image at the receiver is highly dependent on the position of the heliostat
relative to the tower. The error variation is much faster for the day of 03/11 and 10/11
than 14/11/20. An improvement is noted, but a better accuracy could be obtained by
taking into account more error sources in the algorithm. The development of the mini
heliostat will raise awareness about the integration of CSP for the satisfaction of the
thermal energy needs of the rural sub-Saharan population particularly conservation

of cereal and dairy products, water pasteurization.
Acknowledgements We are grateful to the WFS (Word Federation of Scientists) for the support.
References
1. Énergie solaire en Afrique, un avenir rayonnant ? https://www.lemondedelenergie.com/ene

rgie-solaire-afrique-avenir/2019/02/14/ (consulté le nov. 02, 2020).
2. Azoumah Y, Ramde EW, Tapsoba G, Thiam S (2010) Siting guidelines for concentrating solar
power plants in the Sahel: case study of Burkina Faso. Sol Energy 84:1545–1553. https://doi.
org/10.1016/j.solener.2010.05.019
3. Al-Amayreh MI, Alahmer A, Manasrah A (2020) A novel parabolic solar dish design for a
hybrid solar lighting-thermal applications. In: 2020 7th International conference on power and
energy systems engineering (CPESE 2020), 26–29 September 2020, Fukuoka, Japan. Energy
Reports 6(9):1136–1143 https://doi.org/10.1016/j.egyr.2020.11.063
4. Mekonnen BA, Liyew KW, Tigabu MT (2020) Solar cooking in Ethiopia: experimental testing
and performance evaluation of SK14 solar cooker. Case Studies Thermal Eng 22:100766.
https://doi.org/10.1016/j.csite.2020.100766
5. Brouche M, Lahoud C, Lahoud MF, Lahoud C (2020) Solar drying simulation of different
products: Lebanese case, Tmrees, EURACA, 25–27 June 2020, Athens, Greece. Energy Reports
6(6):548–564. https://doi.org/10.1016/j.egyr.2020.09.032
6. Bhargav KR, Gross F, Schramek P (2014) Life cycle cost optimized heliostat size for power
towers. Energy Procedia 49:40–49. https://doi.org/10.1016/j.egypro.2014.03.005
7. Agyekum EB, Velkin VI (2020) Optimization and techno-economic assessment of concentrated
solar power (CSP) in South-Western Africa: a case study on Ghana. Sustain Energy Technol
Assess 40:100763. https://doi.org/10.1016/j.seta.2020.100763
8. Seshie YM, N’Tsoukpoe KE, Neveu P, Coulibaly Y, Azoumah YK (2018) Small scale concen-
trating solar plants for rural electrification. Renew Sustain Energy Rev 90:195–209. https://
doi.org/10.1016/j.rser.2018.03.036
9. Renzi M, Egidi L, Cioccolanti L, Comodi G (2017) Performance simulation of a small-scale
heliostat CSP system: case studies in Italy. Energy Procedia 105:367–372. https://doi.org/10.
1016/j.egypro.2017.03.327
10. Abu-Hamdeh NH, Alnefaie KA (2016) Design considerations and construction of an experi-
mental prototype of concentrating solar power tower system in Saudi Arabia. Energy Convers
Manag 117. https://doi.org/10.1016/j.enconman.2016.02.077
11. Baheti R, Scott P (1980) Design of self-calibrating controllers for heliostats in a solar power
plant. IEEE Trans Autom Control 25:1091–1097. https://doi.org/10.1109/TAC.1980.1102527
12. Jones SA, Stone KW (1999) Analysis of solar two heliostat tracking error sources. In: ASME
international solar energy conference. https://digital.library.unt.edu/ark:/67531/metadc687954
13. Khalsa SS, Ho CK, Andraka CE (2011) An automated method to correct heliostat tracking
errors. SolarPACsES 2011
14. Berenguel M et al (2004) An artificial vision-based control system for automatic heliostat
positioning offset correction in a central receiver solar power plant. Solar Energy 76:563–575.
https://doi.org/10.1016/j.solener.2003.12.006
15. Malan K, Gauché P (2014) Model based open-loop correction of heliostat tracking errors.
Energy Procedia 49:2118–2124. https://doi.org/10.1016/j.egypro.2014.03.224
16. Gross F, Balz M (2020) Potentially confusing coordinate systems for solar tower plants.
SolarPaces 2020. In: AIP conference proceedings, vol 2303, no 1, Dec 2020
17. Sun F et al (2015) Study on the heliostat tracking correction strategies based on an error-
correction model. Sol Energy 111:252–263. https://doi.org/10.1016/j.solener.2014.06.016
18. William B (2001) Stine and michael geyer power from the sun. Netbook http://www.powerf
romthesun.net/book.html
19. Chiesi M, Scarselli E, Guerrieri R (2017) Run-time detection and correction of heliostat tracking
errors. Renew Energy 105. https://doi.org/10.1016/j.renene.2016.12.093
20. SunCalc sun position- und sun phases calculator. https://www.suncalc.org
Power-Domain NOMA for Massive
Connectivity in Smart Grid
Communication Networks
M. Jayachandran and C. Kalaiarasy
Abstract The exploitation of information and communication technology (ICT) and

the development of smart electricity networks have become the main concerns world-
wide. To leverage ICT in the existing electrical power network, 5G wireless systems
are integrated toward the development of smart grid networks. For reliable, efficient,
and secure communication infrastructure in a smarter electricity network, the non-
orthogonal asynchronous transmission is essential. Besides, a massive number of
smart meters (SMs) measure energy consumption and convey instantaneous infor-
mation to the utility through communications networks. However, data rates beyond
10 Gbps, connectivity to support 1 million per sq km device density, and sub-
millisecond latencies are challenging issues when implementing a smart distribution
grid. Therefore, this paper presents the non-orthogonal multiple access (NOMA)
scheme and optimized power control strategy for smart microgrids to integrate
numerous sensor devices with higher data transfer rates and lower latencies. This
new microgrid configuration enables multiple electricity users to transmit and receive
data simultaneously using the same frequency and enables optimized power flow with
high flexibility.
Keywords 5G technology · NOMA · ICT
1 Introduction
The increase of power demand in the existing power network causes severe chal-
lenges in ensuring reliable, efficient, and economic operation. Lack of intelligent
mechanisms, the conventional grids prone to power outages and blackouts. There-
fore, smart grids have gained significant popularity in recent years [1]. Smart grid is
an advanced electricity supply network that enables monitoring, sensing, control, and
M. Jayachandran (B) · C. Kalaiarasy

Puducherry Technological University, Puducherry 605014, India
C. Kalaiarasy
https://doi.org/10.1007/978-981-16-4943-1_19
206 M. Jayachandran and C. Kalaiarasy
communication for improving the reliability, safety, and efficiency of power system
operation. In this regard, modern digital communication technologies are essential
for the reliable and efficient functioning of the smart grid network (SGN) [2, 3].
Wireless communication technologies are becoming more popular for smart grid
applications due to their flexibility and low infrastructure costs. Compared with the
4G technology, 5G has higher system capacity, throughput, and energy efficiency
per service. Multiple access (MA) techniques can be regarded as one of the most
fundamental enablers in wireless networks. They can be categorized into orthogonal
multiple access (OMA) and NOMA methods. In OMA schemes like FDMA, TDMA,
CDMA, and OFDMA, multiple users are not permitted to share the same resource
concurrently in frequency, time, code, and subcarrier domains, respectively. On the
contrary, simultaneous sharing of resources leads to signal interference resulting
in loss of data. For instance, an OFDMA network with 64 subcarriers allows only
64 users simultaneously. This orthogonality imposes the capacity limitation. This
problem can be resolved by allowing the simultaneous transmission of multiple
users in the same frequency carrier which is termed as NOMA. Recently, NOMA has
been recognized as one of the promising multiple access techniques for 5G wireless
networks. NOMA utilizes the power division multiple access techniques (PDMA)
which allocate different power to different users and depends on their distance from
the base station (BS) and channel condition [4]. NOMA schemes can be classified into
two categories, namely code-domain NOMA and power-domain NOMA. Recently,
power-domain NOMA has been proposed which exposes superior capacity than
OMA. [5] The fundamental concept of power-domain NOMA is to ascertain that
multiple users can be shared simultaneously within the same radio spectrum [6]. The
successful operation of NOMA is to form clusters of users with different channel
gains and allow them to transmit on the same radio spectrum with appropriate power
using superposition coding (SC) at the transmitter and decode the message signal of
different users at the receiver using successive interference cancelation (SIC) [7].
Considering the performance of NOMA in fading channels, NOMA in addi-
tive white Gaussian noise (AWGN) channels has recently investigated to exploit
the time-varying nature of multi-user channels. It has been reported that NOMA
achieves superior performance than OMA for various distances [8]. Recently, hier-
archical control is proposed for the microgrids. Energy management system (EMS)
in microgrids consists of three levels of hierarchical control system, namely primary,
secondary, and tertiary control. To optimize the microgrid resources and manage the
power flow between the microgrid and the grid, tertiary control is performed.
In this paper, a communication-enabled smart microgrid infrastructure along with
an optimized control technique using model predictive control (MPC) for the tertiary
control level is proposed. This smart microgrid configuration and the associated
control aim to achieve a fast and smooth grid connection of renewable as well as a
distributed power source.
Power-Domain NOMA for Massive Connectivity … 207
Fig. 1 Schematic diagram of smart microgrid system with wireless communication
The proposed smart microgrid configuration comprises a solar PV array, a wind

turbine, a battery energy storage system, and an electric load. All the units in this
system are parallelly connected to the common AC bus and rely on wireless communi-
cation as shown in Fig. 1. Solar PV and wind turbine are renewable power generation
units. The battery bank is an energy storage system that can smooth the intermittent
performance of RES and fluctuating load profile. To monitor the smart grid, smart
meters are installed for monitoring and measuring the real power, reactive powers,
and power flows. In this system, the power converters play a key role to optimize the
power flow between renewable sources and the grid.
3 a Typical NOMA Scheme
Concerning uplink transmission with a base station and N devices, smart grid units
send messages U 1 , U 2 , … U n to the base station. Then, the base station performs
superposition coding with these data and transmits the following NOMA signal into
the channel:
√ √ √ √
UNOMA = p[ pU1 + pU2 + · · · + pn Un ] (1)
where P is the total transmit power. Since the channel conditions are arranged
as |g1 |2 < |g2 |2 < . . . |gn |2 , the power allocation coefficients must be ordered as
p1 > p2 > . . . pn . At the receiving end, device 1 with channel h1 is farthest from
the base station which receives the weakest signal. Device 2 is the next and so on.
Device N is the nearest base station that receives the strongest signal. The received
Fig. 2 a Uplink NOMA transmission, b downlink NOMA transmission
Fig. 3 NOMA with QPSK modulation in AWGN channel
signal at the device i is given by
Yi,NOMA = UNOMA × h i + wi
√ √ √ √
Yi,NOMA = p p1 U1 + p2 U2 + · · · + pn Un × h i + wi (2)
where wi is the additive white Gaussian noise. The channel conditions are arranged
as |h1 |2 < |h2 |2 < …|hn |2 . Since the highest power is allotted to device 1, this message
dominates the received signal. Hence, decoding is performed to obtain the message.
Next, the device 2 signal is directly decoded device 1’s message, and then, perform
successive interference cancelation (SIC) to remove device 1 message. After SIC,
the device 2 message is dominating term. Then, perform direct decoding to obtain
the device 2 message. This decoding process can be extended to obtain all messages
from smart grid units (Figs. 2 and 3).
4 The Proposed Power Control Strategy
With the knowledge of power demand (Pdemand ), the solar PV power (Ppv ), and wind
turbine power (Pwind ), the net power available in the common AC bus as shown in
Fig. 4 can be calculated as
Fig. 4 Model predictive

control for power
management strategy
Pnet = Ppv + Pwind − Pdemand (3)
For energy saving considerations, this net power must be balanced through battery
energy storage system (BESS) power (PBESS ) and the utility grid power (Pgrid ). When
inverter power is negative, the battery absorbs power from renewable energy sources.
Conversely, when inverter power is positive, the battery supplies power to the load. In
discharging mode operation, a battery with the higher state of charge (SoC) supplies
more power, whereas lower SoC supplies less power. The converse is also true in
the charging process. The SoC of BESS can be obtained as
SoC(k) + PBESS (k + 1) × T s
SoC(k + 1) = (4)
CBESS
where C BESS is the capacity of the batteries. The battery power and utility grid power
can be predicted at the next control instant as
PBESS (k + 1) = u(k + 1) + PBESS (k) (5)
where the control variable u(k + 1) is the change of battery power. The required
battery power (Pnet ) is obtained by minimizing the following cost function to keep
the power balanced within microgrid as
J = | Pnet (k + 1) − PBESS (k + 1)| (6)
In general, the microgrid and utility can operate at different frequencies. The
transition from islanded to the grid-connected mode and grid-connected to islanded
mode are typically planned. To realize a smooth mode transfer, a synchronization
procedure is performed to synchronize the voltage, phase angle, and frequency of
the DG unit to the common AC bus at the utility side. The proposed control strategy
maintains the terminal voltage and frequency of the units to their nominal values.
Hence, the units deliver their nominal active and reactive power to load.
Due to simultaneous transmission, NOMA suffers from interference. As a result,

OMA performs slightly better than NOMA at low SNR as shown in Fig. 5a. However,
Fig. 5 a Capacity performance comparison of NOMA vs OMA, b A power allocation method

for multiple user single-carrier NOMA, c BER performance of multiple user NOMA with QPSK
modulation in AWGN channel; d Outage probability of uplink NOMA
NOMA outperforms OMA by offering high capacity at high SNR. As a consequence

of the number of users (i.e., smart grid devices) of single-career NOMA is increased,
the capacity of the network initially increases as shown in Fig. 5b. The drop-off
point may be regarded as the maximum limit on the number of devices that can be
admitted into the network without any performance degradation. Beyond the drop-
point capacity falls. The drop-off point can be changed by varying transmit power.
In order to enhance the network capacity without performance degradation, transmit
power should be increased. However, the fixed power allocation scheme affects the
BER performance as shown in Fig. 5c. The weakest signal has the highest bit error
rate, while the strongest signal has the lowest BER and is free from interference.
The outage probability is high for both near and far users as shown in Fig. 5d. Thus,
NOMA gives superior performance when the channel conditions between the users
are more distinct.
Besides, the performance of the proposed control strategy is analyzed in a case
study. A typical load profile consumption per day is depicted in Fig. 6a. The power
demand throughout the year is also presented in Fig. 6b, and the average load demand
for this site is approximately 4 kW. The wind and solar meteorological data for
Sundarban (India) location are considered for the case study as shown in Fig. 6c,
d. It is observed from Fig. 6g that the power generated by PV is more than the
power demand during June–September (3651-6570 h). Hence, the utility grid has
not contributed. This excess energy is utilized to charge the battery and distribute it
into the dump load as shown in Fig. 6e. It is noted that high quantities of dump energy
are stored in the battery (77.19 kW). It can be utilized for street lighting, heating,
water pumping, and refrigeration. It is clear from Fig. 6f that the depth of discharge
is maintained up to 80% and the state of charge of the battery is not less than 20%
Fig. 6 a Typical load profile consumption per day, b hourly power demand, c hourly PV power,
d hourly wind turbine power, e battery dump energy during the year, f battery energy during the
year, g utility output energy; h energy contribution of PV, WT, BESS, and utility grid
of charge level. The percentage of contribution of units over a year is depicted in the
pie chart in Fig. 6h.
6 Conclusion
A NOMA-based smart microgrid configuration and MPC-based power control

scheme are proposed for future microgrids. The performance and benefits of adopting
NOMA can support massive connectivity with high data rates and low latencies
for smart grid communication systems. Besides, the proposed power management
strategy optimizes power flow within the microgrid and between renewable energy
sources and the utility grid. Thus, the proposed smart microgrid system provides
high-quality and reliable electricity to the customers irrespective of the fluctuated
renewable energy sources.
References
1. Baldonado, Ahmad A, Rehmani MH, Tembine H, Mohammed OA, Jamalipour A (2017) IEEE
access special section editorial: optimization for emerging wireless networks: IoT, 5G, and smart
grid communication networks. IEEE Access 5:2096–2100
2. Kuzlu M, Pipattanasomporn M, Rahman S (2014) Communication network requirements for
major smart grid applications in HAN NAN and WAN. Comput Netw 67:74–88
3. Hussain M, Rasheed H (2019) Communication infrastructure for stationary and organized
distributed smart meters. In: 2019 2nd International conference on communication, computing
and digital systems (CCODE), pp 17–22
4. Rabie KM, Adebisi B, Yousif EHG, Gacanin H, Tonello AM (2017) A comparison between
orthogonal and non-orthogonal multiple access in cooperative relaying power line communica-
tion systems. IEEE Access 5:10118–10129
5. Islam SMR, Avazov N, Dobre OA, Kwak K (2017) Power-domain non-orthogonal multiple
access (NOMA) in 5G systems: potentials and challenges. IEEE Commun Surv Tutorials
19(2):721–742
6. Agarwal A, Chaurasiya R, Rai S, Jagannatham AK (2020) Outage probability analysis for
NOMA downlink and uplink communication systems with generalized fading channels. IEEE
Access 8:220461–220481
7. Ali MS, Tabassum H, Hossain E (2016) Dynamic user clustering and power allocation for uplink
and downlink non-orthogonal multiple access (NOMA) systems. IEEE Access 4:6325–6343
8. Shahab MB, Kader MF, Shin SY (2016) Simulink implementation of non-orthogonal multiple
access over awgn and rayleigh fading channels. In: 2016 International conference on smart green
technology in electrical and information systems (ICSGTEIS), pp 107–110
Power Management in DC Microgrid
B. Balaji and S. Ganesan
Abstract A microgrid can be characterized as force group of disseminated genera-

tion, burden, and energy stockpiling gadget gathered together nearby to one another.
It offers freedom to use sustainable power hotspots for green and clean climate.
This work mainly deals with the analysis, modeling and simulation of DC microgrid
with solar powered system alone, DC microgrid with battery alone, DC microgrid
with wind alone, DC microgrid with solar powered system and battery, DC micro-
grid with solar powered system and airstream & DC microgrid with solar powered
system, battery and wind. The outcomes are compared in terms of output voltage and
output power. The outcome represents the superior performance of DC microgrid
through solar powered system, battery and wind sources.
Keywords DC Microgrid · Solar powered system (PV) · Wind · Battery and

distributed generation
1 Introduction
The idea of microgrid network was first evolved by RH. Lassete in 2002, that was
an example of appropriated age covering assortment of fuel sources from fossil fuel
sources to environmentally friendly power sources (sun based, wind and biogas).
Microgrid is an organization framework interconnected from assortment of appro-
priated fuel sources into a little organization which can be worked autonomously or
associated with utility matrix. The examination of ac microgrid had been completed
yet there are as yet numerous issues that ought to be tackled identified with strength
B. Balaji (B)
Department of Electrical Engineering, Annamalai University, Chidambaram, India
Department of Electrical and Electronics Engineering, IFET College of Engineering, Villupuram,
India
S. Ganesan
Department of Electrical and Electronics Engineering, Government College of Engineering,
Salem, India
https://doi.org/10.1007/978-981-16-4943-1_20
214 B. Balaji and S. Ganesan
of activity of inverter, similar to resemble component, solidness of frequency and

power supply.
Likewise, the price of inverter was generally costly, particularly for family appli-
cation. Technique for inverter equal had been done, for example, utilization of stage
look circle (PLL) strategy and Droop Voltage. Nonetheless, another issue actually
showed up as well, like difference in recurrence and receptive force because of
progress and sort of utilized burden. It will end up being its own concern in the
microgrid framework. A few explores related with direct current had been created to
defeat issues happened in the substituting current like estimation of misfortunes in
the organization and proportion of effectiveness of rotating current and direct current.
As of late, the DML-family was presented as a better energy-defer elective than
CMOS. DML entryways use 2 distinct methods of activity, dynamic and static, to
specifically accomplish either superior or low-energy activity. Hand crafts of DML
circuits had been demonstrated to be very efficient Emadi [1]. A solitary transport
DC microgrid can address a wide scope of utilizations. Control targets of such frame-
works incorporate elite transport voltage guideline and legitimate burden dividing
between numerous DG’s under different working conditions Peng [2].
A microgrid was difficult to control because of its decreased dormancy and
expanded vulnerabilities. To defeat the difficulties of microgrid control, progressed
regulators should be created. In this, a circulated, 2 level, correspondence financial
control plot, was introduced for numerous transport miniature lattices with each trans-
port having different DG’s associated in equal. The control objective of the upper
level was to figure the voltage references for one transport subsystems Qinmin [3].
This geography, - because of a-thunderous L-C-tank-associated—among the channel
terminals of the switches of 2 inter leaved boost-cells, accomplishes 0 voltage and
0 current—replacements, all things considered, autonomously of the heap current,
with a decreased dv/dt crossways the switches Giorgio Spiazzi [4].
The interleaving method was a notable strategy functional in high force claims
to divide the yield power among 2 modules though diminishing the info current and
yield current wave. Additionally, 0 current exchanging procedure can be utilized
to diminish the exchanging misfortunes and increment influence density Bahrami
[4]. A hilter kilter forward-flyback dc–dc converter had high-force transformation
proficiency ’ηe’ over wide yield power range. To tackles issue of clear in the voltage
rectifier diodes & the issue obligation misfortune in ordinary deviated half-connect
converter, the anticipated converter utilized a voltage doubler construction with
forward inductor Lf in the subsequent juncture, rather than utilizing transformer
spillage inductance that control yields current Lee [5] (Fig. 1).
An advancement of twofold booming DC–DC converter with 0 voltage trading
and 0 current supplanting features proposed. This geography contained-switches and
capacitors cuts at the significant feature of the separation transformer. 2 switches were
worked in a corresponding manner under a PWM conspire Tahavorgar [6]. Another
transformer-less BBC with positive yield voltage was introduced by Miao-2016.
Another transformer-less BBC with straightforward design was proposed in this
examination. Juxtaposed & the routine BBC, the anticipated BBC voltage acquire was
squared junctures of the previous’ and its revenue voltage margin were affirmative [7].
Power Management in DC Microgrid 215
Fig. 1 Block diagram of-AC-DC microgrid-system
Exchanged capacitor based Z source identical DC–DC help converter geographies

were anticipated. Reason on the topological developments was moved LCD cubicle
of 4th request support converter on anterior side a control siphon cell upstream side
Veeracheary [8]. A usable and flexible family energy the executive’s framework was
proposed to create and carry out request reaction (DR) projects. These were under
the half and half age energy stockpiling framework solar powered system, and green
vehicle in the keen network (SG). Prevailing family energy the board frameworks
can’t offer its clients a decision to guarantee client comfort (UC) and not give a
practical arrangement as far as diminished carbon emission [9].
Meanwhile a DC microGrid is comprised of force converters associated from
side to side various line impedances, tuning voltage regulator gave straightforward
& instinctive transaction among clashing objectives voltage guideline & current
sharing. Profoundly adaptable circulated governor system was anticipated to accom-
plish adjusted governor between two control goals, incorporates regulation voltage
regulator and consensus based cc Renke [10].
Block diagram of DC microgrid system is exposed in Fig. 2. Various sources are

solar powered system, wind generator and battery.
Figure 3 outlines the simulink diagram of DC microgrid with battery alone. Figure 4
outlines the Input voltage and its value is 14 V. Figure 5 outlines the circuit diagram
Fig. 2 Block diagram of DC microgrid system
Fig. 3 Simulink diagram of DC microgrid with Battery alone
Fig.4 Input voltage

Fig. 5 Circuit diagram of boost converter
of boost converter with silicon carbide mosfet. Figure 6 outlines the switching pulse
S1 of boost converter and its value lies between 0 and 1 unit. Figure 7 outlines the
voltage across R-load with battery alone, and its value is 81 V. Figure 8 outlines the
Fig. 6 Switching pulse for-S1 of boost converter
Fig. 7 Voltage across R-load with battery alone
Fig. 8 Current through R-load with battery alone

Fig. 9 Output power with R-load with battery alone
current through R-load with battery alone, and its value is 0.82A. Figure 9 outlines
the output power with battery alone, and its value is 65 W.
Figure 10 outlines the simulink diagram of DC microgrid with PV alone. Figure 11
outlines the input voltage, and its value is 16 V. Figure 12 outlines the Voltage across
R-load, and its value is 86 V. Figure 13 outlines the current through R-load with PV
alone, and its value is 0.8A. Figure 14 outlines the output power with PV alone with
PV alone, and its value is 74 W.
Fig. 10 Simulink diagram DC microgrid with PV alone
Fig. 11 Input voltage

Fig.12 Voltage across R-load with solar powered system alone
Fig. 13 Current through R-load with solar powered system alone
Fig. 14 Output power with solar powered system alone
Figure 15 outlines the simulink diagram of DC microgrid with wind alone.

Figure 16 outlines the Input voltage, and its value is 15 V. Figure 17 outlines the
voltage across R-load with wind alone, and its value is 90 V. Figure 18 outlines the
current through R-load with wind alone, and its value is 0.8A. Figure 19 outlines the
output power with wind alone, and its value is 85 W.
Figure 20 outlines the simulink diagram of DC microgrid with solar powered
system and battery. Figure 21 outlines the input voltage, and its value is 16 V.
Figure 22 outlines the Voltage across R-load with solar powered system and battery,
and its value is 100 V. Figure 23 outlines the current through R-load with solar
powered system and battery, and its value is 1A. Figure 24 outlines the output power
with solar powered system and battery, and its value is 100 W. Figure 25 outlines the
simulink diagram of DC microgrid with solar powered system and wind generator.
Figure 26 outlines the input voltage, and its value is 16 V. Figure 27 outlines the
voltage across R-load with PV and wind generator, and its value is 105 V. Figure 28
outlines the current through R-load with PV and wind generator, and its value is
1.2A. Figure 29 outlines the output power with PV and wind generator, and its value
is 110 W.
comparison of output voltage and output power is given in Table 1. By using
DC microgrid with solar powered system, battery and wind, the output voltage is
Fig. 15 Simulink diagram of DC microgrid with wind alone
Fig. 17 Voltage across R-load with wind alone
Fig. 18 Current through R-load with wind alone

Fig. 19 Output power with wind alone
Fig. 20 Simulink diagram of DC microgrid with solar powered system and battery
Fig. 22 Voltage across R-load with solar powered system and battery
Fig. 23 Current through R-load with solar powered system and battery
Fig. 24 Output power with solar powered system and battery
Fig. 25 Simulink diagram of DC microgrid with solar powered system and wind generator

Fig. 27 Voltage across R-load with solar powered system and wind generator
Fig. 28 Current through R-load with solar powered system and wind generator
Fig. 29 Output power with solar powered system and wind generator
Table 1 Font sizes of

Case Output voltage (V) Output power (W)
headings. Table captions
should always be positioned PV (solar powered 86 74
above the tables system)
Battery 81 65
Wind 90 85
Solar powered system 98 97
(PV)+Battery
Solar powered system 105 110
(PV)+Wind
PV+Battery+wind 114 130
enhanced from 86 to 114 V; output power is enhanced from 74 to 130 W. Hence, the
outcome represents that the DC microgrid with solar powered system, battery and
wind is superior to DC microgrid with PV alone, DC microgrid with battery alone,
DC microgrid with wind alone, DC microgrid with PV and battery and DC microgrid
with PV and wind.
4 Conclusion
DC microgrid with PV alone, DC microgrid with battery alone, DC microgrid with

wind alone, DC microgrid with solar powered system and battery, DC microgrid
with PV and wind & DC microgrid with PV battery and wind are modeled, analyzed
and simulated. The outcomes are compared in terms of output voltage and output
power. By using DC microgrid with PV battery and wind, the output voltage is
enhanced from 86 to 114 V; output power is enhanced from 74 to 130 W. Hence, the
outcome represents that the DC microgrid with PV, battery and wind are superior
to DC microgrid with PV alone, DC microgrid with battery alone, DC microgrid
with wind alone, DC microgrid with PV and battery and DC microgrid with PV and
wind. This work mainly deals with the simulation of DC microgrid with PV, battery
and wind. The simulation of DC microgrid with PV battery and wind in closed loop
system can be done in future.
Case Output voltage (V) Output power(W)

PV (solar powered system) 86 74
Battery 81 65
Wind 90 85
PV+Battery 98 97
PV+Wind 105 110
PV+Battery+wind 114 130
References
1. Azer P, Emadi A (2020) Generalized state space average model for multi phase inter leaved
buck, boost and buck-boost DC-Dc converters:-transient steady state and switching dynamics.
IEEE Access 77735–77745https://doi.org/10.1109/ACCESS.2020.2987277
2. Peng J, Fan B, Duan J, Yang Q, Liu W (2019) Adaptive decentralized output-constrained
control of single-bus DC microgrids. IEEE/CAA J Automatic Sinica 6(2):424–432
3. Fan B, Peng J, Duan J, Yang Q, WLiu (2019) Distributed control of multiple-bus microgrid
with paralleled distributed generators. IEEE/CAA J Automatic Sinica 6(3):676–684
4. Bahrami H, Adib E, Farhangi S, Iman-Eini H, Golmohammadi R (2017) ZCS-PWM interleaved
boost converter using resonance-clamp auxiliary circuit. IET Power Electron 10(3):405–412
5. Bahrami H, Adib E, Farhangi S, Iman-Eini H, Golmohammadi R (2017) ZCS-PWM interleaved
boost converter using resonance-clamp auxiliary circuit. IET Power Electron 10(3):405–412.
https://doi.org/10.1049/iet-pel.2016.0267
6. Lee HS, Choe HJ, Ham SH, Kang B (2017) High-efficiency asymmetric forward-flyback
converter for wide output power range. IEEE Trans Power Electron 32(1):433–440
7. Tahavorgar A, Quaicoe JE (2017) A dual series-resonant DC–DC converter. IEEE Trans Power
Electron 32(5):3708–3718
8. Miao S, Wang F, Ma X (2016) A new transformerless buck–boost converter with positive output
voltage. IEEE Trans Ind Electron 63(5):2965–2975
9. Veerachary M, Kumar P (2020) Analysis-and-design of-quasi-z-source equivalent DC-DC

boost converters. IEEE Trans Ind Appl 2:1–1. https://doi.org/10.1109/TIA.2020.3021372
10. Imran A, Hafeez G, Khan I, Usman M, Shafiq Z, Qazi AB, Khalid A, Thoben K-D
(2020). Heuristic based programable controller for efficient energy management under renew-
able energy sources and energy storage system in smart grid. IEEE Access 8(2):39587–
139608https://doi.org/10.1109/-ACCESS.2020.3012735
Embedded System for Lethal Gas
Leakage Exposure and Forewarning
System Using Arm Processor
M. Nirmala and S. Durga Shree
Abstract All over the world various semiconductor manufacturing industries

dealing with the accidents that occur due to the leakage of hazardous or lethal gases
that are involved in the process of making semiconductors and accidents due to the
noise of the machineries used. As it is an emerging problem in the production of
semiconductor, a safety system is being proposed. The proposed embedded based
safety indication system would ensure that the leakage of gas such as hydrogen and
any deviation from the safety limit of sound, accidents happening all over without
any leakage of gases is being detected and an immediate alert to this leakage or
deviation is being sent to the representative of the industry. The message may be sent
to their number, displaying the alert, turning off the supply to the system and ringing
an alarm to indicate the above cases of accident. In this paper, the effectiveness of the
proposed ARM controller-based safety system is implemented using proteus virtual
system modelling is using, would ensure the absolute safety.
Keywords ARM based embedded system · Hazardous gases · Proteus virtual

platform · Global System for Mobile Communication (GSM) · Arduino IDE
1 Introduction
In general, more than 35% of gas concerned accidents happens in industries because
of gas leakage. The real concern is the spillage of hydrogen and sound related acci-
dents in industries. Various rules and regulations are executed for the gas spillage and
abnormal noise identification system. The Proposed model gives an alert message
which is fundamentally required to differentiate a Gas leakage in the commercial
premises. An alert message is sent on to the Liquid Crystal display and a sign with
M. Nirmala · S. Durga Shree (B)

Electrical and Electronics Engineering Department, Kumaraguru College of Technology,
Coimbatore 641049, India
M. Nirmala
https://doi.org/10.1007/978-981-16-4943-1_21
228 M. Nirmala and S. Durga Shree
the assist of GSM unit is skilled to articulate messages to the desired member about
the spillage of Hydrogen and shortly kill the furnish of gas [1]. The wi-fi based
carbon dioxide (CO2 ) monitoring system with GPS receiver is used to acquire statis-
tics information about the gas leakage. But challenge of this gadget is that it makes
use of web for monitoring, also it needs to have 24/7 web connection. This System
solely raises an alarm when gas is leaked, but it cannot perform any motion to forestall
or decrease any the harm [2] Fuel leakage is to installation a fuel leakage detector
at inclined locations. The advantage of the system is computerized alarming system,
which can observe liquefied petroleum fuel leakage in quite a number of locations
[3]. In [4], the GSM based an embedded machine for Gas Cylinder maintenance
where its detection section become aware of the fuel leakage and sends SMS to the
customer via GSM about the gas leakage. A wi-fi LPG leakage monitoring detects
the leakage of the LPG and signals the client the use of GSM about the leakage and
controls exhaust fan [5]. Whenever the device detects the enlarge in the attention of
the LPG leakage it straight away indicators via activating an alarm and concurrently
sending message to the specific cell phones. The fan is switched on to exhaust gaso-
line and an LPG secure valve geared up to the cylinder is closed thru alerts to keep
away from similarly leakage. The machine assures protection and prevents explosion
[6]. A microcontroller totally based gadget the place a fuel sensor (MQ6) is used in
detection of LPG leakage [7]. This unit is additionally built-in with an alarm unit, to
sound an alarm or provide a visible indication of the leakage. The sensor has exces-
sive sensitivity with rapid response time at low priced cost. If leakage is detected,
message to the unique consumer or to household member the use of mobile commu-
nity referred to as GSM is dispatched automatically [8]. It additionally measures the
weight of LPG cylinder and displayed in LCD display. A fuel volume of much less
or equal to 10 kg, it requests for the new cylinder through robotically sending textual
content message to a distributor [9]. Also, when cylinder weighs much less than or
equal to 0.5 kg, it informs the patron by way of sending a message to fill up the
cylinder [10].
Gas leakages and accidents due to noise that exceeds human audible vary are
one of the major causes at the behind of industrial accidents. The leakage of fuel is
robotically detected with the aid of the detection system. The probabilities of furnace
accident are prevented through the gadget which would have been delivered on thru
the leakage of gas [11]. The fuel sensor observes the occurrence of hydrogen gasoline
and sound sensor senses the variation of the sound, then the ARM cortex controls
the stepper motor and also flash this information as a display which strengthen
this industrial accident quitter system [12]. The proposed system is an automatic
fuel exposure, sound revealing and accident prevention system. The fuel leakage
is routinely detected with the aid of ARM system. The exhaust fan is routinely
switched ON if a leakage is detected. To force out the gas from the room, the exhaust
is used. Simultaneously sending an alert message to turn off the supply of gas. The
chances of fire accident are efficiently avoided by the system. ARM controller along
with a gas sensor to identify the presence of hydrogen gas around the surroundings
and its range. A GSM module conveys a warning message to the respective person
responsible. The ARM Microcontroller has low number of transistors and these
Embedded System for Lethal Gas Leakage … 229
transistors are cost effective. One of the most advanced form of these microcontrollers
is a cortex controller, it is mostly used in wireless communication technologies and
other embedded system due to benefits such as low power consumption, etc. ARM
Controller is an energy efficient Microcontroller. It is a high performance 32-bit
processor which offers numerous benefits to the developers. It collectively directs the
data over the web for throwing gasoline out after few milliseconds lag and continue
send messages as “Gas Leakage Detected” to mobile phone.
2 Proposed Lethal Gas Leakage Exposure

and Forewarning System
The main objectives of the proposed system are: (i) To develop a system which detects
the leakage of Hazardous gas like HYDROGEN and alert people quickly using
effective controller as a medium to prevent any kind of accidents. (ii). To prevent the
upshot of any blast in the industry by alerting the area of accident and to maintain the
safety range of sound audible to human ears. (iii) The alerting includes displaying
‘There is a leakage’, sending a message regarding the leakage (to the respective
representative of the gas unit in the industry), cutting-off the supply immediately,
giving an alarm, turning the exhaust gas shown in Fig. 1.
Fig. 1 Block diagram of the proposed lethal gas leakage exposure and forewarning system
The entire block incorporates the detection of the gasoline sensor, then there is a
block manner in which there is an Arduino microcontroller issue which acts to acquire
the cost dispatched by means of the gasoline sensor and eventually the output block
shows the output in the structure of LED lights SMS gateway. The entire block right
here is the gasoline sensor that reads the analog facts enter value, in this situation
the sensor will flip on if the Arduino microcontroller has power, the sensor will then
begin studying and sending analog records to the microcontroller and then enter the
block process. The block manner acts as the recipient of statistics that is dispatched
from the enter block and then processed to do the conditioning that was once in
the past programmed the use of the Arduino IDE software. After conditioning, the
output will be persevered to the output block. The output block consists of a led mild
as a warning of a fuel leak in the room then the fan features as a gasoline suction in
the fan room will mechanically flip on if the block manner sends a command that is
when there is a fuel leak in the room, the fan will mechanically flip off if the analog
cost is the enter block has back to regular and the closing one is the GSM SIM800l
module which features as a message sending media in the tournament of a fuel leak
in the room, sending messages will show up persistently with a prolong of 5 s if the
gasoline prerequisites in the room are nonetheless excessive intensity.
2.1 ARM Based Indication System
The circuit diagram for the proposed ARM based indication system is shown in
Fig. 2.
The proposed device consists of the GSM module, LCD, relay circuit and sensors
as hydrogen sensor and sound sensor which affords greater security to the enterprise
Fig. 2 ARM based indication system

and hence to will increase the effectivity of the system. The gasoline detector has quite
a few advantages, some are convenient installation, low cost and honestly convenient
to use. Another essential gain of the detector is that even if fuel leaks happens when
no people is in the inside of building, GSM module could ship on the spot messages
to stakeholders related to the gasoline leakage and for this reason it helps in warding
off the accidents at the earliest. The proposed gadget is to make certain the protection
restriction and discover the quantity of hydrogen fuel in the ecosystem and the stage
of noise in the environment (human audible range) and to realize the unit (in the
industry) in which a blast has come about except any leakage in gasoline.
The power supply required to run the prototype is very minimum. The selected
components in a way that consumes lesser power than those present in the existing
system. Figure 3 shows the work flowchart of the proposed system. The GSM module
is controlled by the ARM microcontroller. The module indicates an ALERT, when the
gas is leaked. Here alerting system is sending an alert message saying “LEAKAGE
ALERT” to the respective representative of the respective firm which is not a part
of any existing system. The GSM module is provided with a SIM CARD with the
registered number of the respective representative to receive an alert message. A
relay is generally connected here to the hydrogen gas sensor and sound sensor. The
function of the relay is to turn the system into an open system from a closed system
and abruptly turn OFF of the supply. When the system becomes a closed one the
leakage of the respective gas also stops as the system has no supply. The relay
receives an interrupt signal to turn OFF. Safety is assured when a relay is included
as the preventing the leakage of gas would also make a root to get rid of an accident.
Figure 4 shows the simulation circuit of proposed system. The simulation circuit
consists of an M8 hydrogen gas sensor, a sound sensor, and the power supply on
the input side. The power supply is the available AC source which being stepped
down from 230 to 11 V using an step down transformer which is further fed into a
rectifier which converts the AC voltage to pulsating DC and then into pure DC using a
capacitor. Certain level of gas and sound has be coded to the ARM in which whenever
the limit exceeds the coded level the ARM would give signal to the components in the
output side. The GSM module to send an alert message to the particular representative
of the particular unit, the LCD to display ‘ABNORMAL ALERT’, the relay to make
the system an open circuit and thus cut off the supply, the exhaust fan goes on in
order to let the gas into the atmosphere and finally the alarm goes on to indicate that
there is a gas leakage or there is a noise disturbance in the industry.
Figure 5 shows that the simulation results of ARM based system under normal
operating condition. Figures 6 and 7 illustrates that output of hydrogen sensor and
sound sensor under abnormal conditions.
At certain level of hydrogen gas (500) and sound (3000 dB) has be coded to the
ARM in which whenever the limit exceeds the coded level the ARM would give signal
Fig. 3 Flowchart of ARM based indication system
to the components in the output side. The GSM module to send an alert message to
the particular representative of the particular unit, the LCD to display ‘ABNORMAL
ALERT’, the relay to make the system an open circuit and thus cut off the supply,
the exhaust fan goes on in order to let the gas into the atmosphere and finally the
alarm goes on to indicate that there is a gas leakage or there is a noise disturbance in
the industry. And if the condition is normal it displays NORMAL CONDITION.
Fig. 4 Simulation circuit of ARM based indication system
Fig. 5 Simulation result under normal operating conditions (No leakage)
4 Conclusion
As it is known that the leakage of gases in industries and it’s aftermath is an emerging
problem which has even caused many life, the future scope of the proposed system
would be that it would reduce the number of accidents happening all over world due
Fig. 6 Hydrogen sensor output under abnormal condition
Fig. 7 Sound sensor output under abnormal condition
to the leakage of gas in the industries and accidents that happen due to the sound that
exceeds the safety level. Thus, as a result of the proposed system is implemented
using proteus virtual platform and safety can be ensured in the future to the entire
semiconductor company (Einfochips) for which the system is being proposed. More
of modifications can be made in the future to the prototype as a retrofittable device to
the gas tubes in the industry which use hydrogen and in industries which machines
used produce more noise.
References
1. Bhamare P, Dalvi S, Bhamare M, Dube D, Bhonsale M (2016) A survey on GasBo for LPG
gas detection using mobile. Int J Adv Res Dev 4:27–32
2. Falohun AS, Oke AO, Abolaji BM, Oladejo OE (2016) Dangerous gas detection using an
integrated circuit and MQ9. Int J Innov Res Eng Manag 4:197–200
3. Katole KR, Bagade V, Bagade B, Soni A, Kamde H (2016) Hazardous gas detection using
ARDUINO. Int J Innov Res Eng Manage 4:597–600
4. Kasar MS, Dhaygude R, Godse S, Gurgule S (2016) Automatic LPG gas booking and detection
system. Int J Innov Res Eng Manag 4:297–300
5. Amsaveni M, Anurupa A, Preetha RSA, Malarvizhi C, Gunasekaran M (2015) GSM based
LPG leakage detection and controlling system. J VLSI Des Sig Process 3:55–65
6. Keluskar H, Chavan P, Sonalikudale, Salunke GD, Pawar S (2015) GSM based home safety.
Int J Innov Res Eng Manag 4:68–71
7. Soundarya T, Anchitaalagammai JV, Deepa Priya G, Karthick Kumar SS (2014) Cylinder LPG
gas leakage detection for home safety. IOSR J Electron Commun Eng (IOSR-JECE) 9:53–58
8. Prabhu AD, Pathak AD (2017) Gas leak detector using Arduino UNO microcontroller. Int J
Res Appl Sci Eng Technol (IJRASET) 5:45–48
9. Putra MF, Kridalaksana AH, Arifin Z (2017) Designing an LPG gas leak detector with
MQ6 sensor microcontroller-based via android smartphone as an information media. Jurnal
Informatika Mulawarman 12:71–75
10. Yang M, Zhao W, Xu W, Feng Y, Zhao Z, Chen X, Lei K (2019) Multitask learning for
cross-domain image captioning. IEEE Trans Multimedia 21:82–87
11. Hidayat I (2018) Gas leak system using MQ-6 based on wireless sensor network. 17:355–364
12. Parwati CI, Suseno HP, Fatkhiyah E (2015) Modeling telemetry system “Multi-Node” early
warning ammonia gas leak based SMS gateway. Jurnal Teknologi 8:151–160
A Trasformerless Buck-Boost Converter
as Maximum Power Point Tracker
for Battery Charging
C. Balaji, O. Hemakesavulu, A. Dominic Savio, B. Vinothkumar, S. Sakthi,

and P. Sivaperumal
Abstract DC-DC converters are utilized to interface the Photovoltaic source to the
battery for charging purposes. A non-isolated (NI) DC-DC buck-boost converter has
been employed as a maximum power point tracking (MPPT) system to derive the
photovoltaic power present. The MPPT control is utilized to extricate the greatest
power available from the photovoltaic source. The MPPT gets the solar irradiation and
solar cell temperature as contributions to decide the optimized duty cycle to acquire
maximum power. In this work, a non-isolated buck-boost converter is utilized as
MPP tracker for battery charging applications. The performance of the converter is
simulated and validated through the experimental results.
Keywords PV source · Buck-boost converter · Maximum power point tracker ·

Battery charging
1 Introduction
Solar energy has shown promising results and is economic. However, PV cell/module
shows non-linear I-V and P-V characteristics which rely on temperature and solar
insolation [1]. The stochastic nature of PV system has energy conversion problems,
which result lower efficiency of approx. 18% [2]. To increase efficiency and over-
coming the problems in energy conversion, MPP control is used to drive the PV
C. Balaji (B) · A. Dominic Savio · B. Vinothkumar

School of Electrical Engineering, SRM Institute of Science and Technology (Formerly Known As
SRM University), Kattankulathur Campus, Chennai 603203, India
B. Vinothkumar
O. Hemakesavulu
Department of EEE, AITS, Rajampet, AndhraPradesh, India
S. Sakthi
Department of EEE, Jeppiaar Engineering College, Chennai, India
P. Sivaperumal
Department of EEE-Marine, AMET University, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_22
238 C. Balaji et al.
modules at their maximum power point [3]. As in [4], a DC-DC power converter acts
as MPPT (maximum power point tracker) that optimizes the power flow between
PV system and load. Generally, Batteries are connected to the converter to maintain
the required charge voltage for the batteries. A charge controller helps in achieving
higher efficiency in PV battery charging systems [5].
In most of the commercial PV system fed battery applications, the MPP fixed
above the charging voltage of batteries [6]. A buck conversion is needed in this but
with high temperature and low insolation condition, the MPP goes below the required
charging voltage which requires a boost operation [7]. As in [6], to achieve a steady
voltage from an input supply which is lower or greater than the yield especially in
MPPT application, a buck-boost converter is required to harvest maximum possible
power from solar modules at all times irrespective of the type of load being used. In
battery charging applications, if a battery is directly attached to a solar-based module
with no converter, it will harm the battery because of the variances in the voltage
provided to the battery. Many converters are reported in literature for the application
of buck-boost converters as a MPP tracker [8, 9].
This paper presents a MPP tracker realized by a transformerless buck-boost
converter having positive output voltage. Many algorithms are reported in the liter-
ature for tracking MPP. Simple and a popular perturb and observe (P&O) procedure
is adopted [10]. Experiment analysis is done by implementing this converter to a
standalone PV system.
2 The Photovoltaic Power System
PV power conversion system converts energy from the sun into the electricity by
solar cells and supplies power to the load in demand. The block diagram in Fig. 1.
illustrates the components of a typical PV conversion system feeding a DC load. The
components include a PV panel (solar array), DC-DC power electronic converter and
Fig. 1 Components of a typical PV conversion system

A Trasformerless Buck-Boost Converter … 239
a DC load. Additionally, to extract maximum available power, an MPPT controller

is used.
2.1 P-V and V-I Characteristics of the PV Cell
The attributes of a PV cell can be determined from P-V and V-I plots. The voltage
Vs current plot is shown in Fig. 2a. Current corresponding to Vmax is noted. This
current is referred to as the current at MPP. A typical P-V plot is shown in Fig. 2b.
For various voltages at different temperatures and irradiance. The MPP is located
as a point at which the voltage increases and power decreases. Since the P-V char-
acteristics depends on temperature and irradiance, the MPP must be tracked for
changing temperature and irradiance to harvest maximum power from PV cell. The
attributed curve of PV cell in Fig. 2c. It shows two points; short circuit current I sc
and open circuit voltage V oc . The point where I mp and V mp meet is the maximum
power point, which corresponds to maximum power available from the PV cell. The
greatest possible power can be transmitted to the load if the load line crosses this
instantaneous point exactly.
The value of this load resistance can be determined using (1).
Vmp
Rmp = (1)
Imp
Fig. 2 a I-V characteristics, b P-V characteristics of PV cell, c V-I and P-V characteristics of a PV
cell
2.2 Maximum Power Point
The maximum power point tracking (MPPT) guarantees the extraction of the peak
possible power from the solar arrangement and transmission to the load [10]. So
as to transmit most of the power, a DC-DC power converter is employed between
PV module and the load. The load impedance as observed by the supply source
is balanced by controlling the duty cycle and coordinated at the instance of the
maximum power with the source.
3 Buck-Boost Converter
The simple block diagram in Fig. 3. shows the battery charging using PV panel.
The circuit of NI buck-boost converter [11] is shown in Fig. 4a. It comprises of two
MOSFET switches S1 and S2 , pair of diodes D1 and D2 , two capacitors C1 and C2 ,
Fig. 3 Block diagram of PV

fed battery charging
Fig. 4 a Buck-Boost converter, b equivalent circuit of state 1, c state 2

two ferrite core inductors L1 and L2 and a load. MOSFET switching devices S1 and
S2 are operated synchronously.
The PV is a source and a battery is a load. The converter can be operate in buck
and boost mode. The NI buck-boost converter works in two states.
State 1: In this state, the MOSFET switching devices S1 and S2 are closed and
the pair of diodes D1 and D2 are reverse biased. The wound inductors L1 and L2 are
magnetically charged and the parallel plate capacitors C1 and C0 are allowed to drain.
The input voltage Vpv magnetizes the inductor L1 . Also, Vpv and VC1 magnetizes the
inductor L2 .The polarity and current directions in this state are marked in Fig. 4b.
The expressions are given in (2)
di L1 vin di L2 vin + vC1 dvC1 i L dv0 v0

= , = , =− 2, =− (2)
dt L1 dt L2 dt C1 dt C0 R
State 2: In this state, the MOSFET switching devices S1 and S2 are opened and
the pair of diodes D1 and D2 conduct in forward bias. The inductors L1 and L2
are magnetized and the parallel capacitors C1 and C0 are dissipated. The capacitor
C1 is powered by the energy stored in industry L1 through D1 . The parallel plate
capacitor C0 and the load are supplied by the energy stored in inductor L2 . The
current directions and polarity are marked in Fig. 4c. The expressions are given in
(3)
di L1 vC di L2 vC + v0 dvC1 i L + i L2 dv0 iL v0
=− 1, =− 1 , =− 1 , = 2 − (3)
dt L1 dt L2 dt C1 dt C0 C0 R
3.1 Time Domain Waveforms
It is taken into an assumption that the equipment employed are conventional, capaci-
tors are adequately bigger to keep up the consistent voltage across it and the converter
works in steady and stable state for ease of analysis. For the suggested NI buck-boost
converter working in continuous conduction mode (CCM), time-domain waveforms
are presented in Fig. 5a.
The voltage boost of the suggested converter is derived and presented in (4).
2
Vo D
= (4)
Vpv 1− D
It is obvious from (12) that when duty cycle D > 0.5 the suggested converter works
in boost mode. Else, it operates in buck mode. The converter can also be employed to
regulate the output voltage as it has higher gain. The comparison of voltage gains of
Fig. 5 a Characteristic
waveforms under steady
state, b comparison of
voltage gains
(a) (b)
buck-boost converter in this work and the traditional buck-boost converter are shown
in Fig. 5b.
3.2 Modelling of the Converter
The model of the converter is obtained by solving the state equations obtained from
analyzing equivalent circuits of state 1 and state 2. The model is presented in (5).
⎡ d−1
⎤ ⎡ d
⎤ ⎡ ⎤
0 0 L1
0 L1 il1
⎢ 0 2d−1 d−1⎥ ⎢ d ⎥ ⎢ il2 ⎥
⎢ 0 ⎥ ⎢ ⎥
A = ⎢ 1−d L2 L2
⎥, B = ⎢ L2⎥, X = ⎢ ⎥
⎣ C1 ⎦,
⎣ C1 1−2d
C1
0 0 ⎦ ⎣ 0⎦
1−d −1 C2
0 C0
0 C0 R
0
D = [0], C = 0 d − 1 0 2d + 1 (5)
3.3 Design of Inductors and Capacitors
The capacitor voltage ripples and inductor current ripples are calculated using
Charge-Sec Balance and Volt-Sec balance respectively.
VPV VPV
i L 1 = d1 TS , L 1 = d1
L1 i L 1 × f s
VPV + VC1 d1 2 − d1
i L2 = d1 TS , VC1 = VPV , L 2 = VPV
L2 1 − d1 (1 − d1 )i L 2 × f s
iL V0
VC1 = 2 d1 TS , C1 = d1 and
C1 RVC1 f s (1 − d1 )
V0 V0
VC0 = d1 T s, C0 = d1 (6)
RC0 RVC0 f s
Table 1 Buck-boost
Inductor (L1 ) 1.3 mH Capacitor (C1 ) 7.5 pF
converter parameters
Inductor (L2 ) 3.3 mH Switching frequency 20 kHz
Capacitor (C0 ) 18.75 pF Load 40
If the inductor ripple current, the input and output voltages V in and V 0 respectively,
the duty ratio D, the Load Resistance R, and the switching frequency f s are familiar,
the inductance and capacitor values can be calculated using (6). The designed values
are shown in Table 1.
4 Simulation
The PV system with proposed DC-DC buck-boost Converter as Maximum Power

Point Tracker is shown in Fig. 6. The converter is designed for the maximum output
power of 100 W.
The converter is simulated using MATLAB. The voltage of PV panel at MPP is
set at 30 V, and the module voltage is fixed at 24 V. The load resistance of 40 is
used. The circuit is simulated with different duty cycles. Waveforms of voltage in
boost operation is shown in Fig. 7. Variation of output voltages from a converter with
a fixed input at various duty cycles are displayed in Table 2.
A 24 V and 7 Ah lead-acid battery has been employed as a load in buck operation.
The initial charge of the battery is fixed at 50%. The state of charge has been increased
Fig. 6 PV system with proposed MPP tracker and battery

Fig. 7 Waveforms of voltage in boost mode
Table 2 Variation of output

Duty cycle Input voltage Output Output
voltages from a converter
(%age) (V) voltage (V) current (A)
with a fixed input at various
duty cycles 40 24.9 18.9 0.153
45 24.6 21.8 0.1745
50 24.7 24.85 0.199
55 24.3 32.2 0.258
60 24.4 41.9 0.335
Fig. 8 a State of charge of 24 V battery, b state of battery charge of 24 V battery
from 50 to 50.1 in 3 s of simulation time. The state of charge waveform is displayed

in Fig. 8a.
A 24 V and 7 Ah lead-acid battery has been utilized as a load during boost mode.
The initial charge of the battery is set at 50%. The state of charge has been increased
from 50 to 50.08 in 3 s simulation time. The state of battery charge is shown Fig. 8b.
5 Experiment Verification
A laboratory prototype with MPPT system and a battery as a load were built to
check the operational performance of the suggested converter as shown in Fig. 9.
The switching signals are generated using FPGA SPARTAN6 controller board. A
hall-effect sensor (ACS-712) and voltage divider circuit are used to sense current and
voltage respectively. The parameters of the proposed prototype are given in Table 1.
Tests have been done to show the vitality of the proposed buck-boost converter. A
75 W PV module is utilized at standard test conditions with insolation of 1000 W/m2
Fig. 9 Digital photograph of

the experimental setup
a b c
Fig. 10 a Battery voltage under charging condition, b PV voltage (Ch 1) 20 V/div, PV current (Ch
2) 10A/div, PV power (Ch M) 100 W/div, T = 2 s/div, c PV voltage (Ch 1) 20 V/div, output voltage
(Ch 2) 10 V/div, T = 1 s/div
and a temperature of 25 °C. At the point when a battery is directly tied to the PV
module, it is seen that the power extracted from the module is 52 W. However, a
system with the converter as MPP tracker produces 65 W.
The charging voltage that appears across the battery terminals is shown in Fig. 10a.
It shows that the proposed converter serves as a MPP tracker and charges the battery
effectively. The battery is charged using 24 V supply from PV panel.
The PV voltage, current and power when the PV module generates 65 W are
shown in Fig. 10b, c. It can be seen that the MPP tracker continuously tracking and
the PV power is reaching maximum value i.e., 65 W. The experimental test results of
the suggested buck boost converter as a MPP Tracker prove that it could improvise
the energy transmission efficiency effectively.
6 Conclusion
The PV fed battery charging is a developing area of research in the recent years in
the field of photovoltaic based energy generation technologies. This work utilized a
Non-inverting Buck-Boost topology between PV and the battery. The circuit config-
uration and operation of converter has been explained for which theoretical investi-
gation, simulation, and tests were conducted. From the experimental results, it can be
concluded that the converter is a good candidate for PV system fed battery charging
applications. It works effectively as MPP Tracker.
References
1. Chandrasekar B, Nallaperumal C, Padmanaban S, Bhaskar MS, Holm-Nielsen JB, Leonowicz

Z, Masebinu SO (2020) Non-isolated high-gain triple port DC–DC Buck-Boost converter with
positive output voltage for photovoltaic applications. IEEE Access 8:113649–113666
2. Chandrasekar B, Nallaperumal C, Dash SS (2019) A nonisolated three-port DC–DC converter
with continuous input and output currents based on Cuk topology for PV/fuel cell applications.
Electronics 8(2):214
3. Sridhar R, Jeevananthan S, Dash SS, Selvan NT (2014) Unified MPPT controller for partially
shaded panels in a photovoltaic array. Int J Autom Comput 11(5):536–542
4. Mahapatro SK (2013) Maximum power point tracking (MPPT) Of solar cell using buck-boost
converter. Int J Eng Res Technol 2(5):1810–1821
5. Savio DA, Juliet VA, Chokkalingam B, Padmanaban S, Holm-Nielsen JB, Blaabjerg F (2019)
Photovoltaic integrated hybrid microgrid structured electric vehicle charging station and its
energy management approach. Energies 12(1):168
6. Sujitha N, Krithiga S (2017) RES based EV battery charging system: a review. Renew Sustain
Energ Rev 75:978–988
7. Chandrasekar B, Chellammal N, Nallamothu B (2019) Non-isolated unidirectional three-port
Cuk-Cuk converter for fuel cell/solar PV systems. J Power Electron 19(5):1278–1288
8. He X, Zhang X, Liu H, Zhang H (2015) Application of an integrated transformerless buck-
boost converter in photovoltaic MPPT systems. In: 2015 IEEE 11th international conference
on power electronics and drive systems. IEEE, pp 692–695
9. Vivek P, Ayshwarya R, Amali SJ, Sree AN (2016) A novel approach on MPPT algorithm for
solar panel using buck boost converter. In: 2016 international conference on energy efficient
technologies for sustainability (ICEETS). IEEE, pp 396–399
10. Blange R, Mahanta C, Gogoi AK (2015) MPPT of solar photovoltaic cell using perturb &
observe and fuzzy logic controller algorithm for buck-boost DC-DC converter. In: 2015 inter-
national conference on energy, power and environment: towards sustainable growth (ICEPE).
IEEE, pp 1–5
11. Miao S, Wang F, Ma X (2016) A new transformerless Buck-Boost converter with positive
output voltage. IEEE Trans Ind Electron 63(5). In some cases, it is the contact volume editor
that checks all the pdfs. In such cases, the authors are not involved in the checking phase
A Feed-Forward Neural Network Based
MPPT Controller for PEMFC System
with Ultra High Step Up Converter
B. Karthikeyan, D. Karthikeyan, V. P. Arumbu, K. Sundararaju,

R. Palanisamy, and P. Divya
Abstract In the proposed work, a feed forward neural network (FFNN) based MPPT
controller is implemented in which back-propagation algorithm is used to extract
maximum power and reduce peak overshoot from 1.26 kW PEMFC. To step up the
PEMFC voltage, an Ultra High Step Up converter (UPHC) is designed and its steady
state analysis is carried out. The duty cycle of UPHC is controlled using FFNN
MPPT controller to extract optimum power from fuel cell whose result is compared
with conventional Fuzzy Logic controller. A simulation work has been carried out
using MATLAB/Simulink environment and discussed the results.
Keywords Feed-forward Neural Network · MPPT · PEMFC · Step up converter ·

Fuzzy logic controller
1 Introduction
Fossil fuel contribution to the energy economy for fast decades is remarkable but
it is going to be exhausted in near future due to increasing energy demand. Hence
the researchers are focusing on green energy which is pollution free, free of cost,
long existence etc. Among the renewable energy series, fuel cell is fascinating tech-
nology makes the researchers to look into it. There is a wide variety of fuel cell,
in that Polymer Exchange Membrane Fuel Cell is differing because it’s operating
parameters. Some of the notable unique operating parameters are low temperature
operation, efficient, can be started quickly, noiseless operation, harmless by-product.
B. Karthikeyan (B)
Department of Electrical and Electronics Engineering, K. Ramakrishnan College of Technology,
Trichy, India
D. Karthikeyan · V. P. Arumbu · R. Palanisamy · P. Divya
Chennai, India
K. Sundararaju
Department of Electrical and Electronics Engineering, M. Kumarasamy College of Engineering,
Karur, India
https://doi.org/10.1007/978-981-16-4943-1_23
248 B. Karthikeyan et al.
Even though it has lot attractive features, there will be negative side of PEMFC which
makes not suitable for electric vehicle application [1, 2].
The voltage and current relationship for PEMFC is non-linear in V-I plane. There
will be a point in V-I characteristics at which maximum output power is available. To
make it suitable for any practical application, it is required to incorporate a controller
that makes maximum output power at any operating condition. Hence there is need
for Maximum Power Point Tracking Controller (MPPT) to ensure optimum output
power to load. In literate various MPPT controllers were proposed and discussed the
performance of controller, some of noticeable controllers are discussed here [3, 4].
Incremental conductance algorithm using neural network was proposed [5]. A
7 kW PEMFC feeding resistive load through step up converter was studied using
MATLAB/Simulink platform. It was proved that the proposed controller performed
better in terms of fast convergence, reduced peak overshoot and steady state operation
compared to fixed step size IC, variable step size controllers without using neural
network. Extended sliding mode controller was proposed in extracting maximum
power from 200 W PEMFC during dynamic operating condition [6]. The improve-
ments listed in this study compared to conventional controllers are increased tracking
speed, increased boost converter efficiency. To overcome electrical non linearity of
PEMFC, a dedicated power converter for each fuel cell in an array was proposed [7].
It is a different approach because maximum fuel operating point was implemented
instead of maximum power point strategy. There were some improvements done like
20% more power could be extracted than conventional method.
A high voltage gain boost converter along with RBFN controller was proposed
for electric vehicle application [8]. In this analysis, temperature was considered
to be time varying operating parameter. During the dynamic operating condition,
the proposed controller was able to reduce input current ripple content, reduced
voltage stress across the active switches and track the optimum power effectively than
conventional fuzzy based controller. To improve both static and dynamic performance
of the converter, a variable step size IC controller was proposed in which varying
operating parameters are hydrogen partial pressure and temperature [9]. In both the
cases, the effectiveness of proposed controller was better in terms of reduced peak
overshoot, fast response time, reduced ripple content.
A genetic algorithm based optimum algorithm was proposed to extract maximum
power from PEMFC during variation of hydrogen pressure, oxygen pressure and
temperature [10]. The proposed controller could achieve better accuracy, fast
response and less fluctuation of power under different operating condition. Sun flower
optimization algorithm was proposed to reduce sum of squared error between calcu-
lated and actual output voltage [11]. The result obtained is compared with conven-
tional methods for two practical systems of Nedstack PS6 PEMFC and Horizon
500 W PEMFC in which it was proved that the proposed system gives good conver-
gence, accurate and precision. Water cycle algorithm was proposed to tune PID
controller which ensures maximum power extraction from PEMFC [12].
A Feed-Forward Neural Network Based MPPT … 249
Fig. 1 PEM fuel cell Unburnt fuel Fuel(H2)
Anode e-
I e-
Electrolyte
RL
Cathode
e-
byproduct(Hot water) Oxygen(O2)
2 PEM Fuel Cell
Fuel cell is the recent technology in which chemical energy is directly converted into
electrical energy without polluting the environment. Its features like quick start up,
noise free operation, low operating temperature and high efficiency are fascinating
the researchers to use it for electric vehicle application as well as stationary power
production. Among the various types of fuel cells, Proton Exchange Membrane
Fuel Cell (PEMFC) is abrupt choice for electric vehicle application because of its
temperature about 60–90 °C. The platinum catalyst makes the fast reaction rate to
regulate the constant output voltage. As shown in Fig. 1, Hydrogen (H2 ) rich fuel is
supplied at anode to oxidize the hydrogen ions. On the other hand, oxygen (O2 ) rich
atmospheric air is supplied at cathode to reduction reaction take place. At the end,
heat and hot water are obtained as by-product which can re-circulate after cooling
process.
3 Ultra High Step up Converter
The proposed converter consists of three active switches S 1 , S 2 , S 3 in which S 1 , S 2 are

triggered simultaneously and switch S 3 is triggered asymmetrically. Two inductors
L 1 , L 2 are placed in series with each Switch S 1 , S 2 respectively in parallel connected
arms. Diodes D1 , D2 are connected in parallel arms as shown in Fig. 2 which avoids
current reversal during the main switches are turned off. Capacitor C 1 is connected
in parallel with load to make continuous load current.
S1
L1
Fuel cell
D1 D2
C1 RL
S2 L2 S3
Fig. 2 Ultra high step up converter
3.1 Mode of Operation
During the mode 1, both the switches S 1 , S 2 are turned on and inductor L 1 , L 2 store
energy. The energy stored in capacitor C 1 at previous mode supply the load as shown
in Fig. 3.
During mode 2, both switches S 1 and S 2 are turned off and triggering pulse is
applied to the switch S 3 . The energy stored in inductor L 1 , L 2 is released through
Switch S 3 and diode D1 as shown in Fig. 4. The load is supplied from capacitor C 1
which is charged previously.
During this mode, all switches S 1 , S 2 , S 3 are turned off and energy stored
in inductor L 1 , L 2 are released to load which makes high output voltage. The
corresponding current path is shown in Fig. 5.
S1
L1
Fuel cell
D1 D2
C1 RL
S2 L2 S3
RL
Fig. 3 Mode 1
S1
L1
Fuel cell
D1 D2
C1 RL
S2 L2 S3
Fig. 4 Mode 2
S1
L1
Fuel cell
D1 D2
C1 RL
S2 L2 S3
Fig. 5 Mode 3
3.2 Voltage Gain
The conventional boost converter gain is
M = 1/(1 − D) (1)
The proposed converter voltage conversion gain can be written as
M = (1 + D1 )/D2 (2)
The voltage gain of proposed converter is compared with conventional step up

converter as shown in Fig. 6 by changing the values of duty cycle d 1 and d 2 is kept
constant. Both the duty cycle can be varied according to the conversion requirement.
Fig. 6 Voltage gain comparison
It is obvious that the proposed UPHC converter provides high voltage gain compared
to conventional converter.
4 Feed-Forward Neural Network Controller
To extract the maximum power output from UPHC converter and reduce peak over-
shoot at the time of starting, a feed forward neural network as shown in Fig. 7 based
Hidden layer
Input layer Output layer
d1
VFC
IFC d2
Fig. 7 Feed-forward neural network

MPPT controller is implemented in which voltage from fuel cell (V FC ) and current
from the fuel cell (I FC ) are fed as input in the input layer. In the input layer linear
activation function is used to map the input which is given to next layer nodes. The
weight in the hidden layer is found by applying tan sigmoid membership function. At
the output layer linear activation function is used to calculate duty cycle d 1 and d 2 .
The trained network is tested using back propagation algorithm by updating weight
after each epochs.
The simulation work has been carried out using MATLAB/Simulink platform in
which fuel cell parameters are pre-specified values. The characteristics of 1.26 kW
PEMFC is analyzed. The inference made from analysis is the fuel cell is operated at
24.23 V, 52 A to give maximum power output of 1259.6 W.
The proposed Ultra-High Step up Converter fed from 1.26 kW circuit is
constructed by using the hardware components. The operating frequency of the
converter is kept at 50 kHz while the duty cycle is maintained at d 1 = 0.5 and
d 2 = 0.35. The average output voltage obtained at steady state is shown in Fig. 8.
The fuel cell output voltage is influenced by the operating temperature and variation
of temperature is considered as 328 K for (0–0.6 s), 361 K for (0.6–1.2 s) and 338 K
for (1.2–2.4 s) which corresponds to variation of output voltage, current and power
as shown in Fig. 9. While the operating temperature is varied, the output power of
UPHC is also varied in which peak overshoot is present which is shown in Fig. 10.
The peak overshoot of DC link power is reduced by 21.5% with respect to steady state
power 565 W corresponding maximum power reached during first time is 687 W
shown in Fig. 11. From the Fig. 11, it is inferred that Feed Forward Neural Network
outperforms the conventional Fuzzy based controller.
Fig. 8 Output voltage of UPHC converter

Fig. 9 DC link voltage
Fig. 10 DC link power
Fig. 11 DC link power output comparison

6 Conclusion
In the proposed work, Feed-forward Neural Network based MPPT controller was
implemented to extract maximum power from 1.26 kW PEMFC. The Ultra High
Step Up converter (UHSC) was designed and steady state analysis of the step up
converter was done. The switching pulses of UPHC were generated by the FFNN
based MPPT controller thereby optimum power extraction was ensured. A simula-
tion work was carried out using MATLAB/Simulink environment also implemented
hardware prototype and the following highlights were observed.
• A 1 kW UPHC converter was designed and analyzed the steady state operation.
The proposed converter provides better performance than conventional converter.
• Feed forward Neural Network based MPPT controller performs well than
conventional Fuzzy Logic controller.
• Peak overshoot in DC link power was reduced by 21.5% with reference to steady
state power; average time to reach optimum power is reduced by 20.5%.
References
1. Sammes N (2006) Fuel cell technology, reaching towards commercialization. Springer, United
Kingdom
2. Larminie J, Dicks A (2003) Fuel systems explained. J. Wiley, pp 207–225
3. Spiegel C (2011) PEM fuel cell modeling and simulation using MATLAB. Elsevier
4. İnci M, Ömer T (2019) Review of fuel cells to grid interface: configurations, technical
challenges and trends. J Cleaner Prod 213:1353–1370
5. Reddy, Jyotheeswara K, Sudhakar N (2018) Energy sources and multi-input DC-DC converters
used in hybrid electric vehicle applications—a review. Int J Hydrogen Energ 43:17387–17408
6. Harrag A, Bahri H (2017) Novel neural network IC-based variable step size fuel cell MPPT
controller: performance, efficiency and lifetime improvement. Int J Hydrogen Energ 42:3549–
3563
7. Wang MH, Huang ML, Jiang WJ, Liou KJ (2016) Maximum power point tracking control
method for proton exchange membrane fuel cell. IET Renew Power Gener 10:908–915
8. Somaiah B, Agarwal V (2016) Distributed maximum power extraction from fuel cell stack
arrays using dedicated power converters in series and parallel configuration. IEEE Trans Energ
Convers 31:1442–1451
9. Reddy KJ, Sudhakar N (2018) High voltage gain interleaved boost converter with neural
network based MPPT controller for fuel cell based electric vehicle applications. IEEE Access
6:3899–3908
10. Harrag A, Messalti S (2017) Variable step size IC MPPT controller for PEMFC power system
improving static and dynamic performances. Fuel Cells 17:816–824
11. Ahmadi S, Abdi Sh, Kakavand M (2017) Maximum power point tracking of a proton exchange
membrane fuel cell system using PSO-PID controller. Int J Hydrogen Energ 42:20430–20443
12. Harrag A, Messalti S (2018) How fuzzy logic can improve PEM fuel cell MPPT performances?
Int J Hydrogen Energ 43:537–550
Predictive Maintenance of Industrial
Equipment’s Using IOT
S. Nithya, K. Vijayalakshmi, and M. Parimala Devi
Abstract Internet of Things (IOT) is swiftly increasing technology nowadays global

in internet world. IOT is mixture of communication system and embedded device
that is used to attach hardware devices to the community or net. IOT is used for
transmission and reception of information. These structures are used to check the
processes in industries and its applications were through implementing industry
standard protocols using IOT. In this machine small scale commercial packages like
liquid level control, power monitoring and so on can be analyzed wirelessly via Wi-Fi
gadgets, mobiles and laptops. The principle purpose is to epitomize the importance
of IOT as a way to survey small scale commercial packages.
Keywords IoT · PLC · Wi-Fi · GSM · Sensors
1 Introduction
Nowadays, business tracking has important function location to monitor and manip-
ulate the industrial applications or appliances and apparatuses. Business tracking is
exercised for recognising the enterprising of heavy industrial appliances or appara-
tuses. Commercial tracking is used to perform quick processing, decrease energy
utilization and usage, to enhance quality, lessen highly-priced systems and interna-
tional control of enterprise [1, 2]. Many strategies have been conducted and performed
which are to be had to reveal and manipulate industrial approaches. Nowadays, “net
of things” is the most favourable technique for commercial process tracking [3].
S. Nithya (B) · K. Vijayalakshmi

Ramapuram, Chennai, India
K. Vijayalakshmi
M. Parimala Devi
Department of Electronics and Communication Engineering, Velalar College of Engineering and
Technology, Erode, India
https://doi.org/10.1007/978-981-16-4943-1_24
258 S. Nithya et al.
IOT is aggregate of ingrained machine and verbal exchange gadget wherein business
equipments are related to internet with the assist of wireless sensor network [4] and
gadgets or commercial utility can monitor and manipulate via mobiles and laptops.
Forms of items, sensors, electrical apparatuses and items are attached to net very
easily with IOT. IOT is even employed for data transfer and accumulation of records
from electrical appliances or gadgets and connect it to the IOT dashboard via certain
communication protocols [5]. May be connected the use of Bluetooth, Ethernet,
Wi-Fi and so forth to the internet. An arrangement to raise cautions in the event
of discovery of lethal gases, fuel spillage over the safe breaking point can assist
with maintaining a strategic distance from mishaps [6, 7]. A model dependent on
ARDUINO/IOT module has been created dependent on the above idea for location
and ongoing checking of the modern floor. IOT gadgets for the most part have some
cloud administration to oversee the gadget from the web or versatile applications
[8]. The purpose of a gadget being arranged and it can get to easily from anyplace
through web association.
On applied to reveal numerous commonplace parameters utilized in industries
which include strength, liquid degree and dc motor pace manipulate [9]. On this
paper, factory and manufacturing technique tracking is finished by IOT in unique
implementation. On this paper eWON Flexy IoT module as shown in Fig. 1, CPU
315 DP PLC is used for accumulating facts from equipments and transfer it right into
Fig. 1 Concept layout

Predictive Maintenance of Industrial Equipment’s Using IOT 259
the cloud. eWON Flexy is the IoT module [10]. Business process tracking device
includes 4 application percent based coloration mixing, liquid stage manage and
tracking, dc motor velocity manipulate and electricity monitoring.
2 Methodology
The proposed solution is the vision of any organization have to be to maximize

productiveness with guaranteed employee safety and environmental obligation. A
gadget for actual time monitoring is proposed through deployment of sensors at
exclusive points of the company premises as in Fig. 2. The first rate sensors come
across the level of emissions, misalignment, temperature, frequency, current, voltage
and lots of others factors that might also send notification to the ground manager
through Google Cloud. An arrangement to raise cautions in the event of discovery
of issues and damages over the safe breaking point can assist with maintaining a
strategic distance from mishaps. A model dependent on ARDUINO/IOT module
has been created dependent on the above idea for location and ongoing checking of
the modern floor. IOT gadgets for the most part have some cloud administration to
oversee the gadget from the web or versatile applications. The purpose of a gadget
being arranged and it can get to easily from anyplace through web association. on the
web regulating of mechanical procedure have gotten most extensive angle for modern
development and benefit all things considered used to diminishes production, and
costs just like maintenance issue. Right now, forms like meter observing, load, speed
Fig. 2 Concept block diagram

control of motors, PLC processes and analog controls are observed through android
mobiles, PCs etc. using IOT. We use the concept of Machine Management System
(MMS) to control and supervise the facilities remotely and with better efficiency.
IoT gadgets, motors and equipment in industries are prone to wear and tear over
a period of usage. Industrial motors are used for heavy machinery equipments and
use a lot of load and stress to do the required job properly. Hence in this paper we
will talk about the prevention of wear and damage and reduce the downtime and also
increase the up time. The motors in industries are equipped with a smart sensor. The
sensors we will be:
• Vibration sensor
• Temperature sensor.
The sensors are attached on to the motor or any industrial equipment that we
need to monitor. The vibration sensor that is used is a accelerometer sensor. When
the motor undergoes a slight misalignment or a spike in temperature that causes it
to work vigorously, the sensors senses the change in frequency of its rotation in
its radial or axial position in the form of analog waves. The same is applied to the
temperature sensors. The frequency read is sent into the A/D converter that converts
the input signals from analog to digital signals. The Variable frequency drive (VFD)
is kept attached to the motor for controlling the frequency of the motor speed as per
the needs of the industrial needs. The VFD can also show the current and voltage at
which the motor is running.
After the conversion of the signals into digital format, the signals are sent from
the sensors and VFD to the PLC for processing. The signals received are anal-
ysed in the PLC. The signals are compared with the ladder program that is build
with the help of SIMATIC manager. PLC framework has the fundamental practical
section of processor, memory, power gracefully, input/yield interface, interchanges,
and industrial equipment devices. The PLC framework works only if the user inputs
the required instructions and algorithms of work into the program that helps it perform
the needed process executions and atomization. These instructions are input into the
PLC by the user or officials during handling. The info/yield channels give disen-
gagement and sign molding capacities so sensors and actuators can frequently be
straightforwardly associated with them without the requirement for other hardware.
Yields are indicated as being of hand-off sort, transistor type, or triac type. The
interchanges interface is utilized to get and transmit information on correspondences
systems from or to other remote PLCs.
2.1 Ladder Logic
LCs frequently use “Ladder Logic Programming” a short-sighted programming

language that is appropriate for modern applications. Ladder logic is similar to
stacking up relay circuits. The program involves the use of relay circuits to form
connection to various sensors in the system linking to the motors or any industrial
equipment. Presently in our case the ladder logic uses bit logic operation. The LD is
the standardized PLC programming operation. The program uses different graphical
elements in contrast to other coding softwares. Ladder logic symbols look like electric
symbols that gives better user interface that will help the engineers to assign different
symbols. The lines in the PLC ladder logic programme are called rungs. These rungs
can be as many in relation to the number of equipments that is connected to the rungs.
During execution the ladder logic rungs are read and executed one rung at a time,
the PLC executes one symbol at a time. There are many symbols like normally open,
normally closed, output coil etc. as in Fig. 3. The normally open contact specifies
the switched off state, normally closed consists of the closed contact through which
the current flows and the output coil consists of the motor or any equipment. When
the programme is running, it reads the states of these symbols then convert it into
binary values of 0’s and 1’s also considered as low and high respectively.
The program build is used for analysing the values that the PLC receives. The
program has functions known as counter and comparators. The counter will count
the input signal value and the then sends it to the comparator. The comparator is
another function that compares the values of the input digital signals with the safe
running values inputed by the industrial user with the help of SIMATIC manager. If
the values analysed by the PLC from the sensors are found to be abnormally high or
low compared to the values set by the user then the PLC send the processed data to
the IOT. The signals are sent to the IOT which triggers a prompt to send a message
to the GSM module. A software called eBuddy is used for generating the message
that is needed to be sent to the maintenance officials or any other higher authority
manager.
2.2 Vibration Monitoring
Speed and quickening are the two generally normal parameters that are checked
with vibration sensors to distinguish issues on turning hardware. Speed is the speed
estimated a solitary way. It is estimated in ‘mm/sec’. Speeding up is the pace of
progress of speed.
Vibration sensors shown in Fig. 4 are devices or apparatus that measures a wide
range of readings and values that include speed, frequency, velocity, strain, etc.
These sensors utilize the piezoelectric impact by changing over them to an electrical
charge these sensors can also help in finding out the cracks, wears and tears that the
machines undergo if it has any faulty motors. These sensors can be used in many
industries and facilities. There are two types of vibrations axial vibration and radial
vibration. Notwithstanding the sensor qualities, you should think about the neces-
sary equipment and programming to appropriately condition, procure, and imagine
accelerometer estimations. For instance, you have to perform signal handling on
crude vibration signs to show the information in a progressively significant arrange-
ment, for example, the recurrence range. To all the more likely acquaint yourself with
Fig. 3 Ladder Logic
the estimation equipment and programming handling essential for accelerometer

estimations, download the Engineer’s guide to accurate sensor measurements.
With speeding up we can screen a lot higher frequencies. We use speeding up
estimations to identify issues that apply a power, for example, little stuns and effects
Fig. 4 Vibration sensors
that happen at the beginning of bearing harm, gearbox shortcomings, and siphon
cavitation and so on.
2.2.1 Different Levels of Monitoring
i. Machinery Protection sensor

Basic apparatus assurance sensors offer checking of the general vibration levels,
estimated in speed (mm/sec). These are regularly applied to the ISO 10816
norm. This degree of checking won’t analyze a particular machine issue, yet
will distinguish an adjustment in the general vibration level brought about by
deficiencies, for example, unbalanced, misalignment, looseness, belt drives.
ii. Intelligent Vibration Sensor
Smart vibration sensors encourage checking of both the speed levels (mm/sec)
also, increasing speed levels (g). This degree of observing won’t ordinarily
analyze a explicit machine shortcoming. In any case, by observing both speed
and increasing speed it is conceivable to recognize vibration increments brought
about by a more extensive scope of issues, for example, Unbalance, Misalign-
ment, looseness, belt drives, Bearing defects, Rubbing, Gears, Fluid noise,
Impact detection.
iii. Fault analysis
Flaw examination frameworks can be customized to distinguish explicit hardware

issues by checking the individual deficiency frequencies that identify with these
different vibration issues. When distinguished, the client would then be able to be
advised which flaw is beginning to create on the machine. This degree of point by
point examination takes into consideration progressively proficient support systems
to happen. These frameworks can be modified to analyze explicit machine blames,
for example, Unbalance Misalignment, Looseness, Belt drives, bearing defects,
Rubbing, Gears, Fluid noise, Impact detection (Fig. 5).
Fig. 5 Screenshot of conveyor status screen
2.3 Wireless Monitoring
An IOT module is a device that is installed in various devices and appliances in today’s
technological world. It helps in interconnecting and linking different machinery in
industries or factories with each other that when combined forms a cyber physical
system. The IOT helps in sending receiving and transferring information with all of
these devices or machinery. In general it helps in the autonomous communication
between each of these devices without human intervention. The IOTs always depend
on the internet for the transfer of information consistently on network. This is on
the grounds that IOT applications need to send information naturally, progressively
without any sort of user interface with the devices in order to transfer data. These
devices have good reliability, cost effectiveness, and rigidness. Not at all like customer
gadgets that are ordinarily conveyed by the end-client and utilized in controlled
conditions, IOT gadgets are regularly sent in extraordinary situations and remote
like transportation holders or under seawater in harbor the executives arrangements.
IOT situations can be immensely requesting with outrageous temperatures, vibration,
and mugginess.
A GSM module shown in Fig. 6 is a chip or circuit that can provide a connec-
tion between the module and the remote operator or overseer who has established
the connections. The connections can be between any portable devices like tablets,
phones, laptops, chrome books, etc. for interactions between the module and GSM
framework. The modem is a basic part here. A GSM includes a GSM module along-
side some different parts like correspondence interface power gracefully and a few
markers. The Ethernet cables and wireless connection based on wifi and internet
Fig. 6 GSM/GPRS module
we can link these devices together without much issues as it has a wide range of
compatibility. GSM Modules permit remote correspondence with different gadgets
and instruments. Since this technology allows us to remotely and securely connect to
various devices and monitor the company activities it is easier to perform the secu-
rity monitoring, hoe automation, automotive GPS tracker, online and e commerce
transactions. GSM module associates with the network utilizing a SIM cards from
any companies like airtel Vodafone, etc. The basic radio frequencies where a run of
the mill GSM Module works are 900 and 1800 MHz.
Previously mentioned errands, cultivated with the assistance of AT Commands.
AT Commands are pieces of Command Set, which are characterized initially for a
modem. GSM Network likewise actualizes a comparable AT like orders for its GSM
Modules. The processor or controller to which the GSM/GPRS Module is associated
with, is answerable for sending the AT Commands to the module. Accordingly, the
GSM Module performs order explicit errands like noting a call, send a SMS Message,
etc. Even through the AT Commands may appear to be nonexclusive, it is fitting to
allude with the information gave by the maker of the GSM Module for right and
complete rundown of AT Commands.
2.4 Wireless Communication
Nowadays, The eBuddy software must be installed in every official’s phones or tables
to receive the messages. The message is sent to the GSM module that sends it to the
cloud or user. The user receives the messages through the eBuddy app. The messages
sent depends on the gravity of the situation i.e. if the problem faced by the motor or
equipment is low then the messages are only sent to the lower maintenance officials
and if the problem persists for a longer time or if it’s more grave and precarious
then the messages are sent even to the higher officials who can get the details of
the matter and tend to it as soon as possible which prevents any further troubles and
circumstances. The user gets the message in the form of Email or SMS. The received
data is stored in the cloud in the form of spreadsheet. The database of every ongoing
or previous operations of the equipments are stored in the cloud and the officials or
managers can get a monthly or yearly data report on the equipments and its values. A
vitality observing dependent on IoT is utilized to improve the vitality the executives
framework.
The proposed framework can screen and measure the situations and state progres-
sively inside the plant. By determination of exact and precise sensors, a test arrange-
ment to measure the gases was created. The information from the sensors were
gathered progressively and transferred in Google spread sheet shown in Fig. 7.
The communication between PLC and VFD is done with the help of a STARTER
software through an IP gateway. The gateway is a device for controlling and trans-
lating CIP protocol to SEIMENS profibus protocols. Then we configure the VFD to
the SEIMENS starter software in the form of a profibus slave. We have to enter the
motor or equipment details like the power rating, current ratings and voltage ratings
and the type of motor (Figs. 8 and 9).
This product particularly intended for administrator control and observing in
SIMATIC. Procedure perception framework is a standard administrator Interface
with all the significant administrator control and observing capacities which can be
utilized in any part any innovation. SIMATIC software is present day devices for
Fig. 7 Google spread sample

Fig. 8 Light and harmful gases
Fig. 9 Temp and humidity graph
arranging SIMATIC administrator board’s conservative gadgets. Programming that

secures data on the area and reason for mistakes in plants and apparatus and along
these lines offers quick and pointed procedure diagnostics.
The proposed structure can assist with forestalling industry related mishaps
because of gas and fuel spillage by legitimate recognizable proof and cautioning
the individuals around. The proposed configuration can likewise be made appro-
priate for recognizing the contamination level from vehicles and to keep a mind the
smoke outlet from processing plants. The plan can be altered to identify the level of
gases inside pit openings with the goal that valuable lives can be spared. Utilization of
IOT in relationship with ARDUINO/IOT module makes the framework modest and
moderate. The light weight framework can be installed in Jacket, Helmet and Wrist
observes which can be worn by labourers to investigate the highlights of Wearable
Technology [11]. The framework likewise empowers basic man to screen the nature
of air in their region by means of cell phone application.
4 Conclusion and Future Work
In this paper an Intelligent Industry Monitoring framework in light of IOT is

proposed which can successfully screen the Mechanical floor. A model dependent on
ARDUINO/IOT module was created which could detect the centralization of current,
vibrations and temperature. The continuous information acquired from the various
sensors has been transferred to Google cloud. Notwithstanding this other parameters
like temperature, dampness and light force were estimated. For simplicity of inves-
tigation the above information got were plotted. Arrangements were likewise made
to alarm the labourers if there should be an occurrence of any crisis. The frame-
work gives solid, precise investigation to forestall any instance of mishaps. This
framework utilizes ARDUINO/IOT module giving modest arrangements for secu-
rity. Slight adjustment of the model empowers the client to adjust it to any condition.
Prescient support is an up and coming modern need, for which the proposed model
can be ad libbed. If there should arise an occurrence of misalignment and damage
we need to access the issues from point to point which must be broke down further.
Also, the gases diffusing out during spillage may likewise join among themselves
delivering different results which must be managed in detail. These cases can open
an eye for the sprouting analysts.
References
1. Leman AM, Hidayah N (2013) Occupational safety and health: workers and industrial safety
monitoring for sustainable work environment development. Health and Safety, AET
2. Wu-Zhao, Lei-Hong L, Yue-shan H, Xiao-ming W (2013) A community health service archi-
tecture based on the Internet of Things on health-care. In: World congress on medical physics
and biomedical engineering, vol 39, pp 1317–1320
3. Nayak R, Panigrahy MR, Rai VK, Rao TA (2017) IoT based air pollution monitoring system.
Imperial Jo Interdisc Res 3(4)
4. Deshmukh MR, Renuka SJC, Chavan A, Atarde G (2018) IoT based air & sound pollution
monitoring system. IJRIIE 4(3)
5. Xiaojun C, Xianpeng L, Peng X (2015) IoT based air pollution monitoring and forecasting
system. In: International conference on computer and computational sciences (ICCCS), pp
257–260
6. Ramagiri R, Reddy SCM, Reddy S (2015) Development of IoT based vehicular pollution
monitoring system. In: International conference on green computing and internet of things,
(ICGCIoT)
7. Spachos P, Song L, Hatzinakos D (2014) Gas leak detection and localization system. In: 2014
IEEE 11th consumer communications and networking conference (CCNC), Las Vegas, NV, pp
1130–1131. https://doi.org/10.1109/CCNC.2014.6994392
8. Han H, Kang H, Son J (2017) Intelligent operation monitoring IoT tag for factory legacy device.
In: 2017 international conference on information and communication technology convergence
(ICTC)
9. Siddula SS, Babu P, Jain PC (2018) Water level monitoring and management of dams using
IoT. In: 2018 3rd international conference on internet of things: smart innovation and usages
(IoT-SIU)
10. Lee J-L, Tyan Y-Y, Wen M-H, Wu Y-W (2017) Development of an IoT-based bridge safety
monitoring system. In: 2017 international conference on applied system innovation (ICASI)
A Simplified Beginner’s Guidelines
for Design and Fabrication of Prototype
Electrical Vehicle
P. Ramesh Babu, P. Vigneshwar, R. Udaya Simha, S. Tanweer Ahamed,

S. Vengatesh, and V. Vijay
Abstract The aim of this paper is to build a prototype electric vehicle out of struc-
tural materials. It is influential in the development of a modern, safe, and envi-
ronmentally sustainable mode of public mobility. The objective of its design is to
create a lightweight, compact three-wheeled electric vehicle frame. The design phase
entails the creation of a 3D model, a practical prototype, and frame refinement using
CAD software and the material parameters. The electrical and mechanical study is
performed, the results recorded 125 km per charge, and the weight of the vehicle
is 180 kg. The top speed is 40kmph along with >80% efficiency of the BLDC hub
motor.
Keywords Electric vehicle · Lithium-ion battery · BLDC hub motor ·

Suspension · Steering
1 Introduction
We breathe contaminants in excess of the air we require; as a community, we are

planting trees and adhering to rules to mitigate this problem; however, pollution
emitted by automobiles has a detrimental effects which could lead to health prob-
lems including certain tiredness, and even respiratory issues. Adopting sustainability
implies not only the plantation of more trees, as such as the use of clean energy for a
healthier atmosphere and emissions reductions. It’ll last up to 39 years depending on
the (R/P) ratio for fossil fuels (approx.). Green technology should be used now just to
ensure that natural resources are protected for subsequent generations. Electric cars
were favored because they were comfortable, safe to drive, also did not release any
pollutants. Thomas Alva Edison was researching on enhancing energy effectiveness
P. Ramesh Babu (B) · P. Vigneshwar · R. Udaya Simha · S. Tanweer Ahamed · S. Vengatesh ·

V. Vijay
Saranathan College of Engineering, Trichy, Tamil Nadu, India
P. Vigneshwar
https://doi.org/10.1007/978-981-16-4943-1_25
272 P. Ramesh Babu et al.
of a battery in 1901. Meanwhile, Ferdinand Porsche builds the world’s first hybrid-
electric vehicle, powered by a battery and a gasoline engine. There were downtimes
for electric vehicles, as combustion engines were used often but they were affordable
and plentiful, and gas-powered vehicles became successful that they pushed up the
price of fossil fuels. Then NASA came up with the concept of using Lunar Rovers
to operate solely on electricity in order to promote electric vehicles, which narrowed
the odds of electric vehicles. The CitiCar by Sebring-Vangaurd was a notable electric
vehicle with a range of 50–60 miles, and it wasn’t enough. Electric vehicles with the
same range and acceleration as combustion engines are being produced again after
several years of testing.
2 Design and Fabrication
Electric vehicles are classified into three categories: two-wheelers, three-wheelers,

and four-wheelers. The three-wheelers are preferred because they are lightweight
and have a robust build. The number of front and rear wheels on a three-wheeler
defines the vehicle’s stability. Figures 1 and 2 shows the designs of electric three
wheeled vehicle.
2.1 Chassis
Chassis is the metal frame of the vehicle without the body. The frame is an integral
part of vehicle design and fabrication because it determines factors such as vehicle
safety, stability, balance, and ergonomics [1].
Fig. 1 The frame drawing

of roll cage
A Simplified Beginner’s Guidelines … 273
Fig. 2 The CAD model of

the roll cage
2.2 Frame Definition
The design of the vehicle implies a light weight and safe. The chassis must withstand
the weight applied and must not fail at any instant. There are two wheels on the front
and one wheel on the rear which decides the overall stability and balanced weight.
2.3 Frame Dimensions
The overall length of the chassis is 2047.18 mm, overall width is 800 mm, and the
overall height of the chassis is 1000 mm. The wheelbase length is 1600 mm. Since
there are two wheels on the front side, the wheel track length is 1400 mm. The ground
clearance of a vehicle is 150 mm. The total weight of the roll cage is 23.729 kg with
firewall clearance of 55 mm (Fig. 3).
Fig. 3 Wheelbase and wheel track of the prototype model

3 Aisi 4130
The vehicle’s roll cage is made up of metal bars in the form of pipes. Cold drawn steels
costs around Rs. 80–Rs. 200 per kg, which is why it is preferred. AISI 4130 (American
Iron and Steel Institute) alloy steel is used which has extraordinary welding steel that
can be welded in all commercial methods and good mechanical properties. Tungsten
Inert Gas (TIG) welding is used for this type of alloy steel. The properties is mentioned
in the below table,
Properties
Chemical composition Chromium, Cr 0.80–1.10
Manganese, Mn 0.40–0.60
Carbon, C 0.280–0.330
Silicon, Si 0.15–0.30
Molybdenum, Mo 0.15–0.25
Yield strength 460 MPa
Tensile strength 560 MPa
Young’s modulus 210 GPa
Poisson ratio 0.3
Density 7.85 g/cm3
4 Stress Analysis
The impact of force is analyzed using ANSYS 17.2 software and the vehicle’s param-
eters, which necessitates the calculation and display of stress effects on the vehicle
using the maximum principle stress, deformation. The final results will be used to
evaluate what changes can be made to the frame to increase the effect on the driver’s
safety (Figs. 4, 5, 6, 7, 8 and 9).
5 Mechanical Systems
We encounter bumps and speed breakers when driving on the highways, so steering
and suspension help. The tires appear to slip while driving around a curve. To prevent
this, we used the Ackerman mechanism, which reduces it. A few parameters in
steering systems must be determined in order for them to operate correctly. The
Hydraulic suspension system supports the vehicle and reduces the bumps that shocks
absorb when driving on a rough road (Figs. 10 and 11).
Let us see the calculation and formulas for the steering mechanism,
Fig. 4 Front impact on stress
Fig. 5 Front deformation

wheelbase
Ackerman = tan−1 wheelbase (1)
tan δoutside
− trackfront

δinside
Ackermanpercentage = × 100% (2)
Ackerman
Using the formula,

a
tan θo = (3)
d +a
Fig. 6 Side impact on stress
Fig. 7 Side deformation
a
tan θi = (4)
d +a
where θo , θi are outside and inside angles.

For acceleration of the prototype,
Air resistance(Ra ) = (K a × A × V 2 ) N (5)

Fig. 8 Rear impact on stress
Fig. 9 Rear deformation
where K a is proportionality constant of acceleration, A is acceleration and V is

velocity.
Rolling resistance(Rr ) = (K × W × g)N (6)
where K depends on nature of the load, W is total weight and g is gravitational

acceleration constant.
Fig. 10 Double wishbone

suspension
Fig. 11 Dual suspension

with swing arm
Te × Gear ratio × Transmission efficiency

Traction force F = (7)
Wheel Radius
where T e is torque.
F−R
Acceleration a = in m/s (8)
m
where
R = Ra + Rr (9)
Disc brakes mainly focuses on squeezing the pairs of pads against a disc using
calipers. This slows down the rotational speed and hold it stationary. We use brake
oil dot 3 in our electric vehicle [4].
6 Design of Electrical Systems
6.1 Selection of Electric Drive
The electric vehicle system is divided into four sections, which describes how
a concept vehicle works. The four sections are namely—Battery, motor, motor
controller, and Dashboard. The motor requirements are 1 kW, 48 V hub motor,
and their capability determines the prototype vehicle’s performance efficiency and
speed. Normally, we use a regular motor, but to keep it small and lightweight. The
formula for finding current I is,
Pin = (V × I ) in W (10)

Pin
I = in A (11)
V
In order to find the value of Pin , We can apply the values of Efficiency E and
Output Power Pout in the formula of efficiency.

Po
η=
Pin
Po = (Pin × η) in W (12)

2×π × N ×T Pout × 60
Pout = =T = in Nm (13)
60 2×π × N
6.2 Selection of Battery
Lithium-ion battery, they are rechargable battery, cylinderical in shape, and the rating
we are using is 48 V, 75 Ah. It is a closed battery pack with Battery Management
System (BMS) in it. BMS is a electronic integrated circuit that monitors the battery.
It weighs around 19 kg which is 30% less compared to Lead acid batteries [5].
6.3 Power Line Diagram
We used safety devices like fuses, MCB (C-type), and kill switches.

Pin
Fuse rating = 125% in A (14)
pf × V
MCB rating = 125% of Imax in A (15)
7 Final Prototype of Electric Vehicle—Photographs
8 Conclusion
This document’s conclusion is solely for the purpose of designing and fabricating
a prototype electric vehicle as a beginner. Electric cars should be prioritized as the
cost of gasoline rises and pollutes the environment. Until developing the prototype,
this analysis demonstrates the basic specifications. Prototypes will be needed in the
coming years for larger companies to test new EVs concepts.
Acknowledgements The authors would like to express their gratitude to the National Solar Vehicle
Challenge (NSVC) and the Saur Urja Vehicle Challenge (SUVC) for validating our prototype electric
vehicle through competition in India. And we’d like to thank our college for supporting us to work
in various aspects, as well as fabricate the completed prototype vehicle.
• 2nd Runner-up in NATIONAL SOLAR VEHICLE CHALLENGE 2019
• Runner-up in SAUR URJA VEHICLE CHALLENGE 2020.
References
1. Nabil T, El-Naghi BE, Saeed M, Kamal A, Gharib E, Mohsen M, Ahmed I (2019) Design and
fabrication of prototype battery electric three wheeled vehicles. J Asian Electr Veh 17(2):1823–
1834
2. Casalino G, Moradi M, Moghadam MK, Khorram A, Perulli P (2019) Experimental and
numerical study of AISI 4130 steel surface hardening by pulsed Nd: YAG laser. Materials
12(19):3136
3. Wilhelm E, Bornatico R, Widmer R, Rodgers L, Soh GS (2012) Electric vehicle parameter
identification. World Electr Veh J 5(4):1090–1099
4. Paterson J, Ramsay M (1993) Electric vehicle braking by fuzzy logic control. In: Conference
record of the 1993 IEEE industry applications conference twenty-eighth IAS annual meeting.
IEEE, pp 2200–2204
5. Yoshio M, Brodd RJ, Kozawa A (2009) Lithium-ion batteries, vol 1. Springer, New York, pp
2–3
6. Petrushov VA (1998) Improvement in vehicle aerodynamic drag and rolling resistance
determination from coast-down tests. Proc Inst Mech Eng Part D J Autom Eng 212:369–380
7. Nian X, Peng F, Zhang H (2014) Regenerative braking system of electric vehicle driven by
brushless DC motor. IEEE Trans Ind Electron 61(10):5798–5808
8. Sciarretta A, Guzzella L (2007) Control of hybrid electric vehicles. IEEE Control Syst Mag
27(2):60–70
9. Ehsani M, Gao Y, Miller JM (2007) Hybrid electric vehicles: architecture and motor drives.
Proc IEEE 95(4):719–728
10. Duncan J (1983) Linear induction motor equivalent-circuit model, electric power applications.
IEE Proc B 130(1):51–57
11. Chan CC, Chau KT (2001) Modern electric vehicle technology (No. 47). Oxford University
Press on Demand
12. Prabhu VN, Manigandan N (2014) Design and fabrication of solar transport vehicle. OSR J
Mech Civ Eng 11:14–19
Performance Analysis of IPT with DC
to DC Converter for E-Vehicle
Application
V. Senthil Nayagam and L. Premalatha
Abstract This work proposes the utilization of Inductive Power Transmission (IPT),
for charging the E-Vehicle’s battery with the help of an efficient DC TO DC converter.
In this IPT method, the DC TO DC converter is operated with the inclusion of
transmitting and receiving coils. The required power for transmitting coil from the
input source (AC Source or solar energy) is controlled and supplied by the DC TO
DC converter 1. Meanwhile from the receiving coil side of IPT system, the obtained
power supply will be controlled by DC TO DC converter 2 for charging the battery.
In order to identify the efficient DC TO DC converter, two types of converters namely
VMC (Voltage multiplier cell) based DC TO DC converter and LUO converters are
taken into account for comparison.
Keywords DC-DC converter · LUO · IPT · E-vehicle · Battery
1 Introduction
1.1 Existing System Methodology
Most of us in this modern world use a traditional vehicle that uses fossil fuel, which
is the primary factor that causes harmful damage to the ozone layer. It is therefore,
proposed traditional vehicles will be replaced by electric vehicles (EV) [1] to maintain
the quality of the air around us. This solution of using battery-operated vehicles [2, 3]
leads the automobile industries to manufacture various cars and two wheelers which
are more luxurious thus replacing the engines of the conventional vehicles by motors.
This is named as the E-Vehicle [4] and it is popular in the market but not affordable for
V. Senthil Nayagam (B)

Department of Electrical and Electronics Engineering, Sathyabama Institute of Science and
Technology, Chennai, India
L. Premalatha
School of Electrical and Electronics Engineering, VIT Institute of Science and Technology,
Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_26
284 V. Senthil Nayagam and L. Premalatha
all consumers due to its heavy initial cost. All these luxurious E-vehicles are with one
touch control which requires separate source for the control circuit and also supply
required for motor driving. Therefore heavy battery utilization is required for long
travel. Those control circuits consist of inverters and efficient DC to DC converters
[5–7] with high voltage conversion ratio in the input and output stages of the inverter
according to the supply and load demand, so that the battery life can be extended and
cooling arrangements for the battery can be minimized. But the major issue is that
the charging station is supposed to have a variety of charging points for various types
of E-vehicles [8, 9] and its battery capacities. Therefore in order to avoid this issue,
single charging point has to be designed to charge the variety of batteries of various
vehicles. To design this kind of charging point we need to incorporate the efficient DC
to DC converter [10–12] with fewer ripples in the dc supply input for the E-vehicle
battery. If the main source of supply is AC with reduced voltage, then for charging the
battery, boost converter is required for the battery charging of various E-vehicles. If
in case, the main source voltage level is very high than the required charging voltage
level, then step down DC to DC converter used to charge the battery to apply required
voltage to the battery of the various E-vehicle. A single buck-boost converter can be
used to charge the variety of battery range of E-vehicle with variety of input source
but if incase there is failure in the converter any kind of battery charging can be
stopped. So it is recommended to use separate buck converter and boost converter
so as to increase the lifetime and uninterrupted charging of the battery. But there
are some challenges in the implementation of battery-operated vehicles which are
charging the battery with wired connection for a moving vehicle, which is impossible
for longer distance; increasing the life span and reliability; and low-maintenance
operation of the battery. But there are some challenges in the implementation of
battery operated vehicles. For long distance travel, charging the wired battery is
difficult which lessens the battery’s durability and low maintenance of the battery.
Further problems are that the traditional boost converter’s voltage gain will be high
with high duty ratio, diode reverse recovery issue that leads to large current ripples,
high conduction losses, plugging charger manually / unplug operation and wired
charging chances of electrical shock. Inductive charging of battery with a charge
controller such as a powerful DC to DC converter successfully meets the above-
mentioned challenges and their solutions. LUO converter (boost converter) output,
interleaved coupled inductor dependent DC to DC converter (boost converter), buck
converter, boost converter with voltage multiplier cell performance are tested for
battery charging in our work. In addition to this, inductive wireless charging is tested
and its performances are analyzed for various distances between transmitting and
receiver coils. In the process of inductive type of charging the battery, as the distance
between transmitting (primary) and receiving (secondary) coil increases there is
a decrease in output voltage (secondary side) this is due to coupling coefficient
factor. Hence sufficient power can be obtained with the aid of the LUO converter
or interleaved coupled inductor boost converter to charge the vehicle battery. In
the existing system of wireless charging the voltage control takes place only in the
transmitting (primary) side of the coil, because the inverter in primary coil side is
operated in fixed frequency. Hence in the process of E-vehicle battery charging, the
Performance Analysis of IPT with DC to DC Converter … 285
secondary side voltage is unable to control. This causes poor performance and less
efficiency of the battery charging. And also as the inverter is a square wave type,
the harmonics content will be rich in the supply. This leads to the failure of existing
wireless battery charging system.
In the existing inductive charging system without secondary side control of voltage
system consists of AC supply source, rectifier block which will convert AC to DC
and it is supplied as an input to the inverter controlled by DSPIC30F2010 controller.
Then the transmitting coil with compensation network will transfer AC supply from
transmitting coil to receiving coil through air gap. Finally, the battery will receive
the power from the receiving coil through an uncontrolled rectifier. In this system the
voltage on the secondary side is not controlled with respect to the battery. Therefore
uninterrupted charging of battery is not ensured and hence it is difficult to drive effec-
tively under load varying conditions. Moreover losses associated with the inductive
power transfer (IPT) method are very high.
Therefore, in overcoming the above mentioned issues related to the effective
charging of battery, this research work concentrates on proposed technique of novel
wireless charging with reduced losses and also on efficient control of secondary side
wireless transmitted supply.
2 Proposed Methodology
2.1 Process of Identifying Effective DC to DC Converter

for Battery Charging by Receiver Coil of IPT System
The block diagram of coupled inductor based interleaved boost converter with voltage
multiplier cell in IPT implementation shows the overall circuit implementation with
their connection between every circuit unit is presented in the Fig. 1. The AC input
source of 48 V is converted as DC supply using rectifier and this DC supply is fed to
the controlled inverter which is used to convert as AC supply and it is controlled by
the DSPIC30F2010 controller in order to maintain the primary coil voltage according
to the requirement.
The controlled output from the transmitter coil is transmitted wirelessly with the
aid of compensation network to the receiver coil. This received AC supply from the
receiver coil is further converted as DC supply by the rectifier in order to control
the secondary voltage as DC supply and to maintain the required level of voltage for
E-Vehicle battery charging requirement by the coupled inductor based interleaved
boost converter with voltage multiplier cell. This controlled secondary voltage is
applied to the battery to charge or for DC load requirement according to the supply
availability in the feedback voltage of the input side of the coupled inductor based
interleaved boost converter with voltage multiplier cell.
Fig. 1 Block diagram of coupled inductor based interleaved DC to DC converter in IPT

implementation
2.2 Simulation Circuit Coupled Inductor Based Interleaved

DC to DC Converter in IPT
Simulation diagram of coupled Inductor based interleaved DC to DC converter with

voltage multiplier cell is shown in the Fig. 2 that is simulated with the specification
shown in Table 1 and measures various voltage in the circuit.
Simulation diagram of LUO DC to DC converter is shown in Fig. 3. It is simulated
with the specification shown in Table 2 and measures various voltage in the circuit.
Table 3 shows the resonance condition changes of the IPT coils, by the input
and output voltage variation of the coupled inductor based interleaved DC to DC
converter for varying load conditions and duty ratio. From this table, it is clearly
indicated that, the wireless transmitter coil input and receiver coil output voltages
are drastically reduced, as long as, there is an increase in the duty ratio and the load.
Table 4 shows the resonance condition changes as the input and output voltage
varies by the LUO DC to DC converter for varying load conditions and duty ratio.
From this table, it is clearly indicated that, wireless transmitted input and output
voltages are slightly reduced as the duty ratio and load increases when compare to
coupled inductor based interleaved DC to DC converter.
From the waveform shown in Fig. 4, it is observed that, without LUO converter
(without secondary side control), the inverter’s closed loop control (only primary
Fig. 2 Simulation diagram of coupled inductor based interleaved DC to DC converter in IPT

implementation
Table 1 Simulation
Simulation Parameters Value
specifications of interleaved
coupled inductor based DC to Switching frequency 15 kHz
DC converter Input voltage 12~40 V
Output voltage 20~65 V
Output power 40 W
Inductor value 0.150 uH
Capacitor value 1500 uF
MOSFET IRF840 (400 V, 8 A)
DIODE MUR460 (600 V, 4 A)
control) is insufficient to charge the battery connected on the secondary side.

Therefore there will not be any flow of current into the battery to charge the battery.
3 Conclusion
From the obtained simulation results of coupled inductor based interleaved DC to

DC converter with VMC and LUO dc- dc converter in the implementation of WPT
system, for finding effective DC to DC converter −2 in the secondary coil of IPT
system, it is observed that coupled inductor based interleaved converter with VMC
DC to DC converter produced high output voltage for the duty ratio compared to
Fig. 3 Simulation diagram of LUO converter with IPT Implementation
Table 2 Simulation
Switching frequency 15 kHz
specifications of LUO DC to
DC converter Input voltage 12~40 V
Output voltage 20~65 V
Output power 40 W
Inductor value 0.150 uH
Capacitor value 1500 uF
MOSFET IRF840 (400 V, 8 A)
Table 3 Response of coupled inductor based interleaved DC to DC converter according to variation

of duty ratio and load
Load Duty cycle 0.1 0.15 0.2 0.25 0.3 0.4 0.5
100 Input voltage 30.76 30.26 29.35 28.58 24 24.72 21.28
Output voltage 62.95 63.37 63.89 64.06 56.52 64.09 62.48
50 Input voltage 22.56 21.92 20.86 20.06 15 16.26 13.39
Output voltage 34.29 45.19 44.92 44.42 43.97 41.15 38.25
Table 4 Response of LUO converter according to variation of duty ratio and load
Load Duty cycle 0.1 0.15 0.2 0.25 0.3 0.4 0.5
100 Input voltage 17.22 15.85 13.22 11.71 9.98 7.82 6.05
Output voltage 56.86 55.54 52.09 49.47 46 41.1 36.06
50 Input voltage 13.77 10.21 8.79 7.83 6.61 5.02 3.84
Output voltage 36.28 31.8 29.88 28.34 26.06 22.55 19.54
Fig. 4 Secondary side control waveform without and with LUO converter
the LUO DC to DC converter. But during variation of load, distance between the
transmitter and receiver coils, it is found that, high output voltage which is usually
obtained in the coupled inductor based interleaved DC to DC converter with VMC is
reduced drastically when compared to the LUO DC to DC converter. Because, LUO
converter is not much affects the resonance condition, which is the main factor for
efficient wireless power transfer. The comparative results from Tables 3 and 4 which
shows the effect of converters on the resonance condition changes between the WPT
coils, which leads to reduction of input voltage of the transmitting coil. Henceforth,
output voltage of the receiver coil is also reduced, because the WPT coils are works
under mutual inductance principle. During the LUO converter operation of increasing
the receiver coil voltage to the required level of battery charging, which is not much
reduce the input voltage of the WPT transmitting coil when compare to drastic
reduction of input voltage of WPT transmitting coil under VMC based Interleaved
coupled inductor DC to DC converter operation. So, LUO DC to DC converter
performs better than the coupled inductor based interleaved converter with VMC
DC to DC converter in the implementation of IPT for E-Vehicle battery charging.
Therefore LUO DC to DC converter is preferred for wireless power transmission of
battery charging. Similarly, from the primary side control (inverter output control)
and secondary side control (LUO converter output control), it is observed that, both
control is required under closed loop mode of operation, for proper charging of the
battery.
References
1. El-Shahat A, Ayisire E, Yan Wu, Rahman M, Nelms D (2019) Electric vehicles wireless power
transfer state-of-the-art. Energ Procedia 162:24–37
2. Huang C-Y, Boys JT, Covic GA, Budhia M (2009) Practical considerations for designing IPT
system for EV battery charging. IEEE 6:402–407
3. Hoang DT, Wang P, Niyato D, Hossain E (2017) Charging and discharging of plug-in elec-
tric vehicles (PEVs) in vehicle-to-grid (V2G) systems: a cyber insurance-based model. IEEE
Access 5:732–754
4. Manshadi SD, Khodayar ME, Abdelghany K, Üster H (2018) Wireless charging of electric
vehicles in electricity and transportation networks. IEEE Trans Smart Grid 9:4503–4512
5. Berkovich Y, Axelrod B (2011) Switched-coupled inductor cell for DC to DC converters with
very large conversion ratio. IET Power Electron 4(3):309–315
6. Makowski MS, Maksimovic D (1995) Performance limits of switched capacitor DC to DC
converters. In: Proceedings of IEEE 26th power electronics specialist, pp 1215–1221
7. Kumar MK, Sekhar P (2012) High step up voltage gain achieved in DC TO DC converters
using Linear Peak Current Mode control technique. Int J Mod Eng Res 2(3):870–875
8. Jeong S, Jang YJ, Kum D, Lee MS (2019) Charging automation for electric vehicles: is a
smaller battery good for the wireless charging electric vehicles? IEEE Trans Autom Sci Eng
16:486–497
9. Castiglia V, Noia LD, Livreri P, Miceli R, Pellitteri F, Viola F (2017) An efficient wireless
power transfer prototype for electrical vehicles. In: International conference on renewable
energy research and application (ICRERA), vol 10, pp 1215–1220
10. Yie-Tone C, Zong-Xing Lu, Liang R-H (2018) Analysis and design of novel high step-up DC
TO DC converter with coupled inductors. IEEE Trans Power Electron 33(1):425–436
11. Tofoli FL, de Souza Oliveria D, Torrico-Bascope RP Jr, Alcazar YJA (2012) Novel non isolated
high voltage gain DC to DC converters based on 3SSC and VMC. IEEE Trans Power Electron
27:743–749
12. Torrico-Bascope GV, Vasconcelos SA, Torrico-Bascope R, Fernando LM, de Oliveira DS,
Carlos GC (2006) A high step-up DC TO DC converter based on three-state switching cell.
IEEE, pp 998–1003
Operating Cost Analysis of Microgrid
Including Renewable Energy Sources
and a Battery Under Dynamic Pricing
Hephzibah Jose Queen, J. Jayakumar, and T. J. Deepika
Abstract In this paper, minimization of operating cost of DC microgrids is formu-

lated. Utility grid, solar, wind and battery is associated with this formulation. In
this optimization problem, both with and without losses are considered in the power
flow model. Impact of renewable energy sources on reduction in operating cost is
discussed. Based on the solar radiation and air density, power generation from solar
and wind are calculated respectively. Heuristic method is used to solve this mini-
mization problem. To analyze the operating cost, a six-bus customized system is
used. To calculate the cost structure, three different cases are considered. The first
case considers the system without any renewable energy sources or battery storage.
In the second case, a solar and a Wind Energy Conversion system are added to the
system and in the third case, the battery is included with the existing system. With the
inclusion of renewable energy system, dynamic pricing and various load conditions,
the proposed algorithm is minimizing the operating cost considerably.
Keywords Microgrid · Solar power · Wind turbine · Battery and dynamic pricing
1 Introduction
The upcoming energy revolution entails the integration of various distributed energy
sources. This means an integration of renewable energy sources like solar and wind
power. Chendal [1] in his paper states the establishment of micro girds is one option
to integrate these new distributed energy sources in an effective way. Strunz et al.
[7] describes a micro grid is an almost autonomous working, low voltage electricity
network including different residential loads and generating units. To guarantee a
H. J. Queen (B) · J. Jayakumar

Department of Electrical and Electronics Engineering, Karunya Institute of Technology and
Sciences, Coimbatore, India
J. Jayakumar
T. J. Deepika
Department of Electrical and Electronics Engineering, Sona College of Technology, Salem, India
https://doi.org/10.1007/978-981-16-4943-1_27
292 H. J. Queen et al.
stable operation of the microgrid, there is one connection point to the utility grid. If
the energy generation in the microgrid is higher than the demand, electrical energy
is feed in the utility grid respectively if the energy generation in the microgrid is
lower than the required demand, additional energy is needed from the utility grid. To
minimize the operating costs of the system, a battery storage unit can be added to the
grid. Furthermore, dynamic pricing represents a good opportunity to minimize the
total operating costs for a specific time-period (here: 24 h). Dynamic pricing means
that the electricity price varies depending on supply and demand instead of being
fed.
The following calculations show the cost benefit of the integration of renewable
energy sources and the effect of a battery storage unit added to the micro grid.
Hephzibah et al. [2] designed a system to simplify the optimization model, the micro
grid is reduced to a one-bus-system with several units. Therefore, the line losses can
be neglected, and no power flow calculation is needed. Thus, the line losses are not
considered in the optimization model of the operating costs. In a second step, the
line losses of an electricity network with the same units are determined and the costs
for the losses are added to the total operating costs of the grid.
The available units are a PV Panel and a Wind Energy Conversion system, two
residential loads, a battery storage and the connection point to the utility grid. The
input data are the same in systems, the one-bus-system and the six-bus-system. To run
the load flow calculation, the results of the optimization model of the one-bus-system
are used as input values for the battery. The one-bus-system represents a DC bus,
whereas the six-bus-system are an ac network. Therefore, the load flow calculations
are also calculated for an AC system.
2 Input Data
2.1 Solar Panel
The available solar power is dependent on the direct solar radiation sundirect , the
diffuse solar radiation sundiffuse , the size of the solar panel area A and the efficiency
of the solar panel ïsolar . This relation is described by Eq. (1)
PSolar = ηsolar A(Sundirect + Sundiffuse ) (1)
A monocrystalline solar cell has an average effciency of 20%. A solar panel with
an area of A = 80 m2 is connected to the grid via converter with an efficiency of
around 96%. Thus, the overall efficiency of the solar panel is taken as ïsolar = 19.2%.
The German weather service measures weather data in 15 different climate regions
in Germany. Data used in this study case belongs to the first climate region which
is located at the north-sea cost. The measuring station is Bremerhaven, Germany.
Table 1 lists the measured sun radiation values of region one for the 1st September
Operating Cost Analysis of Microgrid … 293
Table 1 Solar radiation data and output power

Time of Radiation (W/m2 ) Solar Time of Radiation (W/m2 ) Solar
day Direct Diffuse power day Direct Diffuse power
(kW) (kW)
1:00 0 0 0 13:00 54 284 5.19168
2:00 0 0 0 14:00 49 265 4.82304
3:00 0 0 0 15:00 39 228 4.10112
4:00 0 0 0 16:00 131 238 5.66784
5:00 0 0 0 17:00 40 159 3.05664
6:00 0 2 0.03072 18:00 14 67 1.24416
7:00 13 77 1.38240 19:00 0 0 0
8:00 45 171 3.31776 20:00 0 0 0
9:00 112 256 5.65248 21:00 0 0 0
10:00 203 298 7.69536 22:00 0 0 0
11:00 249 332 8.92416 23:00 0 0 0
12:00 109 348 7.01952 24:00 0 0 0
[3]. The results of the calculated solar power by Eq. (1) are also shown in the same
table.
2.2 Wind Power
The wind turbine is connected to the bus via a generator and a rectifier. Due to
mechanical (e.g. gearbox and bearings) and electrical (e.g. generator, rectifier and
cables) losses and the coefficient of power (max. 16/27) of a wind turbine, the overall
efficiency ïwind of the turbine, the generator and the rectifier is assumed to be 33%.
Furthermore, the generated power is dependent on the air density ρ air (1.225 kg/m2 )
the swept area of the blades A (2827 m2 ) and the wind speed v. The wind power can
be calculated using Eq. (2).
Pwind = ηwind .1/2. ρair .A.v 3 . (2)
The wind speed data is taken from the same data base as the solar radiation data
by Kothari [8]. The used wind speed data and the resulting wind power values are
listed in Table 2.
294
Table 2 Wind speed data and wind power

Time of day Speed Power Time of day Speed Power Time of day Speed Power Time of day Speed Power
(m/s) (kW) (m/s) (kW) (m/s) (kW) (m/s) (kW)
1:00 5.1 75.81 7:00 2.0 4.57 13:00 2.0 4.57 19:00 3.0 15.43
2:00 4.3 45.44 8:00 1.0 0.57 14:00 5.0 71.44 20:00 2.0 4.57
3:00 3.7 28.95 9:00 2.0 4.57 15:00 7.0 196.02 21:00 3.0 15.43
4:00 3.6 26.66 10:00 4.0 71.44 16:00 5.0 71.44 22:00 2.0 4.57
5:00 3.3 20.54 11:00 4.0 196.02 17:00 3.0 15.43 23:00 3.0 15.43
6:00 3.0 15.43 12:00 4.0 71.44 18:00 3.0 15.43 24:00 3.0 15.43
H. J. Queen et al.
Fig. 1 a Structure of six-bus-system, b structure of one-bus-system (lossless 6 bus)
2.3 Battery Storage Wind Power
The technical data of the battery storage are in accordance with the Tesla Power-
wall 2.0. The maximum power is 3.3 kW and the DC energy is 13.5 kWh [5].
The assumed efficiency for the charging (ïcharge ) and discharging (ïdischarge ) process
is 93% each. The Tesla Powerwall is a lithium-ion battery pack with an integrated
DC/DC converter. Therefore, the system is very easy to install. Moreover, this battery
type is mainly used as a residential storage system in combination with solar panels.
2.4 Residential Loads
The considered 6 bus system with PV, wind, battery, loads and grid connection is
shown in Fig. 1a. The same system is reduced to a one-bus-system as in Fig. 1b
when losses are ignored. Table 3 shows the residential load data. There is one fixed
value for each hour of the day. The load data is in accordance with the values from
Chendan et al. [1].
2.5 Electricity Prices
The electricity prices are in accordance with the prices pricet publish by the European
Energy Exchange for the 1st September 2017 [4]. The prices are listed in Table 4.
296
Table 3 Residential load data

Time of day Load (kW) Time of day Load (kW) Time of day Load (kW) Time of day Load (kW)
1 2 1 2 1 2 1 2
1:00 10.84 20.39 7:00 12.03 21.76 13:00 12.98 24.30 19:00 26.50 47.42
2:00 9.35 17.74 8:00 15.58 28.59 14:00 12.60 23.39 20:00 26.02 46.42
3:00 8.78 16.67 9:00 15.07 27.64 15:00 12.44 23.40 21:00 24.17 43.41
4:00 8.67 16.36 10:00 13.00 24.41 16:00 13.28 25.05 22:00 22.11 40.27
5:00 8.61 16.26 11:00 13.14 25.43 17:00 16.50 30.51 23:00 18.31 33.90
6:00 9.24 17.30 12:00 13.41 25.83 18:00 22.77 41.35 24:00 14.60 27.52
H. J. Queen et al.
Table 4 Dynamic electricity prices

Time Price Time Price Time Price Time Price
of day (EUR/MWh) of day (EUR/MWh) of day (EUR/MWh) of day (EUR/MWh)
1:00 33.09 7:00 36.04 13:00 44.87 19:00 42.20
2:00 31.07 8:00 43.46 14:00 43.09 20:00 43.74
3:00 29.89 9:00 45.83 15:00 39.91 21:00 43.95
4:00 29.64 10:00 45.12 16:00 38.93 22:00 39.40
5:00 29.65 11:00 44.90 17:00 38.56 23:00 36.21
6:00 31.71 12:00 45.02 18:00 40.36 24:00 34.51
3 Cost Analysis: Six-Bus System Without Line Losses
A six-bus system is considered with wind turbine connected at bus one, solar panel at
bus two and battery at bus six. The connection point of the microgrid and the utility
grid is at bus three. Bus four and five represent two residential loads. As the line
losses of the microgrid are neglected, the grid is reduced to a one-bus-system as in
Fig. 1b.
3.1 Mathematical Model of Lossless System
The optimization problem of minimizing the operating costs of the microgrid can be
modeled by the following equations. The objective function (Eq. 3) depends on the
electricity price and power drawn from utility grid. The operating costs of renewable
energy sources are assumed to be zero. Power feed into the system is positive and
drawn power is negative. Equation 4 is the power balance. The system is reduced
to one-bus with several connected units as there are no line losses. At every time
step, the stored energy in the battery must be within its minimum and maximum
capacity energymax modeled in Eq. 6. The power of the battery is thus divided into
charging and discharging power (Eq. 5). According to Eqs. 7 and 8 the charging and
discharging power has to be within the technical limits of the battery.

T
min pricet .Putility,t . (3)
t=0
0 = Putility,t + Pload1,t + Pload2,t + Psolar,t + Pwind,t + Pbattery,t . (4)
Pbattery,t = Pcharge,t + Pdischarge,t . (5)


T

0≤ −Pcharge,t .ηcharge − Pdischarge,t /ηdischarge .period < energymax . (6)
t=0
Pbatterymin ≤ Pcharge,t < 0. (7)
0 ≤ Pdischarge,t ≤ Pbatterymax . (8)
3.2 Lossless System—Case Studies
To evaluate the cost structure of the microgrid, three different cases are considered.
The first case considers the micro grid without any renewable energy sources or
battery storage. The residential loads are met by the utility grid. In the second case,
a PV Panel and a Wind Energy Conversion system are added to the system [9] and
in the third case, the system is enhanced by a battery [7].
3.3 Results—Lossless System Analysis
The results of the optimization model for the three cases are listed in Table 5.
The operating costs obtained in Cases 1, 2 and 3 are e40.84, e8.41 and e8.29
respectively.
4 Cost Analysis: Six-Bus System with Line Losses
A six-bus-system with the same generating units and loads is considered as in Fig. 1a.
4.1 Mathematical Model of System with Loss
The load buses and the battery bus are modeled as PQ buses and the wind turbine
and the solar panel are modeled as PV buses. Thus, the connection point to the utility
grid represents the slack bus. The bus types are listed in Table 6. By means of the
load flow calculation, the required power from the utility grid can be identified. The
original values of R and X are also listed in Table 6. The MVA Base is 100 MV A,
the system frequency 50 Hz and the nominal voltage 11 kV.
Subsequently, the line losses ploss,t at the time t, as in Eq. (9) is the difference
between the injected power into the system and the power drawn from the system.
Table 5 Utility power and power stored for 3 case studies
Time of day PUtility (kW) PBattery (kW) Time of day PUtility (kW) PBattery (kW)
Case 1 Case 2 Case 3 Case 1 Case 2 Case 3
1:00 31.23 −44.58 −44.58 0 13:00 37.28 27.52 27.52 0
2:00 27.09 −18.35 −18.35 −3.3 14:00 35.99 −40.27 −40.27 0
Operating Cost Analysis of Microgrid …
3:00 25.45 −3.50 −3.50 −3.3 15.00 35.84 −164.28 −164.28 0

4:00 25.03 −1.63 −1.63 −3.3 16.00 38.33 −38.77 −38.77 0
5:00 24.87 4.33 4.33 −3.3 17.00 47.01 28.52 28.52 0
6:00 26.54 11.08 11.08 −1.316 18.00 64.12 47.45 47.45 0
7:00 33.79 27.84 27.84 0 19.00 73.92 58.49 58.49 0
8:00 44.17 40.28 40.28 0 20.00 72.44 67.87 67.87 0
9:00 42.71 32.49 32.49 3.3 21.00 67.58 52.15 52.15 0
10:00 37.41 −6.86 −6.86 3.3 22.00 62.38 57.81 57.81 0
11:00 38.57 −6.93 −6.93 2.655 23.00 52.21 36.78 36.78 0
12:00 39.24 −4.36 −4.36 3.3 24.00 42.12 26.69 26.69 0
299
Table 6 Six bus system data

Bus No. Description Bus type To Bus Resistance Reactance
1 Utility grid Slack 2 0.05 0.20
2 Load 1 PQ 3 0.10 0.50
5 0.20 0.50
3 Load 2 PQ 4 0.20 0.80
4 Battery PQ 5 0.10 0.30
5 Solar panel PV 6 0.20 0.40
6 Wind turbine PV 1 0.10 0.15
Simply said the difference between generation and demand. The parameter pb,t
symbolizes the power injected or drawn from the system at bus b at the time t.
For instance, this value is negative for all load buses and positive for all buses with
generating units. The cost incurred due to losses is computed as in Eq. (10). The line
losses are drawn from the utility grid. The parameter dt symbolizes the time period
of 1 h.

nbus
ploss,t = pb,t ∀t ∈ T. (9)
b=1
costloss,t = ploss,t ∗ pricet ∗ dt ∀t ∈ T. (10)
4.2 Results—Six Bus System Analysis with Loss
The cases considered in the six-bus-system are in accordance with those mentioned
in Sect. 3.2. The calculated line losses of the system and the associated costs for the
losses are shown in Table 7. The line losses are very small in comparison to the load
and the generated power by a PV Panel and a Wind Energy Conversion system.
5 Conclusion
An Optimization algorithm is developed to attenuate the operating cost of micro

grid. The cost function is dependent not only the utility power but also the real time
pricing. A six-bus system without losses and with losses is tested to authenticate the
considered algorithm.
The integration of renewable energy sources provides the opportunity to reduce the
operating costs. In this case the operating costs are reduced to less than a fourth of the
Table 7 Line losses and cost for 3 case studies

Time of day Losses (kW) Cost (EUR)
Case 1 Case 2 Case 3 Case 1 Case 2 Case 3
1:00 0.99 4.29 4.29 0.000033 0.000142 0.000142
2:00 0.82 1.95 1.97 0.000025 0.000061 0.000061
3:00 0.76 1.17 1.22 0.000023 0.000035 0.000036
4:00 0.74 1.08 1.13 0.000022 0.000032 0.000033
5:00 0.74 0.91 0.97 0.000022 0.000027 0.000029
6:00 0.80 0.86 0.88 0.000025 0.000027 0.000028
7:00 1.10 1.03 1.03 0.000040 0.000037 0.000037
8:00 1.71 1.58 1.58 0.000074 0.000069 0.000069
9:00 1.61 1.40 1.29 0.000074 0.000064 0.000059
10:00 1.30 1.88 1.88 0.000059 0.000085 0.000085
11:00 1.38 1.95 1.94 0.000062 0.000088 0.000087
12:00 1.42 1.95 1.93 0.000064 0.000088 0.000087
13:00 1.29 1.13 1.13 0.000058 0.000051 0.000051
14:00 1.22 4.14 4.14 0.000053 0.000178 0.000178
15.00 1.22 23.53 23.53 0.000049 0.000939 0.000939
16.00 1.36 4.22 4.22 0.000053 0.000164 0.000164
17.00 1.92 1.77 1.77 0.000074 0.000068 0.000068
18.00 3.44 3.21 3.21 0.000139 0.000130 0.000130
19.00 4.55 4.33 4.33 0.000192 0.000183 0.000183
20.00 4.37 4.42 4.26 0.000191 0.000193 0.000186
21.00 3.80 3.62 3.62 0.000167 0.000159 0.000159
22.00 3.26 3.18 3.18 0.000128 0.000125 0.000125
23.00 2.33 2.24 2.24 0.000084 0.000081 0.000081
24.00 1.59 1.56 1.56 0.000055 0.000054 0.000054
original operating costs. This shows that there is a high potential of saving money and
reducing the power drawn from the utility grid. The integration of a battery does not
reduce the costs by a significant value. Although the battery has a very small impact
in this study case, it could have a bigger influence by using another electricity price
profile. Finally, the simulation results show that the developed algorithm depreciates
the operating cost in both the cases (with and without losses) (Table 8).
Table 8 Comparison of Results

Case study With losses Without loss
Loss (kW) Cost (Euro) Loss (kW) Cost (Euro)
1 43.72 0.001765 40.80 40.8018
2 77.40 0.003080 8.41 8.4131
3 77.30 0.003072 8.29 8.2931
References
1. Chendan Li FC, de Bosio F (2016) Economic Dispatch for operating cost minimization under
real− time pricing in droop−controlled dc microgrid. IEEE J Emerg Sel Top Power Electron
5:587–595
2. Queen HJ, Jayakumar J (2018) Operating cost analysis of a Microgrid including renewable
energy sources without considering the losses. Int J Pure Appl Math 118(20):745–750
3. Federal OCE for building and regional planning in Germany. http://www.bbsr.bund.de/EnE
VPortal/DE/Regelungen/Testreferenzjahre/Testreferenzjahre/03_ergebnisse.html?nn=436654.
Accessed: 2017-10-04
4. eex. https://www.eex.com/de/marktdaten/strom/spotmarkt/auktion#!/2017/09/01. Accessed:
2017-10-06
5. \teslapowerwall2.0dc. https://www.solahart.com.au/media/2849/powerwall-2-dc_datasheet_
english.pdf. Accessed: 2017-10-07.
6. Strunz K, Abbasi E, Huu DN (2014) DC microgrid for wind and solar power integration. Emerg
Sel Top Power Electron IEEE J 2:115–126
7. Barelli L, Bidini G, Bonucci F (2016) A microgrid operation analysis for cost effective battery
energy storage and RES plants integration. J Energ 113:831–844
8. Kothari JDP (2004) Power system optimization
9. Brini S, Abdallah HH, Ouali A (2009) Economic dispatch for power system included wind and
solar thermal energy. Leonardo J Sci 14:204–220
A Comparative Study of Field
Distribution Properties of Different
Types of Oils Using Comsol
M. Divya Bharathi, R. V. Maheswari, and S. Senthil Kumar
Abstract Nature ester oil is the alternative to mineral oil in transformer for cooling
purposes. This paper mainly focuses on a comparative study of field distribution
properties of three types of oil (coconut oil, peanut oil, sunflower oil) with different
gap spacing values. The three types of oil have different conductivity and permit-
tivity values. The electrodes are immersed in oils. The electrodes are made of copper
material. The arc length versus electric potential graph is obtained. Using compara-
tive study analysis we find out the oil which has the best field distribution properties.
Electrical breakdown or dielectric breakdown is a cycle that happens when an elec-
trical protecting material, exposed to a sufficiently high voltage, out of nowhere turns
into an electrical channel and electric flow moves through it. All protecting materials
go through a breakdown when the electric field brought about by an applied voltage
surpasses the material’s dielectric quality. The voltage at which a given protecting
article becomes conductive is called its breakdown voltage and relies upon its size
and shape. Under adequate electrical potential, an electrical breakdown can happen
inside solids, fluids, gases, or vacuum. Notwithstanding, the particular breakdown
components are diverse for every sort of dielectric medium. The electrical breakdown
might be a flashing occasion (as in an electrostatic release) or may prompt a persistent
electric curve if defensive gadgets neglect to intrude on the flow in a force circuit.
For this situation, an electrical breakdown can cause calamitous disappointment of
electrical hardware, and fire risks.
Keywords Natural ester · Electrical Field Distribution · Dielectric medium · Arc

length · Electric Potential
M. Divya Bharathi (B) · R. V. Maheswari

Department of Electrical and Electronics Engineering, National Engineering College, Kovilpatti,
India
S. Senthil Kumar
Department of Electrical and Electronics Engineering, New Prince Shri Bhavani College of
Engineering and Technology, Chennai 600073, India
https://doi.org/10.1007/978-981-16-4943-1_28
304 M. Divya Bharathi et al.
1 Introduction
The transformer assumes a significant function in giving a solid what’s more, effective
power flexibly and is one of the most basic types of gear in electric force transmission
and conveyance frameworks [1, 2].
Most of the high voltage transformers are loaded up with fluid that functions as
electrical protection just as a warmth move medium. The most ordinarily utilized
fluid in power transformers is mineral oil because of its ease and great properties [3,
5]. Anyway, the execution of mineral oil begins to be restricted because of ecological
thought. Initially, traditional transformer oils are generally non-biodegradable; they
can pollute soil and water at the point when a genuine spill happens [4, 6].
This may upset the manor and different creatures. Besides, the mineral oils were
separated from the oil, which will run out later on since oil is non-renewable [5, 7].
Normal ester protecting liquid offers fire security, climate, also, protection maturing
points of interest over mineral oil and are discovered to be appropriate for the utiliza-
tion in transformer protection framework. Past fixed cylinder maturing contemplates
showing that the warm maturing pace of paper protection in regular ester is funda-
mentally more slow than that in mineral oil. These two characteristic esters have
presently been utilized in little force and appropriation transformers over the US,
and further upgrades are progressing in the trust that they will be generally applied
in enormous force transformers [8, 9]. Moore introduced the prerequisites and desires
for regular ester liquids for application in power transformers. It has become evident
that more examination is needed to learn the drawn-out safe activity of transformers
where the common ester is utilized.
2 Methodology
Analyze the field distribution properties of three kinds of oil (ex. coconut oil (satu-
rated oil), sunflower oil (unsaturated oil), peanut oil (monosaturated oil) using
COMSOL multiphysics software.
In COMSOL the HV electrode applied potential (50, 70, 80 and 100 V). The
ground electrode was at 0 potential. The electrodes are immersed in these three types
of oil. The distance between the electrodes is varied in the form of 2, 4, 6 and 10 cm.
The electrodes are made of copper material. The conductivity and permittivity values
of the three types of oil are shown in Tables 1 and 2.
Table 1 Oil conductivity

S. No. Oil Conductivity
values
1 Coconut oil 2.09*10−7 −1 m−1
2 Sunflower oil 2.48*10−7 −1 m−1
3 Groundnut oil 1.74*10−7 −1 m−1
A Comparative Study of Field Distribution … 305
Table 2 Oil permittivity

S. No. Oil Permittivity
values
1 Coconut oil 2.9
2 Sunflower oil 3.35
3 Groundnut oil 2.64
Fig. 1 Coconut oil with a

point to point Electrode in
2 cm gap spacing
The values of permittivity of three kinds of oil are shown in Table 2.
This paper deals with the field distribution of three types of oil with different voltage
and different gap spacing. The electrode configuration is the point to point, sphere
to sphere, plane to plane, plate to plate.
3.1 Coconut Oil (Point to Point Electrode Configuration)
Here the field distribution of coconut oil with a point to point electrode configuration
using different voltage and different gap spacing results using COMSOL is given
Fig. 1.
In Fig. 2a the x-axis is arc length and y-axis is electric potential. The point to point
50 V 2 cm gap spacing has the lowest arc length versus electric potential curve. The
point to point 100 V 10 cm gap spacing curve has the highest magnitude in this arc
length versus electric potential graph.
3.2 Coconut Oil (Sphere to Sphere Configuration)
In Fig. 2b sphere to sphere electrode configuration, the all curves starting point has
applied voltage. Then time elapses the value of arc length versus electric potential
Fig. 2 Simulation result. a Coconut oil point to point configuration, b coconut oil sphere to sphere,
c coconut oil plane to plane, d coconut oil plate to plate, e sunflower oil point to point, f sunflower
oil sphere to sphere
of all these curves drastically down to 0. Then all the curves are rise up to the given
potential of that considered voltage.
Dielectric breakdown is additionally significant in the plan of coordinated circuits
and other strong state electronic gadgets. Protecting layers in such gadgets are
intended to withstand ordinary working voltages, however higher voltage, for
example, from electricity produced via friction may demolish these layers, deliv-
ering a gadget pointless. The dielectric quality of capacitors restricts how much
energy can be put away and the sheltered working voltage for the gadget.
3.3 Coconut Oil Plane to Plane Electrode
In Fig. 2c all these curves start from 0 electric potentials then all these curves are
increased up to consider electrode given potential then all the curves are linearly
decreasing to 0 electric potentials. The gap spacing is increased then the electric
potential decreases.
Utilizing present-day limited component procedures, the electric fields created
by anodes can be precisely determined. This permits plans to be tried and improved
before produce. Extraction and pillar speeding up cathodes can be demonstrated and
reproduced radiates followed through them.
3.4 Coconut Oil Plate to the Plate Electrode
In Fig. 2d the arc length versus electric potential curves starts from 0 then it was
increasing linearly then reach the particular electrode configuration given electric
potential. Then the voltage is stable at a particular time. Then the electric potential
is linearly reduced to 0. Corona shields are a kind of field molding cathode used to
shield sharp focuses that would somehow experience the ill effects of the corona.
The sharp focuses are generally unavoidable parts like jolts or different installations
and fittings.
A corona shield builds the range of these focuses by in a real sense concealing
them with adjusted terminals that are electrically associated so they are successful
at a similar potential. Different names for corona shields are monitor rings and
stress rings. Faltering is the place where particles assault cathode surfaces, thumping
molecules off the surface. This cycle causes the disintegration of the terminal surface.
Cathodes in plasmas experience the ill effects of faltering.
3.5 Sunflower Oil Point to Point
In Fig. 2e the high voltage electrode the voltages are applied is 50, 70, 80 and 100 V.
The gap spacing is increased linearly. The arc length versus electric potential graph
has these characteristics point to point 50 V 2 cm < point to point 70 V 4 cm < point
to point 80 V 6 cm < point to point 100 V 10 cm.
3.6 Sunflower Oil Sphere to Sphere
In Fig. 2f at 17–18 cm arc length, the sphere to sphere 100 V 10 cm gap spacing has
0 electric potential then the configuration has linearly increases and reaches 100 V.
In sphere to sphere 50 V 2 cm configuration at arc length 7 cm to 11 cm, the value
of electric potential is 0 V. Then that curve increases and reaches 50 V.
The plan and manufacture of electrodes for direct current (dc) high voltage photoe-
mission electron firearms can fundamentally impact their presentation, most promi-
nently regarding the greatest attainable inclination voltage. Legitimate electrostatic
plan of the triple-point intersection shield anode limits the danger of electrical break-
down (arcing) along with the protector link plug interface, while the cathode shape
is intended to keep up <10 MV/m at the ideal working voltage focusing on almost
no field discharge once molded.
3.7 Sunflower Oil Plane to Plane
In Fig. 3a the plane to plane 80 V 6 cm electrode configuration 10–12 cm arc length

the electric potential is 80 V then that curve decreases linearly. In this graph plane
to plane 70 V 4 cm electrode configuration at 2 cm arc length, the electric potential
is 13 V. At 3 cm arc length, the electric potential is 25 V. At 9–11 cm arc length the
electric potential remains 70 V then the electric potential decreases depending upon
arc length.
3.8 Sunflower Oil Plate to Plate
In Fig. 3b plate to plate 50 V 2 cm electrode configuration 8–15 cm arc length, the

electric potential is maximum almost close to 50 V. Then it linearly decreases. At
19.93 cm arc length, the electric potential is 0 V. The plate to plate 100 V 10 cm
electrode configuration 13–23 cm arc length the electric potential is near to 100 V.
Then the curve linearly decreases.
Fig. 3 Simulation result. a sunflower oil plane to plane, b sunflower oil plate to plate, c peanut oil
point to point, d peanut oil sphere to sphere, e peanut oil plane to plane, f peanut oil plate to plate
The shape and material of the high-voltage electrode of an air pressure plasma
age framework were advanced. The exploration was performed to accomplish the
greatest consistency of plasma treatment of the outside of the low-voltage cathode
with a measurement of 100 mm. To create low-temperature plasma with a volume of
around 1 cubic decimetre, a beat volume release was utilized started with a corona
release. The consistency of the plasma in the district of the low-voltage cathode was
evaluated utilizing a framework for estimating the appropriation of release current
thickness. The framework’s low-voltage Electrode-gatherer—was a plate of 100 mm
in distance across, the directing surface of which was isolated into 64 radially found
sections of the equivalent surface region. The current at each portion was enlisted by
a fast estimating framework constrained by an ARM-based 32-digit microcontroller.
To encourage the understanding of results got, a PC program was created to imagine
the outcomes. The program gives a 3D picture of the current thickness circulation
on the outside of the low-voltage cathode.
3.9 Peanut Oil Point to Point
In Fig. 3c point to point 70 V 4 cm electrode configuration the maximum voltage

almost equal to 70 V occurring at the arc length value of 7–11 cm. In point to point
80 V 6 cm electrode configuration. The maximum voltage almost 80 V occurring at
10–12 cm arc length. Then the curve linearly decreases.
3.9.1 Peanut Oil Sphere to Sphere
In Fig. 3d sphere to sphere 50 V 2 cm electrode configuration 1 and 14 cm arc length

electric potential is maximum between 1 and 14 cm arc length points the electric
potential value decreases. Sphere to sphere 100 V 10 cm electrode configuration the
maximum electric potential occurring at 2 and 30 cm arc length. The response can
be begun and halted by interfacing or detaching the two anodes. On the off chance
that we place a variable opposition in the circuit, we can even control the pace of
the net cell response by basically turning a handle. By interfacing a battery or other
wellspring of current to the two anodes, we can drive the response to continue in
its non-unconstrained, or invert course. By putting an ammeter in the outer circuit,
we can quantify the measure of electric charge that goes through the terminals, and
consequently the number of moles of reactants that get changed into items in the cell
response.
3.9.2 Peanut Oil Plane to Plane
In Fig. 3e plane to plane 70 V 4 cm electrode configuration the maximum electric

potential of 70 V occurring at the arc length value of 7–10 cm arc length. Plane
to plane 80 V 6 cm electrode configuration the maximum electric potential 80 V

occurring at the arc length value of 11–14 cm arc length.
In most electrochemical tests our premium is focused on just one of the anode
responses. Since all estimations must be on a total cell including two anode frame-
works, it is normal practice to utilize a reference terminal as the other portion of the
cell. The significant prerequisites of a reference cathode are that it be anything but
difficult to get ready and keep up, and that it’s likely to be steady.
3.9.3 Peanut Oil Plate to Plate
In Fig. 3f plate to plate 50 V 2 cm electrode configuration the maximum voltage of

50 V occurring at the arc length value of 14.92 cm. In plate to plate 100 V 10 cm
electrode configuration the maximum voltage of 100 V occurring at the arc length
values of 13–22 cm arc length.
A galvanic cell (here and there more suitably called a voltaic cell) comprises two
half-cells joined by a salt extension or some other way that permits particles to pass
between the different sides to look after electroneutrality. The customary method
of speaking to an electrochemical cell of any sort is to compose the oxidation half
response on the left and the decrease on the right.
4 Conclusion
A comparative study using three kinds of oil with different conductivity and permit-
tivity and find out the field distribution of those oil using comsol software. From
these results, coconut oil has the highest field distribution results compared to peanut
oil and sunflower oil. Peanut oil has followed after coconut oil. The sunflower oil has
the least field distribution values. So the result is coconut oil > peanut oil > sunflower
oil.
Acknowledgements The authors are also grateful to the National Engineering College, Kovilpatti,
Tamilnadu, India, for their support.
References
1. Cui Y, Ma H, Saha T, Ekanayake C, Martin D (2016) Particle tracing modelling on moisture

dynamics of oil-impregnated transformer. IET Sci Meas Technol 10(4):335–343
2. Madavan R, Balaraman S (2017) Comparison of antioxidant influence on mineral oil and natural
ester properties under accelerated aging conditions. IEEE Trans Dielectr Electr Insul 24(5):2800–
2808. https://doi.org/10.1109/TDEI.2017.006527
3. Liao R, Lin Y, Zhang Y, Xia H (2016) Independent effects of aged oil and aged paper on moisture
evaluation of power transformers. Electr Power Components Syst 44(5):556–564
4. Mariprasath TS, Victor K (2015) Pongamia Pinnata as alternate liquid dielectrics in distribution
transformer: a critical study on the property of viscosity. Adv Energ Power 3:1–7. https://doi.
org/10.13189/aep.2015.030101
5. Rao UM, Sood YR, Jarial RK (2016) Performance analysis of alternate liquid dielectrics for
power transformers. IEEE Trans Dielectr Electr Insul 23(4):2475–2484. https://doi.org/10.1109/
TDEI.2016.7556527
6. Al-Amin H, O’Brien J, Lashbrook M (2013) Synthetic ester transformer fluid: a total solution
to windpark transformer technology. Renew Energ 49
7. Bakrutheen M, Karthik R, Madavan R (2013) Investigation of critical parameters of insulating
mineral oil using semiconductive nanoparticles. In: Proceedings of IEEE international confer-
ence on circuit, power and computing technologies, ICCPCT 2013, pp 294–299. https://doi.org/
10.1109/ICCPCT.2013.6529039
8. Thien YV, Azis N, Jasni J, Kadir Z, Yunus R, Ishak M, Yaakub Z (2016) The effect of polarity
on the lightning breakdown voltages of palm oil and coconut oil under a non-uniform field for
transformers application. Ind Crops Prod 89:250–256. https://doi.org/10.1016/j.indcrop.2016.
04.061
9. Mehta DM, Kundu P, Chowdhury A, Lakhiani VK, Jhala AS (2016) A review of critical evalu-
ation of natural ester vis-a-vis mineral oil insulating liquid for use in transformers: Part II. IEEE
Trans Dielectr Electr Insul 23(3):1705–1712. https://doi.org/10.1109/TDEI.2016.005371
Managing the Smart Grid with Demand
Side Management Using AntLion
Optimization
Banala Venkatesh and S. Padmini
Abstract The demand Side Management (DSM) considered as the utmost key
strategy applied to a Smart Grid (SG), which prepares consumers to yield accu-
rate results regarding their consumption of electrical energy and supports the utility
to curtail the peak load demand and restructure the demand curve. In this paper, a
load shifting strategy has been applied for the DSM with huge number of controllable
devices. To curtail the peak demand, decrease the electricity bill and to minimize the
PAR, the load-shifting problem has been approached in hourly manner, throughout
the 24 h in a day. Antlion Optimization (ALO) algorithm has been used to accom-
plish the objective of load shifting through minimization problem to the DSM and
has been employed in residential loads of SG.
Keywords Smart Grid · Demand side management · Appliance scheduling · ALO

algorithm · Load shifting
1 Introduction
In the modern age, things surrounding are getting smarter with respect to time. In
near prospect, all our traditional power girds will convert to smart grids. The demand
of electricity growing day by day will provide opportunities for future innovation
to design our system very much effective, capable and smart. The main theme of
converting the traditional grid to smart grid is to raise demand in calculation, comput-
erization and communication to maintain the network by monitoring and controlling
with the minimal manual interference.
The main features of the smart grid are capability, consistency; effectiveness,
sustainability and consumer participation. Huge size grids has been mandatory for
improved consistency. With the better quality, sustainability of assets beside with
B. Venkatesh (B) · S. Padmini

SRM Institute of Science and Technology, Kattankulathur, Chengalpattu, India
S. Padmini
https://doi.org/10.1007/978-981-16-4943-1_29
314 B. Venkatesh and S. Padmini
consumer participation practical capability can be enlarged by means of Demand side

management (DSM) [1]. The adjustment of customer demand for Electrical energy
over different methods such as economic encouragements and interactive alteration
over teaching is called DSM [2]. In DSM, customers are motivated to alter its regular
electricity consumption plan on the source of power saving techniques, electric billing
and economic encouragements [3]. So a practical generation capability has made
and which reduces the dependency for large production. By applying demand side
management, carbon emissions will get decreased by less utilization of peak load
Generating plants, for example Diesel generators in peak load hours. Peak load hour
and off-peak load hour are the two different kinds of energy demand. Peak load hours
are from 11 am to 3 pm and from 6 to 9 pm and the remaining time is off peak load
hours. Excess generation must be provide in order to meet the peak demand. DSM
is the best option to reduce the necessity of more generation in the peak load hours.
The various techniques are used to apply DSM for the smart grid system as
shown in Fig. 1. DSM is applied to various functions of SG in areas of electrical
power market, infrastructure construction and electrical vehicles, by this way a DSM
leads to most important parameter in SG [4]. By controlling and manipulating the
energy demand, modified demand profile reduced maximum load demand can be
obtained, that leads to the better stability of the system less cost of generation and
less environment pollution. Energy utilization of customer is affected by penalties
and emoluments in perfect levels of supply chain by means of smart tariff with
Fig. 1 Techniques of demand side management

Managing the Smart Grid with Demand … 315
DSM. However, DSM provides an opportunity to enhance the system performance,

stability and security by improving the present infrastructure capability by helping
to minimize the carbon technology throughout the system by playing an important
role.
DSM helps in improving load factor that is accountable for maintain low elec-
tricity prices [5]. For reducing the highest demand through load shifting schemes are
used which takes the advantage of loads altering from high load hours to off high
load hours, strategic conversion is a non-traditional method which reduce the total
consumption after the load shape changes at consumers. In recent years, Several
DSM schemes and algorithms are evolved [6–9]. Traditional methods are dynamic
programing and linear programming are mostly used. A huge variety of devices along
with several metaheuristic methods, such as computational methods could not tackle
such a huge number of controllable devices.
In this manuscript, we have articulated the problems of reduction of the peak load,
electricity bill and Peak to average ratio, which will be conversed in later sections.
A related method has been discoursed in paper offered by Logenthiran [10]. In this
paper, an advanced method is proposed using Antlion optimization technique (ALO)
for solving the DSM problem. The data of load and devices of smart grid, mentioned
in paper [10] has been taken.
2 Problem Formulation
2.1 Minimization Function
The primary gaol in the DSM is to make the initial load curve as similar to our
objective load curve. The modification can be done by applying DSM Techniques.
Let us take the initial load curve as loadin . Let us assume the altered load curve
obtained as loadfi by applying the techniques of DSM [11]. So, Then our objective
can be framed as

minimization F1 : Max Load f i (1)

h

minimization F2 : Load f i h .E h (2)
H =1
minimization F3 : Min(PAR) (3)
From the above functions, F 1 represents the highest peak value of final load curve,
that we want to be minimized laterally with the function F2 denoting the final cost
of energy over a Twenty fourhour duration interval and E h is the electricity price in
hth hour and Function F 3 denotes the peak to average ratio of the system.
From the above Eqs. 1 and 2 the final load curve is obtained as by estimation the
load alteration as function F h
min Fh : [|RLMh | − |Loadh |] (4)
Subject to for all h = 1, 2, 3, …, N. where N is the time step number in hour wise
block.
The estimated load correction for each time step hour h can be obtained from
RLM i.e., Reducible load Margin, which is given by
RLMh = Forecasth − objh (5)
Equation (5) governs the estimated load, which desires to be remove fully or
partially or added to the initial load curve at various hour slots that makes it as close
as to objective load curve.

≥ 0 Forecasth > objh
RLMh = (6)
≤ 0 Forecasth < objh
The minimization of Peak to Average ratio leads to improvement in system

performance and system stability, it is expressed as
Loadmax Loadmax
PAR = = Load (7)
Loadmean n
N
Here, total hourly time slot is denoted as N , peak load is denoted as Loadmax ,
Average value of the load is represented as Loadmean .
2.2 Constraints
The following are the constraints developed for the above minimization problem.
• The type of the shifted device should be greater than zero always in all time slots
[11]
xkh > 0 ∀k, h (8)
• The available number of controllable devices of that particular must always be

less than the number of devices to be shifted

h
xk ≤ controllable(k) (9)
k=1
2.3 Calculation of Controllable Devices
There are two types of controllable devices are there in the DSM scheme which is
a most important thing. One is Non-shiftable devices, those are fixed to run under
defined time slot e.g. refrigerator, T.V. etc. Another one is Shiftable devices, which
are flexible to run under any time slot e.g. washing machine, dryer, dishwasher etc.
in this work the shiftable devices from residential area of the smart grid test system is
considered for solving the DSM Problem. The devices setup in the residential areas
for controlling the load typically consumes less amount of power and the running
time slot is very slow. In residential area, 2604 manageable devices contains of 14
various devices are installed [10].
3 AntLion Optimization (ALO)
Here antlion method is taken which is referred from the paper [12]. The technique
ALO is a newly designed Evolutionary metaheuristic searching method, which
imitates the antlions hunting method in nature. By this method, ant and antlions
defined by means of searching managers are represented to discover answers by
stages of capturing the target prey, which contains the arbitrary walking direction of
ants, construction of setups, making the ants to fall in trap, capturing the target prey,
and reconstruction the traps.
Ants passage randomly in Wildlife while hunting for nutrition, so at each step an
arbitrary walk by an ant is represented as
N −1

N
Z k = [0; p(1) + p(2); . . . ; p( j); p( j) (10)
j=1 j=1
where k = 1, …, dimen, dimen is the dimension of ant, N is the iteration number, Z

= [Z 1 ; …; Z dimen ], Z k is a (N + 1) × 1 matrix, and p(j) is a Random function and
can be expressed as:
1 if rand > 0.5

p= (11)
−1 if rand < 0.5
Here rand is a arbitrary number produced by even spreading among 0 to 1.

The walking of ants has to be altered to the accurate position in actual search space
permitting to the limited boundaries. Position can be calculated using Eq. (12):

Z k − lk
Yk = ∗ (ok − n k ) + n k (12)
m k − lk
where l k and mk is the min and max values of Z k , nk , and ok denotes the min and
max of antlion in the kth dimension respectively, Y = [Y 1 ; …; Y dim ],
Y k is a (N + 1) × 1 matrix. Z k is normalized in range of [0, 1] using Z k −lk /Z k − m k ,
then it is converted into domain [ni , oi ] using Eq. 12.
The ants’ movements are affected by antlions’ traps. This can be described as:
n = n 0 + Antlion, o = o0 + Antlion (13)
where n’ and o’ is the min and max of varying boundary at current iteration,
In the ALO algorithm, Elitism procedure is used. That means that a Global best
antlion is chosen as elite antlion during process of optimization.
Random walking of the ant will be governed by the position update of each ant,
which is selected by Roulette wheel and elite. It can be found as:
R A + RE
Ant = (14)
2
where Ant is the novel location, RA is the Roulette wheel nominated antlion, RE is
elite antlion. Updated position of ant must be change if the ant crosses its boundaries.
The antlion should takes its position when the ant grasps the quarry and is greater
than the antlion. This procedure is known as prey catching, and it is given as:
Antlion = Ant, if f (Ant) < f (Antlion) (15)
4 Test Smart Grid System
To validate the efficiency of the suggested method, the proposed DSM strategy is
applied on residential area smart grid. Figure 2 shows the Utility of the smart grid.
The operating voltage of electrical network is 400 V. The connection among the
two grids has a 0.003 pu resistance, 0.01 pu. reactance. The residential area system
microgrid length is 2 km. This network data is for utilized for planning of resources
without any congestion (Table 1).
The main goal of DSM is decrease the utility bills of customers in all the three
areas. Therefore, a nonlinear relationship should be there between objective load
curve and electricity tariff. The equal electricity tariffs are applicable to all areas in
the smart grid. Simulations were carried out with a maximum allowable delay of 12
h. The performance of the algorithm would be better if there is a longer delay, since
the number of loads related to load shifting increases, leads to improved results.
Fig. 2 Smart grid network
Table 1 Data for the residential smart grid area

Name of the device Hourwise power utilization of each Device (kW)
1st hour 2nd hour 3rd hour
Dryer 1.2
Dish washer 0.7
Washing machine 0.5 0.4
Oven 1.3
Iron 1.0
Vaccum 0.4
Fan 0.20 0.20 0.20
Kettle 2.0
Toaster 0.9
Rice cooker 0.85
Hair dryer 1.5
Blender 0.3
ALO Technique has been applied for the residential area for the DSM problem in
Sect. 3. Different set of devices are there in each section with various features like
different operating hours, schedule and power consumption.
Figure 3 Show the power analysis of demand for 24 h duration. From the graph the
maximum energy consumption by the device is 2000 W and minimum consumption
of the device is 40 W. Figure 4 shows the power consumption analysis of various
devices (Table 2).
Fig. 3 Demand analysis
Fig. 4 Individual power analysis

Table 2 Maximum load and electricity bill reduction

Technique Maximum Maximum Percentage Electricity Electricity Percentage
load without load with of reduction bill without bill DSM of reduction
DSM (KW) DSM (KW) (%) DSM (Rupees) (%)
(rupees)
PSO 1540 1334.1 15.4 3080 2668.2 15.4
ALO 1540 1309.4 17.64 3080 2618.8 17.64
6 Conclusion
DSM technique in SG is presented in this paper with ALO technique. DSM has
gaining a lot of attention, because of its benefits in smart grid. During peak hours,
extra demand of power gets reduces, reduces the electricity bill of the customers and
reduction in the PAR value. In this paper, ALO algorithm has been adopted which
obtains a better result compared to PSO and reduces the peak demand, substantial
savings in energy bill and PAR value, carried on residential loads of the smart grid.
References
1. Guelpa E et al (2019) Demand side management in district heating networks: a real application.
Energy 182:433–442
2. Tang R, Wang S, Li H (2019) Game theory based interactive demand side management
responding to dynamic pricing in price-based demand response of smart grid. Appl Energ
250:118–130
3. Khan A, Javaid N, Ahmad A, Akbar M, Khan Z, Ilahi M (2018) A priority-induced demand
side management system to mitigate rebound peaks using multiple knapsack. J Ambient Intell
Humanized Comput 10(4):1655–1678
4. Su H, Zio E, Zhang J, Chi L, Li X, Zhang Z (2019) A systematic data-driven Demand Side
Management method for smart natural gas supply systems. Energ Convers Manag 185:368–383
5. Kumar K, Saravanan B (2019) Day ahead scheduling of generation and storage in a microgrid
considering demand Side management. J Energ Storage 21:78–86
6. Yilmaz S, Chambers J, Patel M (2019) Comparison of clustering approaches for domestic
electricity load profile characterisation—implications for demand side management. Energy
180:665–677
7. Walzberg J, Dandres T, Merveille N, Cheriet M, Samson R (2019) Accounting for fluctuating
demand in the life cycle assessments of residential electricity consumption and demand-side
management strategies. J Cleaner Prod 240:118–251
8. Luo X, Fong K (2019) Development of integrated demand and supply side management strategy
of multi-energy system for residential building application. Appl Energ 242:570–587
9. Peltokorpi A, Talmar M, Castrén K, Holmström J (2019) Designing an organizational system for
economically sustainable demand-side management in district heating and cooling. J Cleaner
Prod 219:433–442
10. Logenthiran T, Srinivasan D, Shun T (2019) Demand side management in smart grid using
heuristic optimization. IEEE Trans Smart Grid 3(3):1244–1252
11. Gupta I, Anandhini GN, Gupta M (2016) An hour wise device scheduling approach for demand
side management in smart grid using Particle swarm optimization. In: National power systems
conference. https://doi.org/10.1109/npsc.2016.7858965
12. Mirjalili S (2015) The ant lion optimizer. Adv Eng Softw 83:80–98
Comparison of PI Based and ANN Based
Dynamic Voltage Restorer Controller
for Voltage Sag Mitigation in Distribution
System
T. Jane Tracy and N. Rathina Prabha
Abstract In recent years, one of the major concerns in the distribution system is
the quality of power at the consumer side. Out of all power quality issues, voltage
sag is the most frequent one. The Dynamic Voltage Restorer (DVR) is one of the
effective ways for protecting sensitive loads from Voltage sag/swell conditions. In
this paper design and analysis of DVR for mitigating voltage sag is done by means
of MATLAB/SIMULINK. The results of conventional DVR controller using PI are
compared with Artificial Neural Network (ANN).
Keywords Artificial Neural Network (ANN) · Custom Power System Devices

(CUPS) · Dynamic Voltage Restorer (DVR) · Total Harmonic Distortion (THD)
1 Introduction
Nowadays the interest on power quality has increased very high. This is mainly due
the problems created by the poor power quality by the consumers and the utilities [1].
Among the entire power quality disturbance Voltage Sag is considered as the most
frequently occurring one in a distribution system. The major features that categorize
the voltage sag are its magnitude and duration of occurrence.
There are many custom power devices used to mitigate the voltage sag. Among
this DVR is considered as the most effective way to protect the sensitive loads in a
power system from voltage sag and swell [4, 5]. DVR is the series connected custom
power device used for the injection of required voltage in series to the system to
regulate the load.
T. Jane Tracy (B) · N. Rathina Prabha

Department of Electrical and Electronics Engineering, Mepco Schlenk Engineering College,
Sivakasi, India
N. Rathina Prabha
https://doi.org/10.1007/978-981-16-4943-1_30
324 T. Jane Tracy and N. Rathina Prabha
2 Analysis of Voltage Sag
Voltage sag indices are experienced by customers when they share a common supply
impedance with an over current event on the supply system [7]. The voltage dip must
be quantified using some of the indices. Some of the important indices are as follows.
2.1 Detroit Edison Sag Score (SS)
The Detroit Edison sag score is defined as follows
Va + Vb + Vc
S= (1)
3
where V a , V b, V c are the RMS value of the phase voltage in terms of per unit. This
method does not consider the time duration of voltage sag.
2.2 Voltage Sag Lost Energy Index (VSLEI)
When there is a three phase fault in a distribution system the Voltage Sag Lost Energy
Index can be given as [10].

Va 3.14 Vb 3.14 Vc 3.14
W = 1− × Ta + 1 − × Tb + 1 − × Tc (2)
Vnom Vnom Vnom
where V a, V b, V c are the voltages in three different phases. The value of T a , T b , T c

represent the time duration of sag for all the three phases.
3 Schematic Diagram of Dynamic Voltage Restorer
Figure 1 shows the schematic diagram of DVR. The DVR injects the missing voltage
caused due to the disturbance, faults or any problem caused in the distribution system.
The components of DVR are as follows [4]:
Comparison of PI Based and ANN Based … 325
Fig. 1 Schematic diagram of DVR
3.1 DC Energy Storage Device
The source for the VSI is given from a DC energy storage device. These energy
storage devices are used to supply the required real power when DVR is providing
compensation.
3.2 Control Circuit
The control circuit for DVR compares the load voltage with the reference value to
produce the control signal for the PWM generator [6]. This helps to maintain the
voltage magnitude at the point of common coupling.
3.3 Voltage Source Inverter (VSI)
The purpose of the Voltage Source Inverter is to convert the DC voltage available
from the energy storage device into AC voltage based on the control circuit output.
3.4 Passive Filter Unit
The main purpose of the passive filter circuit is to filter out the harmonics present in
the voltage generated from the VSI.
3.5 Series Injection Transformer
The series injection transformer is particularly designed to inject the required

compensation voltage to the sensitive load side. The main functions of the series
injection transformer are coupling, protection and voltage injection.
4 Controlling Method of DVR
The control circuit in a DVR plays a vital role. It gives a fast response during voltage
sags or variation in load [12]. The control of DVR is done by detecting, measuring,
synchronizing and performing transformation. This is done to do the voltage compen-
sation during any voltage sag condition. In this paper a feed forward based control
with Phase Locked Loop (PLL) is used. The load voltage given to the sensitive load
is measured. The three phase abc voltage is converted into dq0 voltage by means of
Park transformation. The PLL is used to synchronize the load voltage with the refer-
ence voltage and helps to derive the unit vectors sin θ and cos θ. The transformation
is given by the equation
⎡ ⎤ ⎡ ⎤⎡ ⎤
Vd cos θ cos θ − 2π3
1 Va
⎣ Vq ⎦ = ⎣ − sin θ − sin θ − 2π 1 ⎦⎣ Vb ⎦ (3)
3
1 1 1
V0 2 2 2
Vc
The reference signal is compared with the load voltage so as to obtain the input
signal for the controller. The equation is given by
⎡ ⎤ ⎡ ⎤⎡ ⎤
VDVRa cos θ 2π sin θ 2π 1 Vd
⎣ VDVRb ⎦ = ⎣ cos θ − sinθ − 3 1 ⎦ ⎣ Vq ⎦ (4)
3
VDVRc cos θ + 3 sin θ + 3 1
2π 2π
V0
Now the generated three phase voltage is used to produce the PWM modulation
signal. This produces the required sinusoidal voltage from the voltage source inverter.
This will help in the improvement of the voltage profile into the nominal load voltage.
5 Simulation and Results
The performance of Dynamic Voltage Restorer for the different fault condition is
analyzed by the simulink model shown in Fig. 2.
In Fig. 2, 11 kV source is connected to the load via a transformer. The DVR is
connected at the point of common coupling. The specifications of the given system
are as shown in Table 1.
Fig. 2 Matlab simulink model of DVR for protection of sensitive load
Table 1 System parameters

Parameter Value
Source 11 kV, 30 MVA
Transformer 11 kV/0.4 kV, 1MVA
Parallel RLC Load 400 V, 4.42 kV
DC Supply 500 V
Series injection transformer 10 V/100 V, 1.5 kVA
Line impedance 1 , 1 mH, 20 µF
In this paper, sag for a time period of 100 ms is considered. Analysis and perfor-
mance of DVR for single line to ground fault, double line to ground fault and three
phase fault is done. The results are obtained for both PI based DVR controller and
Artificial Neural Network based DVR controller are obtained.
Figure 3 shows the voltage sag from t = 0.2 to t = 0.3 s. Here the voltage for
the different sections in the distribution system is shown. i.e. for the 11 kV line, for
400 V line and the sensitive load. The graph clearly shows that there is voltage sag
in the transmission line. But when it reaches the load terminal, the voltage remains
Fig. 3 Voltage sag for 100 ms

Table 2 Single line to ground fault

Sag% V rms (p.u) SS VSLEI Total Harmonic Distortion (THD)
VA VB VC V abc 11 kV V abc 400 V V abc load
10 0.94 1.01 0.95 0.972 0.016 2.62 1.68 0.84
25 0.87 1.01 0.90 0.927 0.239 7.10 4.30 0.84
50 0.74 1.01 0.82 0.858 1.959 16.48 9.51 1.02
75 0.61 1.07 0.75 0.804 6.602 29.19 15.7 1.16
85 0.57 1.04 0.73 0.777 8.917 35.53 18.6 1.25
90 0.95 1.03 0.71 0.759 10.89 39.01 30.1 1.21
Table 3 Double line to ground fault

Sag% V rms (p.u) SS VSLEI Total Harmonic Distortion (THD)
10 0.91 0.96 0.97 0.943 0.063 3.24 3.38 0.89
25 0.77 0.9 0.92 0.862 1.084 8.93 9.27 1.03
50 0.55 0.79 0.85 0.730 9.198 21.42 22.2 1.40
75 0.35 0.69 0.80 0.614 28.60 39.3 41.6 2.16
85 0.29 0.65 0.79 0.575 39.22 48.43 52.1 2.4
Table 4 Three phase to Ground Fault

Sag% V rms (p.u) SS VSLEI Total harmonic distortion (THD)
10 0.91 0.91 0.91 0.91 0.156 2.9 3.38 0.84
25 0.77 0.77 0.77 0.77 3.012 8.00 9.31 1.03
50 0.56 0.56 0.56 0.56 23.27 19.25 22.2 1.82
75 0.34 0.34 0.34 0.34 79.83 35.99 41.6 2.75
85 0.28 0.28 0.28 0.28 107.4 45.14 52.1 3.10
constant. This is due to the presence of DVR which injects the voltage required by
the load.
During the voltage sag from time t = 0.2 s to t = 0.3 s, the THD analysis is
done and the results show that the THD value is 13.66% for the source voltage but
it reduces to 1.94% for the load voltage due to the presence of DVR.
5.1 For PI Based DVR Controller
Tables 2, 3 and 4 show the RMS value of voltage for the three different phases for
different percentage of voltage sag. The THD value at the load side clearly indicates
that with the presence of Dynamic Voltage Restorer the harmonics has been greatly
reduced.
5.2 For ANN Based DVR Controller
The artificial neural network works on the three steps i.e. training, validation and
testing. The output obtained from the PI based controller is used to train the artificial
neural network and the output is obtained for the three above mentioned faults.
Tables 5, 6 and 7 give the RMS voltage for single phase to ground fault, double
line to ground fault and three phase to ground fault. The THD for load voltage is
measured. The results show that the THD is reduced greatly at the load side with the
help of DVR.
Table 5 Single line to ground fault

10 0.95 1 0.97 0.974 0.009 2.7 1.53 0.8
25 0.88 1 0.93 0.936 0.149 7.32 4.02 0.81
50 0.78 1 0.89 0.89 1.019 14.8 7.78 1.01
75 0.64 1 0.84 0.826 4.449 29.8 14.6 1.08
85 0.57 1 0.82 0.8 7.103 30.49 16.0 1.14
Table 6 Double line to ground fault

10 0.92 0.96 0.97 0.947 0.052 3.23 3.37 0.96
25 0.76 0.89 0.92 0.853 1.202 8.92 9.27 1.13
50 0.55 0.79 0.85 0.732 9.144 21.4 22.3 1.35
75 0.35 0.69 0.81 0.616 28.91 39.26 41.6 2.05
85 0.29 0.65 0.78 0.572 39.29 48.37 52.2 2.12
Table 7 Three phase to ground fault

10 0.91 0.91 0.91 0.91 0.140 2.9 3.37 0.85
25 0.76 0.76 0.76 0.76 3.221 7.99 9.27 1.42
50 0.54 0.54 0.54 0.54 26.01 19.23 22.3 1.72
75 0.34 0.34 0.34 0.34 81.76 35.98 41.6 2.50
85 0.29 0.29 0.29 0.29 104.1 45.10 52.2 2.77
6 Conclusion
The working of Dynamic Voltage Restorer for the mitigation of voltage sag in the
load side is analyzed in this paper. Here, the DVR is controlled by PI based controller
and ANN based controller. The results show that when the voltage sag is very low (i.e.
less than 50%) then the performance of PI based controller and ANN based controller
remains the same. But when the voltage sag is very high (i.e. greater than 50%) then
the performance of ANN is better when compared with PI based controller.
References
1. Wang ZQ, Zhou SZ, Guo YJ (2002) Comparisons on ways of magnitude characterization of
power quality disturbances. In: Proceedings of the 2002 large engineering systems conference
on power engineering. IEEE, 0-7803-7.520-3
2. Safdarian A, Fotuhi-Firuzabad M, Lehtonen M (2019) A general framework for voltage sag
performance analysis of distribution networks. Energies 12:2824
3. Wang B, Ye J, Manandhar U, Ukil A, Gooi HB (2017) A DC microgrid integrated dynamic
voltage restorer with model predictive control. 978-1-5386-0971-2/17 IEEE
4. Gayatri M, Parimi AM, Pavan Kumar AV (2015) Application of dynamic voltage restorer in micro
grid for voltage sag/swell mitigation. In: 2015 IEEE power, communication and information
technology conference (PCITC). https://doi.org/10.1109/PCITC.2015.7438096, October 2015
5. Benachaiba C, Ferdi B (2008) Voltage quality improvement using DVR. Electr Power Qual
Utilisation J XIV(1)
6. Raut PA, Kalgunde MN (2015) An overview and design of dynamic voltage restorer to improve
power quality in microgrid. In: 2015 international conference on energy systems and applications
(ICESA 2015). 978-1-4673-6817-9/15, 2015 IEEE
7. Babu BP, Indragandhi V (2019) A review on dynamic voltage restorer in power systems
concerned to the issues of power quality. IOP Conf Ser Mater Sci Eng. https://doi.org/10.1088/
1757-899X/623/1/012015, VCADPCA 2019
8. Hamoud F, Doumbia ML, Cherit A (2017) Voltage sag and swell mitigation using DSTATCOM in
renewable energy based distributed generation systems. In: 2017 twelfth international conference
on ecological vehicles and renewable energies (EVER). 978-1-5386-1692-5117, 2017
Design of Quadratic Boost Converter
for Renewable Applications
B. Abinayalakshmi, S. Muralidharan, and J. Gnanavadivel
Abstract This project work aims in exploring a quadratic boost converter which has
low voltage ripple achieved using extreme duty cycles and PWM strategy. Normal
quadratic boost converter topologies require a higher amount of stored energy by
available in the capacitors used. The energy stored capability is directly proportional
to the size of the capacitors. By using the minimum stored energy method, the
converter size reduces. Also this transformer less proposed topology is suitable for
renewable energy applications.
Keywords Pulse Width Modulation (PWM) · Direct current (DC) · Quadratic

boost converter (QBC) · Switching time (TS)
1 Introduction
In this paper, the non isolated power converter topology is used [2]. Low storing
energy capacitors cause reduced switching losses and increased efficiency [3]. The
non isolated power converter is a single circuit having the current flow between the
input and the output. Compared to the boost converter, the quadratic boost converters
provide double the level of the output voltage at the same duty ratio [5]. This converter
use the smaller size capacitor for the particular output voltage ripple specifications
[1–10]. Reduced voltage ripples content and improved power quality DC-DC QBC
converters [11–13].
B. Abinayalakshmi (B) · S. Muralidharan · J. Gnanavadivel

Mepco Schlenk Engineering College, Sivakasi, India
S. Muralidharan
https://doi.org/10.1007/978-981-16-4943-1_31
332 B. Abinayalakshmi et al.
Fig. 1 Proposed QBC

circuit diagram
2 Proposed QBC Circuit Diagram
At constant frequency, both the two transistors are commutating with the same duty
cycle ratio. The 180° phase shift occurs between both of them (Fig. 1).
3 Dynamic Model of Quadratic Boost Converter
The traditional state averaging technique applies to express the dynamic model of
this converter
topology.
This converter topology can be expressed as,
L 1 di L 1 /dt = dV g −(1 − d)V
c1 (i)
L 2 di L 2 /dt = d Vg + Vc1 −(1 − d)Vc2 (ii)
C1 dVc1 /dt = (1 − d)i L 1 −di L 2 −i 0 (iii)
C2 dVc2 /dt = (1 − d)i L 2 −i 0 (iv)
Vg, VC1 , VC2 —input voltage, voltage across the capacitors C 1 , C 2 respectively.
i L 1 , i L 2 —current flows through the L1 and the L2 inductors respectively.
i 0 , R—converter output current, load resistance respectively.
The output voltage and the output current can be express as,
V0 = Vg + Vc1 + Vc2 (1)
i 0 = V0 /R (2)
Design of Quadratic Boost Converter for Renewable … 333
Dvg
Vc1 = (3)
1− D
Vc2 = Dvg /(1 − D) (4)
I L 1 = I0 /(1 − D) (5)
I L 2 = I0 /(1 − D) (6)
Vc2 = DVg /(1 − D)2 (7)
4 Modes of Operation
MODE I
During this operation, the capacitor C 1 is discharging and the C 2 is charging. The
S 1 switch is on and S 2 switch is off (Fig. 2).
MODE II
During this operation, both the C 1 and the C 2 are discharging. Both S 1 and S 2 switches
are on.
MODE III
During this operation, both the C 1 and the C 2 are charging. Both S 1 and S 2 switches
are off.
MODE IV During this operation, the C 1 is charging and the C 2 is discharging. The
S 1 switch is off and S 2 switch is on.
4.1 Selections of Inductors
(1) L1 Inductor The L 1 voltage is equal to the input voltage, when the S 1 switch
is closed. During the time period of the DTs, the current rises with the Vg/L 1
slope and also total ripple increases twice of the L 1 .

L 1 = Vg /2i L 1 DTs
Fig. 2 Modes of operations
(2) L 2 Inductor The L 2 voltage is equal to the V C1 + Vg, When the switch S 2 is
closed. During the period of DTs, the current rises with the (V C1 + Vg)/L 2
slope.

L 2 = Vg /2i L 2 (1/1 − D)DTs
4.2 Selection of Capacitors
In this converter, the four possible switching states are occur at particular duty cycle.
During the switching period three switching states are involved. For our convenient
consider as (C 1 = C 2 ). 1. D > 0.5, 2. D < 0.5, 3. D = 0.5.
(a) During t 2 , the converter operation
V0 /t| t2 = −((I L 2 − I0 )/c2 ) − (I0 /c) = −((I L 2 − 2I0 )/c)
During t 2 = (1 − D) T s , Due to switching in t 2 , the output voltage can be

express as,
V0 /t| t2 = −((I L 2 − I0 )/c2 ) − (I0 /c)(1 − D)TS = −((I L 2 − 2I0 )/c)(1 − D)TS
(b) During t 1 and t 3 , the converter operation
V0 /t| t1 , t3 = −((I L 2 + I0 /c1 ) − (I0 /c2 ) = −((I L 2 + 2I0 /c)
Therefore, t 1 and t 3 = DTs − (1 − D) T s = (2D − 1)T s , when 180° phase shift

occurs
V0 /t| t1 , t3 = −((I L 2 + 2I0 /c)((2D − 1)TS /2
(c) During t 4 , the converter operation
V0 /t| t4 = −((I L 1 − I L 2 − I0 /c1 ) − (I0 /c2 )) = −(I L 1 − I L 1 − 2I0 )/c
V0 | t4 = −(I L 1 − I L 2 − I0 /c1 ) − (I0 /c2 )(1 − D)TS

= −((I L 1 − I L 1 − 2I0 )/c)(1 − D)TS
(d) During t 1 and t 3 , the converter operation
V0 /t| t1 , t3 = ((I L 1 − I0 /c1 ) + (I L 2 − I0 /c2 )) = (I L 1 + I L 1 − 2I0 )/c
V0 | t1 , t3 = ((I L 1 + I L 2 − 2I0 /c)(1 − 2D)Ts /2
5 Specifications
See Table 1.
Table 1 Design values of

Duty cycle I S (A)
proposed converters
Output power (Po ) 100 W
Duty cycle (d) 0.553
Output voltage (V o ) 100 V
Input voltage (V s ) 20 V
Proportional gain constant (K p ) 0.00001
Integral gain constant (K i ) 0.25
Resistance () 100
Line frequency 50 Hz
Switching frequency (f s ) 50 kHz
Inductor L 1 (µH) 368
Inductor L 2 (µH) 824
Inductor1 Current (IL1 ) 5
Inductor2 Current (IL2 ) 2.23
Capacitor C (µF) 10
6 Open Loop Analysis of QBC
See Fig. 3 and Tables 2, 3.
Fig. 3 Simulink diagram of open loop analysis of QBC

Table 2 Various duty cycle for open loop analysis of QBC

Duty cycle I S (A) V s (V) I 0 (A) V 0 (V) Efficiency
0.1 0.2940 20 0.2299 22.99 89.88
0.2 0.4851 20 0.2941 29.41 89.15
0.3 0.8429 20 0.3878 38.78 89.20
0.4 1.577 20 0.5318 53.18 89.66
0.5 3.271 20 0.7680 76.80 90.15
Table 3 % of load for open loop analysis of QBC

% of load Resistance I S (A) V s (V) I 0 (A) V 0 (V) Efficiency
25 400 1.567 20 0.2479 99.15 78.42
50 200 2.717 20 0.4834 96.68 86.00
75 133 3.920 20 0.7239 96.27 88.89
100 100 5.096 20 0.9587 95.87 90.17
7 Simulations of Closed Loop Analaysis of QBC
Figure 4 shows the simulation of closed loop analysis of quadratic boost converter.
Tables 4 and 5 shows the efficiency for various input voltage and various % of
load.
7.1 Output Voltage Waveform of Quadratic Boost Converter
See Fig. 5.
7.2 Output Current Waveform of Quadratic Boost Converter
See Fig. 6.
8 Conclusion
This project work was started with an objective of designing QBC with high power
factor and low voltage ripples. Reduction in output voltage ripple is obtained by using
the voltage ripple cancellation technique. The smaller capacitors help in achieving
the same voltage ripple specifications by using this converter. In quadratic boost
Fig. 4 Simulink diagram of closed loop analysis of QBC
Table 4 Various input voltage for closed loop analysis of QBC

I S (A) V s (V) I 0 (A) V 0 (V) Efficiency
14.60 10 1.008 100 68.49
8.064 15 1.008 100 83.33
5.943 20 1.008 100 84.80
4.674 25 1.008 100 86.26
3.856 30 1.008 100 87.13
Table 5 % of load for closed loop analysis of QBC

% of load Resistance I S (A) V s (V) I 0 (A) V 0 (V) Efficiency
25 400 1.854 20 0.2501 100 70.76
50 200 3.241 20 0.4997 100 77.09
75 133 4.593 20 0.7508 100 81.73
100 100 5.943 20 0.9981 100 83.97
Fig. 5 QBC output voltage waveform
Fig. 6 Output current waveform of QBC
converter, single switch with high voltage rating and the high current rating is
preferred. Thus the proposed QBC converter achieved the decreased output voltage
ripples and the increased efficiency.
References
1. Hwu KI, Peng TJ (2012) A novel buck–boost converter combining KY and buck converters.
IEEE Trans Power Electron 27(5)
2. Tofoli FL, de Castro Pereira D, de Paula WJ, de Sousa Oliveira D (2015) Survey on non-
isolated high-voltage step-up dc–dc topologies based on the boost converter. IET Inst Eng
Technol Power Electron 8(10)
3. Jiang W, Chincholkar SH, Chan C-Y, Peng TJ (2018) Modified voltage-mode controller for
the quadratic boost converter with improved output performance. IET Power Electron Inst Eng
Technol 11(14):11–27
4. Weng X, Xiao X, He W, Zhou Y, Shen Y, Zhao W, Zhao Z (2019) Comprehensive comparison
and analysis of non-inverting buck boost and conventional buck boost converters. IET Inst Eng
Technol Power Electron 2019(16):3030–3034
5. Chen Z, Yang P, Zhou G, Xu J, Chen ZY (2016) Variable duty cycle control for quadratic boost
PFC converter. Trans Ind Electron 63(7)
6. Morales-Saldana JA, Galarza-Quirino R, Leyva-Ramos J, Carbajal-Gutierrez EE, Ortiz-Lopez
MG, Xu D (2007) Multi loop controllers design for quadratic boost converter. IEEE Trans Ind
Electron 1(3)
7. Andrade AMSS, Hey HL, Schuch L, da Silva Martins ML (2018) Comparative evaluation of
single switch high-voltage step-up topologies based on boost and zeta PWM cells. IEEE Trans
Ind Electron 65(3)
8. Hanisah N, Baharudin, Mansur TMNT, Hamid FA, Ali R, Misrun MI (2017) Topologies of
dc-dc converter in solar PV applications, Indonesian J Electr Eng Comput Sci 8(2):368–374
9. Taghvaee MH, Radzi MAM, Moosavain SM, Hizam H, Hamiruce Marhaban M (2013) A
current and future study on non-isolated dc-dc converters for photovoltaic applications. Renew
Sustain Energ Rev 17(c):216–227
10. Ghamrawi A, Gaubert J, Mehdi D (2016) New control strategy for a quadratic boost converter
used in solar energy system. In: 2016 IEEE international energy conference (ENERGYCON),
Leuven, Belgium, pp 1-6.https://doi.org/10.1109/ENERGYCON.2016.7514125
11. Selvaperumal S, Muralidharan S, Nedumal Pugazhenthi P, Prabhakar G, Bastin Solai Nazaren
J, Parthiban PG, Karuppasamy P (2017) Performance investigation of SHE PWM implemen-
tation of GA based LCL resonant inverter in marine applications. Indian J Geo Marine Sci
46(09):1889–1898
12. Gnanavadivel J, Senthil Kumar N, Jaya Christa ST, Muralidharan S (2019) Design and Imple-
mentation of FPGA based high power LED lighting system for ships. Indian J Geo Marine Sci
48(05):724–729
13. Menaka S, Muralidharan S (2017) Novel symmetric and asymmetric multilevel inverter topolo-
gies with minimum number of switches for high voltage of electric ship propulsion system.
Indian J Geo Marine Sci 46(09):1920–1930
Recognition of Partial Discharge Signal
Using Deep Learning Algorithm
J. Ashmin Sugaji, M. Ravindran, and R. V. Maheswari
Abstract In transmission frameworks, insulators have a huge impact on the better

presentation of the devices. The outside insulators are introduced to an environment
that has a high temperature, clamminess similar to pollution from the beachfront and
industries. On deposition of contaminants, pollution builds gradually, and spillage
current starts to stream on a surface level. Partial Discharge (PD) deteriorates the
insulation and leads to the breakdown of the device. The effect of conductive pollu-
tion on PD is seen through tests performed on both earthenware and non-ceramic
protectors at different pollution levels. To achieve complete information about PD,
it is gotten through a PD acknowledgment system that records the PD waveforms
close to the regular PD. A couple of sorts of PD signals are difficult to recognize
at a starting stage. To crush the test, a Convolutional Neural Network (CNN) based
profound learning procedure for PD plan affirmation is presented in this paper. The
acquired PD signal is changed into a 3-D (F-q-n) picture. To anticipate such a PD the
3-D (F-q-n) picture is feed as a input to Deep Learning Algorithm. It uses Convo-
lutional Neural Networks (CNN) for picture gathering. In this, Alex Net is used for
perceiving the unmistakable PD.
Keywords Partial discharge · Deep learning · Pattern recognition · Spillage

current · Spark over voltage
1 Introduction
As indicated by IEC (International Electro-technical Commission) Standard 60,270,

partial discharge (PD) is an electrical discharge that does not completely bridge the
space between two conducting electrodes. Insulators are protective devices that are
utilized in transmission and distribution lines for protection between two conductors.
The fundamental driver of the disappointment of overhead line encasing is flashover
which happens in the middle of the line and earth during overvoltage in the framework
J. Ashmin Sugaji (B) · M. Ravindran · R. V. Maheswari

Department of Electrical and Electronics Engineering, National Engineering College, Kovilpatti,
India
https://doi.org/10.1007/978-981-16-4943-1_32
342 J. Ashmin Sugaji et al.
[2]. During this flashover, the gigantic warmth created by arcing causes penetrates
in the insulator body. In transmission frameworks, high voltage gear, covers assume
a significant part in the presentation of the hardware. It is a lot important to keep
up the protection in the electrical gear for the working activity during high voltage
power transmission [3].
Near mountain locality, farming, or beachfront territory, particles are kept on
encasing and contamination develops progressively [9]. This decreases the protection
strength when the protectors are dry [12]. In any case, when mist or light downpour
wets the dirtied protector, a conductive layer is framed on the defiled cover surface,
which starts spillage current [1, 10]. Spillage current begins to stream along the
surface and there are arcing releases [4]. The bend expansion causes the flashover on
the separator [2]. Such electrical releases show up to a span of substantially less than
1 s. PD estimations have arisen as an incredible demonstrative device for protection
framework condition observing [7]. Utilizing PD signal investigation the PD action
can be classified. The greater parts of the protectors are in the polluted structure
[11]. Because of the presence of air, pollutant bubbles (void) are made inside the
protecting material. PD consistently happens in a void. Because of the event of
release disappointment happens in the protection framework [6, 8]. Because of the
above reason, PD identification and estimation are fundamental for the expectation
of protection life for HV power hardware [5]. PD ordinarily occurs at imperfection
destinations like voids, depressions, breaks, joints, and delamination.
2 Experimental Setup
2.1 Test Specimen
Test specimens used are ceramic and non-ceramic 11 kV insulators shown in Fig. 1.
The dimensions of the test specimens are tabulated in Table 1.
1. TYPE A—Porcelain Disc Type Insulator
2. TYPE B—Glass Disc Type Insulator
3. TYPE C—Composite Disc Type Insulator.
2.2 Experimental Setup
The artificial contamination execution test with uniform contamination dissemination

is done in a fake chamber. The supply is given as demonstrated in Fig. 2. It shows
the test arrangement that having coupling capacitance, high voltage test transformer,
and fake chamber to hold the example.
The transformer evaluated voltage is 2*0.22/100/0.22 kV, appraised current
2*22.8/0.1A, the evaluated yield is 10kVA. The testing transformer is utilized to
Recognition of Partial Discharge Signal Using … 343
TYPE A TYPE B TYPE C
Fig. 1 Types of ceramic and non-ceramic insulators
Table 1 Dimensions of
Type Creepage Length Height (cm) Diameter (cm)
specimen
(cm)
A 29.7 5.3 24
B 27.3 3.8 23
C 32 11 10
Fig. 2 Experimental setup and control panel
deliver AC, DC, and impulse voltage. AC supply is given to the example through the
transformer. The test arrangement is controlled utilizing a control work area shown
in Fig. 2. The flashover voltage and spillage current are estimated from the control
work area.
3 Methodology
Before the tests, all the samples are thoroughly washed with refined water to remove
any signs of dirt or grease. To prevent dust or other contaminants, the samples are
allowed to dry naturally inside. After normal drying, at first, the samples are uncov-
ered for flashover test in unpolluted condition. At that point, the separators are covered
with quantitative contamination utilizing the brushing method. Figure 3 shows the
insulators polluted with NaCl and Fig. 4 shows the industrial polluted insulators.
Kaolin is a non-dissolvable substance. Following 24 h of common drying of the
samples, the samples are suspended in the chamber. The sample utilized for testing
is suspended vertically on the test arrangement. Then the high voltage is applied
gradually from zero voltage until a total flashover happens on the insulator surface.
At flashover, the voltage is recorded.
The dissolvable contamination material utilized in this work is NaCl. Kaolin
is utilized to address the non-dissolvable toxin. The method utilized for artificial
contamination is a strong layer technique. The brushing strategy is done, before
the beginning of the contamination test, using a weighing machine, the volume of
NaCl (Sodium Chloride), cement, and kaolin is measured. Contamination should be
washed and cleaned with refined water and a cloth after each use.
Fig. 3 Coastal polluted ceramic and non-ceramic insulators
Fig. 4 Industrial polluted ceramic and non-ceramic insulators

Fig. 5 Block diagram
Figure 5 addresses the block diagram of the process, where the insulators are
polluted using salt and cement. After the contamination PD is measured and finally
it is recognized using a deep learning algorithm.
3.1 Measurement of Pollution Severity
The degree of seriousness of the pollution has been indicated in two essential
manners. For industrial pollution, they are:
(1) The salt deposit density, SDD, the measure of salt pollutant on the insulator
surface in units of mg of salt per cm2 of protector surface.
(2) To normalize, for industrial pollution, the equivalent salt deposit density or
ESDD is utilized, which is characterized as the measure of NaCl that would
yield a similar conductivity at complete weakening as the non-NaCl salt.
4 Experimental Results
The flashover voltage decreases with the increase in pollution levels. Table 2 shows
the breakdown voltage and leakage current of the test specimens in unpolluted
conditions.
Table 2 Flashover test in

Type Breakdown Voltage(kV) Leakage current(mA)
unpolluted condition
A 110.93 57.21
B 85.27 40.56
C 97.38 41.75
4.1 Partial Discharge Analysis
At first, the test examples were tried without applying any contamination, with an
applied voltage and under dry condition. Figure 6 shows the partial discharge at
unpolluted condition.
(a) Porcelain (b) Glass
(c) Polymer
Fig. 6 PD Waveform at unpolluted condition

(c) Polymer
Fig. 7 PD Waveform at 1:1 ratio condition
Partial discharge test with 1:1 proportion condition: The ceramic and non-
ceramic insulator samples were tried by applying the contamination at 1:1 proportion
condition. Figure 7 shows the PD at 1:1 proportion dirtied condition.
4.2 3D Plot
The PD signals are changed into 3-D (F-q-n) pictures for design purpose. The PD
information comprise of Charge (q), Number of Discharge (n), and Phase angle (F).
This gives a phase-resolved PD (PRPD) pattern 3D (F-q-n) plot shown in Fig. 8 and
is changed over into a picture.
(c) Polymer
Fig. 8 3D plot for polluted insulators
The breakdown voltage for 11 kV porcelain insulator with coastal pollution is

shown in Table 3. The breakdown voltage diminishes with the increment in contam-
ination levels. Figure 9 shows the relation between salt ratio and breakdown voltage
for porcelain insulators.
The breakdown voltage for 11 kV glass insulator with coastal pollution is shown
in Table 4. The breakdown voltage diminishes with the increment in contamination
levels. Figure 10 shows the relation between salt ratio and breakdown voltage for
glass insulators.
Table 3 Breakdown voltage for 11 kV porcelain insulator with coastal pollution

Ratio Weight of NaCl Weight of Amount of ESSD (g/cm2 ) Breakdown
(g) Kaolin (g) Water (ml) voltage (kV)
0:0 0 0 0 0 110.93
1:1 10 10 15 0.011 75.136
2:2 20 10 15 0.016 67.42
Fig. 9 Relation between ratio of salt and voltage failure for disc type insulator
Table 4 Breakdown voltage for 11 kV glass insulator with coastal pollution

Ratio Weight of NaCl Weight of Amount of ESSD (g/cm2 ) Break down
(g) Kaolin (g) water (ml) voltage (kV)
0:0 0 0 0 0 85.27
1:1 10 10 15 0.016 73.64
2:2 20 10 15 0.018 67.83
Fig. 10 Relation between ratio of salt and voltage failure for Glass insulator
Breakdown voltage for 11 kV polymer insulator with coastal pollution is shown

in Table 5. Figure 11 shows the relation between salt ratio and breakdown voltage
for polymer insulator.
Table 5 Breakdown voltage for 11 kV polymer insulator with coastal pollution

Ratio Weight of NaCl Weight of Amount of ESSD (g/cm2 ) Breakdown
(g) Kaolin (g) water (g) voltage (kV)
0:0 0 0 0 0 97.38
1:1 10 10 15 0.0156 96.41
2:2 20 10 15 0.017 95.84
Fig. 11 Relation between ratio of salt and voltage failure for polymer insulator
In this paper, three cases are tried. The insulators are covered with a quantitative
brushing strategy. AC artificial contamination tests for ceramic and non-ceramic
insulators under different contamination levels with an alternate proportion and PD
are estimated. When the salt deposit density grows for each extent the breakdown was
reduced. Under such conditions, these are the inference from the acquired outcomes.
• The breakdown voltage of clean insulators is higher than the polluted insulators.
• Spark over-voltage diminishes with an increment in contamination level.
• The breakdown voltage of the insulators is caused by an increase in the ESDD
levels.
References
1. Chandrasekar S, Kalaivanan C, Cavallini A, Montanari GC (2009) Investigations on leakage

current and phase angle characteristics of porcelain and polymeric insulator under contaminated
conditions. IEEE Trans Dielectr Electr Insul 16(2):574–583
2. Boudissa R, Djafri S, Haddad A, Belaicha R, Bearsch R (2005) Effect of insulator shape on

surface discharges and flashover under polluted conditions. IEEE Trans Dielectr Electr Insul
12(3):429–437
3. Cavallini A, Chandrasekar S, Montanari GC, Puletti F (2007) Inferring ceramic insulator pollu-
tion by an innovative approach resorting to PD detection. IEEE Trans Dielectr Electr Insul
14(1):23–29
4. Chandrasekar S, Kalaivanan C, Montanari GC, Cavallini A (2010) Partial discharge detection
as a tool to infer pollution severity of polymeric insulators. IEEE Trans Dielectr Electr Insul
17(1):181–188
5. Simonyan K, Zisserman A (2014) Very deep convolutional networks for large-scale image
recognition. arXiv preprint arXiv: 1409.1556
6. Liao R, Fernandess Y, Tavernier K, Taylor GA, Irving MR (2012) Recognition of partial
discharge patterns. In: Power and energy society general meeting, pp 1–8
7. Lewin PL, Petrov LA, Hao L (2012) A feature based method for partial discharge source
classification. In: Conference record of the 2012 IEEE international symposium on electrical
insulation (ISEI), pp 443–448
8. Schroder P, Neubauer Y, Arumugam S, Schoenemann T (2014) Dielectric and partial discharge
investigations on ceramic insulator contaminated with condensable hydrocarbons. IEEE Trans
Dielectr Electr Insul 21(6):2512–2524
9. Duan L, Hu J, Zhao G, Chen K, Wang SX, He J (2019) Identification of partial discharge
defects based on deep learning algorithm. IEEE Trans Power Deliv 34(4):1557–1568
10. Othman NA, Piah MAM, Adzis Z (2017) Space charge distribution and leakage current pulses
for contaminated glass insulator strings in power transmission lines. IET Gener Transm Distrib
11(4):876–882
11. Maadjoudj D, Mekhaldi A, Teguar M (2018) Flashover process and leakage current char-
acteristics of insulator model under desert pollution. IEEE Trans Dielectr Electr Insul
25(6):2296–2304
12. Douar MA, Beroual A, Souche X (2016) Assessment of the resistance to tracking of polymers
in clean and salt fogs due to flashover arcs and partial discharges degrading conditions on one
insulator model. IET Gener Transm Distrib 10(4):986–994
Design and Analysis of LK Model Based
FEFET Memories
S. S. Vijayavelu, K. Mariammal, M. Adhitya Narayan,
and P. Subash Rathinam
Abstract Industries scaled down the size but the supply voltage was still in the 1 V
range for a decade. Reducing power consumption needs supply voltage to be reduced
especially for Dynamic Random-Access Memory (DRAM) which is a storage device
whose dynamic power consumption is proportional to the square of the power supply.
In this work, design of low power consuming DRAM and Static Random-Access
Memory (SRAM) with Ferro Electric Field Effect Transistor (FEFET) as transistor
nodes which has higher subthreshold swing due to negative capacitance, hence a
lower power supply can be applied and changing process parameter to find Static
Noise Margin (SNM) of SRAM and characteristics of DRAM. The simulations are
carried out with cadence for 32 nm Verilog-a model and are analyzed with Python.
Keywords FEFET · Subthreshold swing · Verilog-a · DRAM · SRAM
1 Introduction
In view of present-day smartphones which have more use than computers that
are plugged into power sockets, smartphones need to have a good mechanism of
consuming low power. Smartphone’s operating frequency diversely shifting towards
higher operating frequency and larger bandwidth. It drastically increases the rate of
consumption of power stored in the battery as dynamic power consumption is propor-
tional to the frequency. In order to reduce the power consumption, we need to reduce
the other factor that is the power supply but the conventional Metal Oxide Semicon-
ductor Field Effect (MOSFET)s or FinFETs have a lower subthreshold swing, i.e.,
higher S value of above 60 mV/decade. This higher value of S makes the conven-
tional FETS unusable for lower power supply because of lower I ON /I OFF . Hence
lower power consuming nodes are needed for today’s high-frequency application.
One approach of reducing the power supply without increasing Off current (I OFF ) is
decreasing the S (inverse of the subthreshold slope) which is ideally can’t be lower
than 60 mV for conventional FETS. S parameter can be given by the expression,
S. S. Vijayavelu (B) · K. Mariammal · M. A. Narayan · P. S. Rathinam

Department of Electronics Engineering, Madras Institute of Technology, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_33
354 S. S. Vijayavelu et al.
Fig. 1 FE on MOSFET
model FET
NC as
Voltage
+
-
Negative
capacitance

S = 1 + CCoxD ∗60 mV. One way of reducing the S values is reducing depletion
capacitance but that only reduces up to 60 mV. In this paper, DRAM and SRAM are
simulated from Ferro Electric FET (FEFET) which is designed with the idea of Ferro
Electric (FE) on MOSFET model [9] as shown in Fig. 1. Ferro electric materials have
the effect of hysteresis and need careful layout in order to reduce the hysteresis. One
way is to match the capacitance of MOSFET and Ferro electric material to stable
injection of charge carriers over a gate-controlled energy barrier. The subthreshold
slope S associated with the FET cannot be reduced below the Boltzmann limit which
is of 60 mV/decade, thus leading to enable low voltage and low power operation
[2]. Negative Capacitor Field Effect Transistors (NC-FETs) are certain to sustain
Moore’s law by reducing the supply voltage. The reduction in voltage is achieved by
integrating a negative capacitor in the gate, the internal voltage amplification turns a
transistor on and off at voltages lower [3]. Negative capacitance field effect transistors
(NC-FET) are low voltage and low power field effect transistors (FET) which have
the same structure as MOSFET except that ferro electric thin film is used as a gate
insulator. It can help us to reduce the leakage and to increase the on-state current of
the transistor. It can maintain a reasonable operating speed at very low voltage (0.5 V
or less) [4]. NC-FETs include ferro electric-FET that uses a ferro electric material
as the gate insulator. In this, the gate insulator behaves like a negative capacitor. In
this NC region, however, the gate insulator is thermodynamically unstable unless
stabilized using a positive series capacitor. Interestingly, this stabilization provides
the necessary voltage amplification to reduce S below Boltzmann limit [5]. Charge
conserved capacitor model (CLK) is considered for the simulation as a precaution
over if the spice simulator is faulty [6]. FEFET based FDSOI leading edge CMOS
is designed and it showed endurance cycles up to 105 cycles can be obtained and
it is shown that endurance can be increased [7]. A simple 1Transistor process with
1012 on/off endurance cycle is achieved by ZrHfO on the p-MOSFET. This shows
good results in FEFET based memories [8]. HfO2 is better than ZrHfO in endurance
and ferro electricity. By applying the simple idea of mounting a FE with parameters
obtained from HfO2 experiments on standard Berkley Short channel Insulated Gate
Design and Analysis of LK Model Based FEFET Memories 355
Model (BSIM4) [9]. The Landau Khalatnikov (LK) Equation is used to find Elec-
tric field and polarization of Ferro Electric materials (FE). The model for negative
capacitance is obtained by the time dependent LK equation which is mounted as an
external capacitor [10] given in equation number (1) value [1]. The Current–Voltage
characteristics of the field effect transistors FET is obtained by the thermal.
dP
E −ρ = α P + β P3 + γ P5 (1)
dt
Static Parameters (α, β, γ ), Kinetic Coefficient (ρ). Considered Values [9] are as
follows,
α FE = -8.65e10 (cm/F).
β FE = 1.92e20 (cm5 /F/C2 ).
gamma = 0 (cm9 /F/C4 ).
rho = 9 (´ cm).
This paper is organized as follows,
Section I consists of Introduction and literature survey. Section II presents the
design of SRAM and DRAM. Section III discusses the simulation results of SRAM
and DRAM and section IV Concludes the work.
2 SRAM and DRAM Design
This section presents the design of SRAM and DRAM. N-channel and P-Channel
Transistors are utilized for the design of these memories. Incorporation of FEFETS
enables the complete SRAM and DRAM design attractive in terms of power as it
requires minimal power consumption when examining with the other transistor-based
designs (Fig. 2).
Fig. 2 6 T SRAM
Fig. 3 DRAM circuit design
SRAM is two crossly coupled inverters with access transistors to write and read
the data. The cross coupled inverters act as a flipflop to store data as long as power
supply is applied and mostly used in caches (Fig. 3).
DRAMs are used in Main memory which uses a capacitor to store charge
dynamically. Power consumptions by DRAMs are proportional to supplied Voltage.
Simulations are done with FEFET which is obtained by connecting Negative Capac-
itance Verilog-a model and Verilog-a model for 32 nm FET by changing the
parameters in Generic BSIM-4 model with those of 32 nm FET [9].
From (1), We can derive for ferro electric Voltage which is modeled for NC model
given by Eq. (2)

T FE d Qp α Qp β Qp 3 γ Qp 5
V FE = ρ + T FE + + (2)
AF E dt AF E AF E 3 AF E 5
First 32 nm modified BSIM4 is obtained (Fig. 4) and NC is obtained (Fig. 5) to

get FENMOS and FEPMOS by connecting together.
As from (2) Gate charge information is needed to find VFE. So extra node is
created which transfers information about the Gate Charge (Fig. 6 and 7).
A simple Inverter Circuit is first shown which works at 0.3 V power supply. Shown
in Fig. 8 and results are shown in Fig. 9.
One thing that changes the characteristics of Ferro Electric Material is thickness
(t FE ). Depending upon the thickness, Ferro Electric materials may show increased
hysteresis and needs to be carefully developed that gives good SNM characteristics
and good transfer characteristics.
6 T SRAM Cell with FEFET as transistor nodes is designed in cadence as shown
in Fig. 10, and its one inverter output is shown for different thicknesses of the Ferro
Electric Material shown in Fig. 11and12.
Fig. 4 NMOS simulation

layout
Fig. 5 Negative capacitance

(NC) from LK equation
Fig. 6 NMOS and NC

connected to form FENMOS
Fig. 7 Modified simulation

layout for FENMOS
Fig. 8 Simple inverter circuit with 0.3 V supply
In Cadence, SNM calculation for various thicknesses in SRAM is tedious and

time consuming, So, we use Python (Fig. 11) to find SNM with data from Cadence
for Various Thickness (Figs. 13,14, 15, 16, 17, 18, 19 nad 20).
From the plots, we can find that Static Noise Margin is good and butterfly is wide
open for little value of thickness of the Ferro Electric materials which is because, As
the thickness increases hysteresis will play a role in one dimensional LK model of
the FE. To get good SNM, we have to take narrow FE.
With the plot available, READ and WRITE SNM is obtained for different
thicknesses of the material, (Figs. 21, 22)
Fig. 9 inverter input, output voltage characteristics
Fig. 10 Cadence design of

6 T SRAM
Fig. 11 VTC of one inverter output of SRAM

Fig. 12 (equivalent python plot of VTC)
Fig. 13 SNM plot for 1.5 nm tFE

Fig. 14 SNM plot for 2.5 nm tFE
Fig. 15 SNM plot for 3.9 nm
With the Figures of read and write SNM (Figs. 20 and 21), we can find optimum
value to get good SNM is around 8 nm. 3T1C DRAM Circuit is simulated with same
FEFET as transistor Nodes (Fig. 23) and corresponding simulations are obtained
(Fig. 24).
Volatile DRAM is simulated with FEFET with FE @10 nm which doesn’t give
hysteresis, If FE is at thickness higher than 10 nm, hysteresis will occur and can be
used as non-Volatile DRAM as shown in Fig. 25.
Fig. 16 SNM plot for 6.3 nm
Fig. 17 SNM plot for tFE 10 nm
One another difficulty is to sense the bit. In order to Sense the output and Refresh,
we need a sense amplifier to work upon 0.3 V which can be obtained with FEFET
with FE thickness less than 10 nm.
4 Conclusion
SRAM and DRAM design using FEFET is suggested in this work. The simulations
are carried for the suggested designs with FEFET as transistor nodes. For different
thicknesses of the Ferro Electric Material, SRAM and DRAM are Characterized and
for SRAM, Butterfly plots are plotted to get Static Noise Margin of the SRAM found
out optimum thickness of FE to work upon is around 8 nm to get better SNM because
of wide opening in butterfly plot such that we can detect higher or lower state of stored
Fig. 21 Read SNM
data precisely and also get good volatile DRAM, thickness of the FE is found to be
below 10 nm. As shown in Fig. 25, to get good Non-Volatile memories with the help
of hysteresis, t FE > 10 nm is a region of interest for non-volatile operations.
Fig. 22 Write SNM
Fig. 23 Cadence simulation design

Fig. 24 DRAM transient plot
Fig. 25 Hysteresis plot
References
1. Li Y, Lian Y, Yao K, Samudra GS (2015) Evaluation and optimization of short channel ferro-
electric MOSFET for low power circuit application with BSIM4 and Landau theory. Solid State
Electron 114:17–22
2. Salahuddin S, Datta S (2008) Use of negative capacitance to provide voltage amplification for
low power nanoscale devices. Nano Lett 8(2):405–410
3. Jain A, Alam MA (2014) Stability constraints define the minimum subthreshold swing of a
negative capacitance field-effect transistor. IEEE Trans Electron Devices 61(7):2235–2242
4. Kobayashi M, Hiramoto T (2015) Device design guideline for steep slope ferroelectric FET
using negative capacitance in sub-0.2 V operation: operation speed, material requirement and
energy efficiency. In: Proceedings symposium VLSI technology, June 2015, pp T212–T213
5. Sivasubramanian S, Widom A, Srivastava Y (2003) Equivalent circuit and simulations for the
Landau Khalatnikov model of ferroelectric hysteresis. IEEE Trans Ultrason Ferroelectr Freq
Control 50(8):950–957
6. Cirit MA (1989) The Meyer model revisited: Why is charge not conserved? [MOS transistor].
IEEE Trans Comput-Aided Design Integr Circuits Syst 8(10):1033–1037
7. Dünkel S et al. (2017) A FeFET based super-low-power ultra-fast embedded NVM technology
for 22nm FDSOI and beyond. In: IEDM Technology Dignostics, pp 19.7.1–19.7.4
8. Cheng C-H, Chin A (2014) Low-leakage-current DRAM-like memory using a one-transistor
ferroelectric MOSFET with a Hf-based gate dielectric. IEEE Electron Device Lett 35(1):138–
140
9. A Verilog-A Compact Model for Negative Capacitance FET “Muhammad Abdul Wahab and
Muhammad Ashraful Alam, Purdue University
10. Aziz A, Ghosh S, Datta S, Gupta SK (2016) Physics-based circuit-compatible SPICE model
for ferroelectric transistors. IEEE Electron Device Lett 37(6):805–808. https://doi.org/10.1109/
LED.2016.2558149
Convolutional Encoder–Decoder
Architecture for Speech Enhancement
Utkarsh Maheshwari, Piyush Goel, R Annie Uthra, Vinay Vasanth Patage,

Sourabh Tiwari, and Saksham Goyal
Abstract Signal processing faces the quandary of not being able to separate non-
stationary noise from speech signal. Traditional methodologies relied on spectral
subtraction for the same; however, such techniques relied on approximation of spec-
tral mask of the noise. This paper proposes an effective and novel convolutional
encoder–decoder architecture to effectuate clean speech from the input audio through
denoising the audio input. The architecture uses skip connections to increase informa-
tion flow from encoder to decoder, which helped the authors bolster the performance
of the network. The generated output is evaluated on objective and subjective metrics
like signal-to-noise ratio (SDR), perceptual evaluation of speech quality (PESQ) and
short time objective intelligibility (STOI). The proposed system outperforms the
state-of-the-art systems with respect to SDR, PESQ and STOI. The architecture
finds applications in various fields such as speech recognition, machine translation
and telecommunication.
Keywords Speech denoising · Neural networks · Speech enhancement · Babble

noise · Encoder–decoder networks
U. Maheshwari · P. Goel (B) · R. A. Uthra

Department of Computer Science Engineering, SRM Institute of Science and Technology,
U. Maheshwari
R. A. Uthra
V. V. Patage · S. Tiwari · S. Goyal
Samsung Research Institute, Bangalore, India
S. Tiwari
S. Goyal
https://doi.org/10.1007/978-981-16-4943-1_34
370 U. Maheshwari et al.
1 Introduction
Speech denoising can be defined as the process of removing unwanted audio signals
from the speech that affects the intelligibility of the original message. The aim is
to enhance the cognitive quality of speech signals, especially by removing back-
ground noise. Noise can be broadly classified into stationary and non-stationary noise.
Stationary noises are the ones with single-tone sine waves and constant frequency.
Based on the energy levels of the noise, stationary noises can be further classified
as white, pink, brown and black noises. Various techniques such as spectral gating,
MMCA and MRCA have been used in the past to reduce them. These methods rely on
calculating the spectrum of clean speech and blocking off any frequency higher than
the spectrum. While such methods perform well for stationary noises, they still have
difficulty coping with normal sounds such as the unstoppable noise or the whirling
sound found in the crowd at the same time. Most real-world scenarios contain what is
described as non-stationary noises. Non-stationary noises have varying frequencies,
thus making techniques such as spectral gating impractical. Babble noises like street
noise, barking dogs and background chatter make the speech incomprehensible for
the listener. However, recent advancements in deep learning have made it possible
to reduce such noises as well. Models based on neural networks work much better
than traditional methods for all kinds of noises. This could be attributed to the ability
of neural networks to generate more complex and nonlinear decision boundaries.
Furthermore, in-depth learning methods outperform the traditional methods in both
quantitative and qualitative metrics.
Inspired by these recent developments, the authors of this paper propose one such
novel architecture aimed to improve the speech quality. A symmetric convolutional
encoder–decoder architecture with sequence remodelling network is proposed. The
network takes a monophonic sound signal as an input and separates clean and noisy
signals. The network also consists of U-Net-type skip connections from encoder to
decoder. In addition, our network also performs voice activity detection (VAD), to
find silences in between. Many metrics have been shown to be inconsistent with
human judgements, we use the benchmark techniques like short time objective intel-
ligibility (STOI) and perceptual evaluation of speech quality (PESQ) to compare our
model with the state of the art. Speech enhancement finds applications in hearing
aids, robust speech recognition algorithms, mobile communication, etc. This paper
aims to propose one such deep learning algorithm to solve the problem of speech
denoising. Instead of focusing on the noise, we consider the task of denoising as
essentially to learn a mapping from noisy audio to clean audio.
The paper is organised as follows. In Sect. 2, we briefly describe prior work done
in the field. In Sect. 3, we propose our encoder–decoder-based architecture with
overall design. In Sect. 4, results of experiments are discussed, and finally in Sect. 5,
we conclude our study.
Convolutional Encoder–Decoder Architecture for Speech Enhancement 371
2 Literature Review
Prior to the use of deep learning architectures, spectral gating techniques were used
for speech enhancement. Such techniques primarily relied on the use of approxi-
mation of noise masks which were then removed from speech signals. The authors
of the [1] used various minimum mean-squared estimator (MMSE) methods to first
approximate spectra of the noise from non-speech parts and then subtracting the
noise from the input signal. However, MMSE-based methods were not suitable for
non-stationary noises. Additionally, such methods were inadequate for non-acoustic
noises. Another approach was proposed by the authors of [2] based on minima-
controlled recursive averaging (MCRA). As the name suggests, hypothetical gains
related to the presence of speech were recursively averaged to arrive at optimal spec-
tral gain function. Improved versions of MCRA were then proposed which instead
of relying on static smoothing parameter was time and frequency dependent, thus
providing better results for non-stationary noises with a low signal to noise ratio.
However, both MMSE and MCRA-based methods were not suitable for non-
stationary noises, and hence a nonlinear algorithm was required to develop a more
dynamic mask. Neural networks overcome this problem as they are capable of devel-
oping more intricate decision boundaries. Such networks instead of approximating
a noise mask considered the task of denoising as a mapping from noisy input to
clean output. One such architecture was proposed by the authors of [3]. A deep
neural network architecture trained on Aurora2 data set used pairs of clean and
noisy signals. The authors used global variance (GV) and dropout layers. Since
speech is a sequential signal, long short-term memory was used by authors of [4].
LSTM approximates masks using differential function between two subsequent time
instances. U-Net-based architecture was used by [5]. U-Net initially developed for
medical imaging tasks was augmented with long-range skip connections and trained
on instrumental sounds to develop spectrogram masks. All the papers mentioned
above used signal distortion ratio (SDR) as an objective function. Authors of [6]
proposed three architectures to finally meet the objective of universal sound separa-
tion. Three architectures, CLDNN based on ConvTasNet [7], TDCN++ with skip
connections and normalisation layers and iTDCN++, a two stage iterative TDCN
network was proposed.
3 Proposed System
The proposed system achieves speech enhancement through two steps, data prepro-
cessing followed by audio denoising as shown in Fig. 1. Data preprocessing converts
the audio signal to MFCC that is fed to the audio denoising module. The audio
denoising module separates the speech and noise through symmetric encoder and
decoder.
Fig. 1 Proposed system architecture
Fig. 2 MFCC features

represented on the MEL
scale with amplitude versus
frequency
3.1 Data Preprocessing
In the proposed system, the sound signals were first converted into monophonic
signals each of sample rate 16000 kHz. The signals are then converted into STFT
giving values of amplitude change in accordance with time. Next MFCCs of the
sound signals were extracted, where the hop length is taken to be 512, the number of
coefficients to be extracted is taken as 13, and the number of segments is taken as 1.
MFCC accounts for the human perception for sensitivity at particular frequencies by
converting the theoretical values into a MEL scale, thus they are suitable for speech
recognition. Figure 2 depicts a sample preprocessed noise signal with the various
MFCC coefficients on MEL scale.
3.2 Audio Denoising
The proposed denoising network is capable of enhancing monophonic input signal

and separating the input signal into its speech and noise constituents. The network
Fig. 3 Symmetric encoder–decoder architecture
consists of a symmetric encoder–decoder architecture along with a two-layered

LSTM network in between as shown in Fig. 3. Before feeding the audio to the
encoder, the signal is upsampled by a factor of two and then the output of the decoder
is downsampled by the same rate. The encoder contains five layers numbered one to
five with increasing number of output channels as we go deeper. In order to sepa-
rate non-stationary noise, nonlinearity needs to be introduced. 1 × 1 convolution is
effective for the same. Therefore, each layer of encoder consists of ReLu followed by
1 × 1 convolution, normalisation layers and GLU activation. A two-layered LSTM
network each with 128 hidden units takes the input from the fifth layer of the encoder,
and the output is then fed to the decoder. The architecture of the decoder is symmet-
rical to the encoder with five layers numbered five to one. Each layer of decoder
performs 1 × 1 transpose convolution followed by a GLU activation. Number of
output channels decreases symmetrically along the decoder except in the last layer.
The last layer consists of a single channel as the output. Skip connections as proposed
by Ronneberger et al. [8] connect the output of the kth layer of the encoder to the
kth layer of the decoder.
The network is trained for 300 epochs on Valentini data set with Adam optimiser.
The objective function is a ratio between distortion and clean signal.
Three metrics were used to evaluate the performance of the model on different data
sets. Signal to distortion ratio (SDR) is an objective metric that measures the l2
errors present between the model output and ground truth audio. Two additional
subjective metrics were evaluated. These were perceptual evaluation of speech quality
(PESQ) and short time objective intelligibility (STOI). These metrics measure the
intelligibility of speech. The network was trained on Valentini data set. The data set
consists of 400 sentences read out by 109 speakers added with babble noise from
CSTR corpus.
Fig. 4 SDR comparison between the proposed and state-of-the-art models
4.1 SDR Comparison
SDR was compared with convolutional encoder–decoder (CED) network proposed

by Park and Lee [9], deep neural network (DNN) proposed by Xu et al. [3] and
Wiener filter. SDR performance of wiener filter was least as the algorithm relies
on MMSE-based approach. Both CED and DNN showed improvements over tradi-
tional approaches as they are capable of nonlinear decision boundaries. The proposed
algorithm because of skip connections performed better than CED and DNN, and
the results are depicted in Fig. 4.
4.2 PESQ Comparison
The PESQ result was compared with SEGAN [10], Wave U-Net [11] and MMSE-
GAN [12]. The proposed architecture performed better than GAN-based architec-
tures. The results are summarised in Table 1.
Table 1 PESQ comparison

Model PESQ value
Proposed symmetric encoder–decoder model 2.63
SEGAN 2.16
Wave U-Net 2.4
MMSE-GAN 2.53
Table 2 Data set comparison

Data set PESQ STOI SDR
DNS 2.48 0.86 6.19
MCV + UrbanSound 2.34 0.84 6.98
Valentini 2.63 0.87 6.62
4.3 Data set Comparison
The results were compared between three data sets: DNS, Valentini and Urban
Sound8K augmented with Mozilla Common Voice to generate clean and noisy speech
pairs. The results for the same are summarised in Table 2.
4.4 Waveforms
Figure 5 displays the waveforms of noisy input and the clean output of the model.
The figure also consists of a waveform of the noise present in the input audio. These
waveforms can be used to estimate the performance of the model.
Fig. 5 From left to right: a noisy input b cleaned output c separated noise
5 Conclusion
This paper proposes an idea for an encoder–decoder-based speech enhancement

model. The technique is to train a network to learn the mapping between noisy and
clean training samples and then use the trained weights to enhance audio signals
that are not seen before in the data set. The model is evaluated on three different
subjective and objective metrics and three data sets to validate the performance. The
idea is to use the proposed algorithm for speech recognition, machine translation and
telecommunication.
References
1. Boll S (1979) Suppression of acoustic noise in speech using spectral subtraction. IEEE Trans
Acoust Speech Signal Process 27(2):113–120. https://doi.org/10.1109/TASSP.1979.1163209
2. Cohen I, Berdugo B (2009) Speech enhancement for non-stationary noise environments. Signal
Process 81:2403–2418. https://doi.org/10.1016/S0165-1684(01)00128-1
3. Xu Y, Du J, Dai L, Lee C (2015) A regression approach to speech enhancement based on deep
neural networks. IEEEACM Trans. Audio Speech Lang. Process. 23(1):7–19. https://doi.org/
10.1109/TASLP.2014.2364452
4. Weninger F, Hershey JR, Roux JL, Schuller B (2014) Discriminatively trained recurrent neural
networks for single-channel speech separation. In: 2014 IEEE global conference on signal and
information processing (GlobalSIP), December 2014, pp 577–581. https://doi.org/10.1109/Glo
balSIP.2014.7032183
5. Jansson A, Humphrey E, Montecchio N, Bittner R, Kumar A, Weyde T (2017) Singing voice
separation with deep U-Net convolutional networks, pp 7
6. Kavalerov I et al. (2019) Universal sound separation. ArXiv190503330 Cs Eess Stat, August
2019, Accessed 28 Jan 2021. [Online]. Available http://arxiv.org/abs/1905.03330
7. Luo Y, Mesgarani N (2019) Conv-TasNet: surpassing ideal time-frequency magnitude masking
for speech separation. IEEEACM Trans Audio Speech Lang Process 27(8):1256–1266. https://
doi.org/10.1109/TASLP.2019.2915167
8. Ronneberger O, Fischer P, Brox T (2015) U-Net: convolutional networks for biomedical image
segmentation. ArXiv150504597 Cs, May 2015, Accessed 09 Mar 2021. [Online] Available
http://arxiv.org/abs/1505.04597
9. Park SR, Lee J (2016) A fully convolutional neural network for speech enhancement.
ArXiv160907132 Cs, September 2016, Accessed 27 Jan 2021. [Online]. Available http://arxiv.
org/abs/1609.07132
10. Pascual S, Bonafonte A, Serrà J (2017) SEGAN: speech enhancement generative adversarial
network. ArXiv170309452 Cs, 2017 Jun, Accessed 09 Mar 2021. [Online]. Available http://
arxiv.org/abs/1703.09452
11. Macartney C, Weyde T (2021) Improved speech enhancement with the wave-U-Net.
ArXiv181111307 Cs Eess, Nov. 2018, Accessed 09 Mar 2021. [Online]. Available http://arxiv.
org/abs/1811.11307
12. Phan H et al (2020) Improving GANs for speech enhancement. IEEE Signal Process Lett
27:1700–1704. https://doi.org/10.1109/LSP.2020.3025020
Tuning of MIMO PID Controller Using
HCLPSO Algorithm
T. Jeyaraman, D. Joelpraveenkumar, M. Kaliraj, M. Krishna Chandar,

and M. Willjuice Iruthayarajan
Abstract In this paper, the heterogeneous comprehensive learning particle swarm

optimization is proposed for the tuning of multivariable proportional integral deriva-
tive (PID) for Wood and Berry system. This simulation work is done for both the
decentralized and centralized PID controller. For comparison, results from the tuning
of multivariable PID controller by particle swarm optimization (PSO) algorithm is
considered. Here, our objective is to minimize the integral absolute error (IAE) value
of the system. For the simulation of the system and algorithm, MATLAB/SIMULINK
software is used. Statistical performance of evolutionary algorithms such as best
value, mean value, and standard deviation are going to be evaluated based on ten
independent initial conditions. In this work, it is observed that HCLPSO give more
consistent performance compared to PSO algorithm.
Keywords MIMO · PID · Centralized · Decentralized · HCLPSO · Tuning
1 Introduction
In this modern world, most of the industries are having multiple inputs and multiple
outputs systems. These systems are very complicated to get the desired output so
that a suitable controller and tuning methods are needed. Most of the MIMO system
is controlled by using a PID controller [1]. PID controller [2] is the most popular
controller due to its advantages over others. It is most widely used in industrial
applications. It uses a control loop feedback mechanism to control the variables. PID
stands for the mathematical term proportional, integral, and derivative. Proportional
means constant or right value. Integral means the summation of a function over a
given interval and derivative is the rate of change of value for a given interval. In
this work, decentralized and centralized PID controllers for the tuning of the MIMO
system. In decentralized controller [3], there are n PID controllers and in centralized
controller [4] n*n PID controllers. In various engineering applications, the tuning of
T. Jeyaraman · D. Joelpraveenkumar · M. Kaliraj · M. K. Chandar · M. W. Iruthayarajan (B)

Department of EEE, National Engineering College, Kovilpatti 628503, India
https://doi.org/10.1007/978-981-16-4943-1_35
378 T. Jeyaraman et al.
the PID controller for various algorithm is mentioned in [5]. In recent times, hetero-
geneous comprehensive learning particle swarm optimization (HCLPSO) algorithm
[6] performs very well. The tuning of PID controller for SISO system for HCLPSO
algorithm was explained in the paper [7]. In this paper, we are going to tune the PID
controller for MIMO system by using HCLPSO algorithm. Among population-based
tuning algorithms, HCLPSO has the learning capability to become the output of the
system as consistent, converged, and has better performance. In HCLPSO, the total
population is grouped into two subpopulations such as exploration and exploitation.
For the optimal results, we have to balance between these two subpopulations. This
paper mainly focuses on minimizing the error value such as integral absolute error
(IAE) and the performance analysis of the PID controller for the binary distillation
column plant that was described by Wood and Berry [8]. The PID controller continu-
ously calculates the error value which is the difference between the desired set point
and the measured process variable and applies the correction by changing the propor-
tional, integral, and derivative value. For the performance analysis, we compare the
output results of HCLPSO to the output results of PSO.
2 MIMO System
The MIMO system is the multiple inputs and multiple outputs system which requires
control techniques for improving the performance of the system. The MIMO system
is more difficult to exploit than a SISO system. We have used a 2 × 2 Binary
distillation column plant described by Wood and Berry.
The generalized transfer function for the MIMO system is given below,
⎡ ⎤
g11 (s) · · · g1n (s)
⎢ ⎥
G(s) = ⎣ ... . . . ... ⎦ (1)
gn1 (s) · · · g11 (s)
2.1 Wood and Berry System
Wood and Berry have derived the mathematical model for binary distillation column
plant. The typical column plant consists of a vertical column, re-boiler, condenser,
reflux drum, and it contains a feed stream and two product streams. The column plant
requires a minimum four number of feedback control loops. These control loops are
used to control the distillate concentration, bottom concentration, level of re-boiler,
and level of reflux rate. Each controller loop requires a minimum of one input and
output so that the system is considered as a MIMO system.
The derived transfer function for the column plant is given by
Tuning of MIMO PID Controller Using HCLPSO Algorithm 379

12.8e−s −18.9e−3s
G(s) = 1+16.7s 1+21s
6.6e−7s 19.4e−3s (2)
1+10.9s 1+14.4s
2.2 Decentralized Controller
In a decentralized controller, there are n controllers are used. We used the Wood and
Berry system with 2 PID controllers. The decentralized controller system is most
frequently used in industries. The computational time is less for a decentralized
controller system compared to a centralized controller system. Simulink model for
the decentralized controller is shown below (Fig. 1). The general transfer function
for the decentralized controller is given by,
⎡ ⎤
k1 (s) · · · 0
⎢ .. . . .. ⎥
K (s) = ⎣ . . . ⎦ (3)
0 · · · k n (s)
PID
GC1 GP11 D11
GP12 D12
GP21 D21
PID
GC2 GP22 D22
Fig. 1 Simulink model for decentralized controller

PID
GC1 GP11 D11
PID
GC3 GP12 D12
PID
GC4 GP21 D21
PID
GC2 GP22 D22
Fig. 2 Simulink model for centralized controller
2.3 Centralized Controller
For a MIMO system, a full multivariable controller is used, such a controller is

called a centralized controller. This controller will give satisfactory responses for
the desired system. The centralized controller requires n*n controllers, whereas the
decentralized controller requires only n controllers. In this controller, two PIDs are
coupled to each other to obtain the desired output for the system. We can design a
full matrix for a centralized controller system and find out the controller parameter
values by suitable tuning methods. Simulink model for the centralized controller is
shown below (Fig. 2).
The general form of n* n multivariable PID controller is given by,
⎡ ⎤
k11 (s) · · · k1n (s)
⎢ ⎥
K (s) = ⎣ ... . . . ... ⎦ (4)
kn1 (s) · · · knn (s)
3 HCLPSO Algorithm
In HCLPSO, the swarm is divided into two heterogeneous subpopulations. The first
subpopulation is enhanced for exploration and the second subpopulation is enhanced
for exploitation. In both exploration and exploitation subpopulations, the exemplar
is generated using comprehensive learning (CL) strategy with the learning proba-
bility. Each subpopulation is assigned to focus only on either exploration or exploita-
tion. To generate exemplars, in exploration subpopulation exemplars, the personal
best experience of the same subpopulation is used and in the exploitation subpop-
ulation, the personal best experience of the entire swarm population is used. As
exploration subpopulation learns only from the same population, the diversity can be
retained. The velocity of an exploitation-enhanced subpopulation is updated using
the following equation:
Vid = ω∗ Vid + c1∗ randd1∗ p bestdf i(d) − X id + c2∗ rand2∗ g bestd −X id
d dimension
V id updated velocity of an ith particle
ω inertia weight
c1, c2 acceleration factor
rand1 d, rand2 uniform random between 0 and 1
Xid dth value of ith particle in the population
Pbestd fi(d) best position of the ith particle
Gbestd best position of the whole swarm population
Vid = wVid + c∗ randid ∗ pbestdf i(d) − X id
where f i (d) =[f i (1), f i (2),…., f i (D)] indicates the ith particle follows its own or
other pbest for each dimensions. The particle is specified according to the learning
probability values(pc i ) .
4 Parameter Analysis
For the decentralized controller, we made the performance analysis. In HCLPSO,

there are several tuning parameters such as learning probability, grouping, and the
velocity of the particle. The output result was analyzed by changing these parameters,
keeping the two parameters as constant. So that one parameter is kept as constant
and change the other parameter and analyze the output. In order to find the effect
of learning probability to the performance of HCLPSO, the velocity of the particle
is kept as constant. The obtained statistical performance such as best value, mean
value, and optimal value among ten independent runs for the velocity = 0.15, learning
probability = 0.25 and velocity = 0.15, learning probability = 0.15 are represented
in Table 1.
From Table 1, we infer that
Table 1 Performance of HCLPSO algorithm based on learning probability

Parameters Functional count Best value Mean value Std deviation value
Velocity = 0.15 and 9000 8.1462 8.2849 0.1714
learning probability y = 3000 8.6521 9.5202 0.7022
0.25
Velocity = 0.15 and 9000 8.1129 8.5659 0.3783
learning probability y = 3000 8.6511 10.2654 1.9256
0.15
Table 2 Performance of HCLPSO algorithm based on velocity of the particles

Parameters Functional count Best value Mean value Std deviation value
Velocity (V max ) = 0.25 3000 9.7702 10.3743 0.5655
Learning probability (Pc) 9000 9.6836 9.7862 0.2620
= 0.2
Velocity (V max ) = 0.2 3000 9.8895 10.5557 0.5392
Learning Probability (Pc) 9000 9.6824 9.8263 0.3014
= 0.2
• If we decrease the learning probability value, the standard optimal value will
increase and output will not converge.
• If we increase the learning probability value, the standard optimal value will
decrease and output will converge, and it will give the consistent performance.
The obtained statistical performance such as best value, mean value, and standard
deviation value for the velocity = 0.25, learning probability = 0.2 and velocity =
0.2, learning probability = 0.2 are represented in Table 2. In order to find the effect
of maximum velocity of the performance of the HCLPSO algorithm, the learning
probability is kept as constant.
From the above Table 2, we infer that
1. If we decrease the velocity of the particle, it will give the best optimal solution.
2. From the above analysis, we conclude that the best optimal solution and system
consistent
5 Results
From this paper, we analyze the parameter changes and finally conclude that the
best optimal solutions will get for the particular parameters. We made several trials
and runs and obtained the suitable parameters for the design of centralized and
decentralized PID controllers for HCLPSO algorithm.
For decentralized controller, the parameters are learning probability (pc) = 0.25,
grouping ratio (g1) = 0.3, and velocity (Vmax) = 0.2. The overall obtained results for
Table 3 Comparison of HCLPSO algorithm with PSO algorithm for decentralized controller
PSO obtained results Max_FES Best optimal Mean optimal Standard deviation value
3000 9.6824 10.2209 0.5433
6000 9.6824 10.104 0.4444
9000 9.6824 10.0822 0.5468
Learning probability(pc) = 0.25, Grouping ratio(g1) = 0.3, Velocity(Vmax) = 0.2
HCLPSO obtained 3000 9.9101 10.5299 0.4284
results 6000 9.6826 9.7301 0.0666
9000 9.6825 9.6851 0.0018
Table 4 Comparison of HCLPSO algorithm with PSO algorithm for centralized controller
PSO obtained results Max_FES Best optimal Mean optimal Standard deviation value
3000 8.3255 9.0636 0.7900
6000 8.1245 10.0422 2.7854
9000 8.0189 9.0186 1.5178
12,000 8.0273 10.9396 4.4975
Learning probability(pc) = 0.25, Grouping ratio(g1) = 0.3, Velocity(vmax) = 0.15
HCLPSO obtained 3000 8.6521 9.5202 0.7022
results 6000 8.2661 8.8084 1.1179
9000 8.1462 8.2849 0.1749
12,000 8.0945 8.2580 0.1289
the decentralized controller using HCLPSO algorithm for various function evolutions
are reported in Table 3.
For centralized controller, the parameters are learning probability (pc) = 0.25,
grouping ratio (g1) = 0.3, and velocity (vmax) = 0.1. The overall obtained results
for the centralized controller using HCLPSO algorithm are represented in Table 4.
From the Tables 3 and 4, the obtained best performance of both the HCLPSO and
PSO algorithms are similar, but the performance of HCLPSO algorithm gives more
consistent performance compared to PSO algorithm when the number of function
evaluations are higher due to their learning ability. Convergence characteristics of
both the algorithms are reported in Figs. 3 and 4 which shows both the algorithms
have converged their optimal values.
6 Conclusions
• From the Table 1, we conclude that by increasing the learning probability value,
the output results are more consistent and the deviation is minimum.
Fig. 3 Convergence graph for decentralized controller
Fig. 4 Convergence graph for centralized controller
• From the Tables 3 and 4, we conclude that PSO and HCLPSO have moreover
give same optimal solution, but HCLPSO gives lesser value of standard deviation
value for increasing the function evaluation of the particle. Since the HCLPSO has
learning probability, it gives more consistent solution for greater value of function
evaluation.
• From the Tables 3 and 4, we conclude that centralized controller has better optimal
solution compared to the decentralized controller.
Acknowledgements We thank the college management for providing the lab facility to do the
project work.
References
1. Juang JG, Huang MT, Liu WK (2008) PID Control using presearched genetic algorithms for a
MIMO system. IEEE Trans Syst Man Cybern Part C (Appl Rev) 38(5):716–727
2. Ochi Y, Yokoyama N (2012) PID controller design for MIMO systems by applying balanced
truncation to integral- type optimal servomechanism. IFAC Proceed Volumes 45(3):364–369
3. Harjare VD, Patre BM (2015) Decentralized PID controller for TITO systems using characteristic
ratio assignment with an experimental application. ISA Trans 59(385–397)
4. Dhanya Ram V, Chidambaram M (2015) Simple method of designing centralized PI controllers
for multivariable systems based on SSGM. ISA Trans 56: 252–260
5. Willjuice Iruthyarajan M, Baskar S (2009) Evolutionary algorithms based design of multivariable
PID controller. Expert Syst Appl 36(5):9159–9167
6. Lynn N, NagaratnamSuganthan P (2015) Heterogeneous comprehensive learning particle swarm
optimization with enhanced exploration and exploitation. Swarm Evol Comput 24:11–24
7. Muniraj R, Willjuice Iruthayarajan M (2018) Tuning of robust PID controller with filter for PLL
system using HCLPSO algorithm. Int J Pure Appl Math 118(16):181–197
8. Wood R, Berry M (1973) Terminal Composition control of a binary distillatillation column.
Chem Eng Sci 28(9):1707–1717
Design Guide for Small-Scale
Grid-Connected PV System Using PVsyst
Software
A. Lavanya, Kushagra Bhatia, J. Divya Navamani, A. Geetha,

and K. Vijayakumar
Abstract Recently, small-scale grid-connected PV systems are acquiring familiarity

in institutions and industries mainly due to their clean and climate-friendly attribute.
A 104 kW grid-connected PV system at Kattankulathur, Tamil Nadu, is discussed
in this paper. A software tool is essential to estimate and predict solar energy power
generation for a particular location. This paper mainly deliberates the design aspects
using a suitable software tool for the grid-connected PV system. The best approach
for designing and evaluating energy production in PV systems is PVsyst software.
Therefore, this paper demonstrates the step-by-step procedure followed for creating
the 104 kW PV system. The approach towards loss distribution and performance
ratio analysis is also presented. This paper is proposed as a guide for the researchers
working in this field of designing a PV system based on the load requirement.
Keywords PVsyst · Grid connected · Loss distribution · Performance ratio · Case

study
1 Introduction
India has a solar potential of about 748 GW, assuming three per cent of the wasteland
area covered by PV modules as evaluated by National Institute of Solar Energy.
National Solar Mission (NSM) plays a most important role in endorsing the solar
energy to the central place in the country’s National Action Plan on climate change
[1]. NSM was launched on 11 January 2010. NSM is a noteworthy initiative of India’s
Government with dynamic contribution from states to encourage environmentally
sustainable development while addressing India’s energy safety trials. It will also
establish a substantial involvement by India to the global effort to encounter climate
change challenges [2–5]. The mission’s objective is to build India as a universal front
A. Lavanya (B) · K. Bhatia · J. Divya Navamani · A. Geetha · K. Vijayakumar

SRM Institute of Science and Technology, Kattankulathur, TamilNadu 603203, India
K. Bhatia
https://doi.org/10.1007/978-981-16-4943-1_36
388 A. Lavanya et al.
Fig. 1 Popular design

software for PV system
runner in solar energy by making the strategy circumstances for solar technology
dissemination across the country as rapidly as likely. The mission’s goal is to install
100 GW grid-connected solar PV plants by the year 2022 is the motivation for
this paper. India’s total installed generating power capacity is 375.323 GW. As of
December 2020, from this coal share is 55.4%, renewable energy is 23.5%, nuclear
1.8%, and hydro energy involvement is 12.3%. The comprehension of a photovoltaic
(PV) case study is to complete the step-by-step procedure for sizing, which permits
the assumed data confined in the practical manual, as topographical localization of
the location and the energy requirements, to assess predominantly the size of the
PV array and the volume of the batteries. To achieve this operation, sizing tool is
required. An evaluation between the fixed-tilt plane and seasonal tilt adjustment of
PV system is finished, built on their efficiency using PVSYST, was proposed in [6,
7]. Based on the statement acquiesced by the Ministry of New and Renewable Energy
(MNRE) and TERI an institute for energy and resources, some of the world’s most
commonly used PV simulation software for PV system design is shown in Fig. 1.
2 Step-by-Step Guild to Simulate a Grid-Connected PV

System
In this guild, the objective is to simulate, study and analyse a grid-tied PV system.
To start, the basic data collected in order to build and design the system is utilized.
All the information will be given in each and every step along the way. PVsyst V6.88
is employed to demonstrate the considered PV plant simulation.
Design Guide for Small-Scale Grid-Connected … 389
Fig. 2 Start window PVsyst
Step 1: Start and select the initial in the software

Start-up PVsyst software; select ‘Grid-Connected’ under ‘Project design’ to start
the project as shown in Fig. 2.
Step 2: New project and site file
As shown in Fig. 3, this is what the work space looks like, two subheadings,
namely ‘Project’s Designation’ and ‘System Variant’. To start, first thing needed is a
‘Site file’. Site data consists of location details such as latitude, longitude, sea level
and time zone. Upload ‘Site file’ of the correct format if in hand or create a new
file. PVsyst allows us to generate our custom ‘Site file’ and ‘Meteo file’. Click the
highlight icon in Fig. 3. A new window will open as shown in Fig. 4, ‘Geographical
Site Parameter’, three sub-tabs: ‘Geographical Coordinates’, ‘Monthly Meteo’ and
‘Interactive Map’.
Step 3: Creating our own Site and Meteo File
Enter the ‘Site Name’, ‘Country’ and the ‘Region’. To get the Meteo data, enter
‘location coordinates’; Click on the ‘Get from Coordinates’ to finish. The software
will automatically get the ‘Altitude’ and the ‘Time Zone’ through the site coordinates.
Import Meteo Data. The ‘import’ option is open under ‘Meteo data Import’ and select
from the data. Use ‘Meteonorm 7.2’. Click ‘import’. A new window tab ‘Monthly
Meteo’ will be opened which will show all the data parameters that will be applied
to our system. Refer Fig. 5 for all the parameters. Select the ‘Irradiance units’ as
‘kWh/m2, day’ as shown in Fig. 5.
Step 4: System variant configuration, calculation
Fig. 3 New project window
Fig. 4 Geographical site parameters

Fig. 5 Data parameters
In Fig. 6, under ‘Main parameters’, there are two options: ‘Orientation’ and
‘System’.
Click on ‘Orientation’, a new window pops as shown in Fig. 7. Enter the ‘Plane
tilt’ (in degrees) and ‘Azimuth’ (in degrees) under ‘Field Parameters’, as shown in
Fig. 7 highlighted. Click ‘Ok’ when done.
Fig. 6 Main project window, updated

Fig. 7 Orientation tab
Click on ‘System’, here all the necessary data parameter that is to be entered such
as ‘Presizing’, ‘PV modules’, ‘Inverter’. Under ‘Presizing’, option is there to either
use the ‘Planned Power’ or ‘Available Area’. Use either of the two for presizing.
Under ‘Select the PV modules’, access a large database of existing PV module
configurations by manufacturer. Estimate the inverter size using the load side data,
or use the inverter configuration present on-site if known. All the selection will be
shown in Fig. 8. Click ‘Ok’ once the configuration is complete.
Step 5: Getting things ready for simulation
Things are almost ready for simulation. There are a few additional parameters that
can be explore, such as ‘Detailed losses’ and ‘Self consumption’; refer Fig. 9. After
a successful run, results via a report format are obtained, which can be downloaded
as a PDF, or click ‘Detailed result’ to view Figs. 11, 12 and 13 for the result.
As a result of this study, it is determined that the PVsyst software considered proposes
the configuration based on the initial steps as given below in Fig. 10 (Table 1)
Horizontal global irradiation, diffuse irradiation, ambient temperature, global inci-
dent and effective radiation, effective energy at the output of the array, energy injected
into grid and performance ratio details are given in Table 2.
Fig. 8 Complete configuration details
Fig. 9 Simulation run

Fig. 10 System design

procedure
Table 1 PV array characteristics

PV array characteristics
PV module Si-poly model poly 325 Wp 72 Manufacturer lubi electronics
cells
Number of PV modules in series 17 modules Total number of PV modules Nb. modules 340
in parallel 20 strings
Unit nom. power 325 Wp Array global power nominal (STC) 111 kWp
At operating cond. 101 kWp (50 °C) array Total area nodule area 660 m2
operating characteristics (50 °C) Cell area 596 m2
U mpp 585 V I mpp 172 A
Inverter model solar inverter RPI M50A Manufacturer delta energy
Characteristics operating voltage 200–800 V Unit nom. power 50.0 kWac
Max. power (= > 35 °C) 55.0 kWac
Fig. 11 PVsyst results

Normalized productions (per installed kWp): Nominal power 111 kWp
Fig. 12 Performance ratio PR. Normalized productions (per installed kWp): Nominal power 111
kWp
Fig. 13 PV loss estimation
4 Conclusion
The small-scale grid-connected PV system intended for the desired load using the
PVsyst software is investigated. PVsyst software analyses the PV system power
generation and also the system losses. The entire study is focused to design a grid-
connected photovoltaic system for the load assumed in Kattankulathur location,
Tamil Nadu, India. 340 PV modules of 325 Wp are estimated for the desired load.
From the system considered for design, the energy production is 166.9 MWh/year,
with a specific prod. 1510 kWh/ kWp/year and the performance ratio PR is 80.59%.
Horizontal global radiation obtained will be 1927 kWh/m2 , and efficiency at STC is
estimated as 16.77%.
Table 2 Performance of designed small-scale grid-connected PV system

GlobHor DiffHor T_Amb Globlnc GlobEff EArray E_Grid PR
kWh/m2 kWh/m2 °C kWh/m2 kWh/m2 MWh MWh
January 149.9 65.83 25.10 135.0 128.0 12.45 12.24 0.820
February 164.8 57.87 26.19 152.0 145.4 13.87 13.62 0.811
March 196.4 73.77 28.06 188.8 181.8 17.06 16.73 0.802
April 194.0 74.16 29.68 193.0 186.6 17.30 16.94 0.794
May 192.5 85.75 31.74 197.1 190.4 17.57 17.21 0.790
June 169.1 88.48 30.72 174.0 168.0 15.69 15.38 0.800
July 161.0 91.88 30.43 164.5 158.4 14.86 14.57 0.801
August 162.0 89.33 29.48 162.8 156.8 14.75 14.47 0.805
September 160.8 76.35 28.80 157.1 151.2 14.21 13.94 0.803
October 137.3 81.81 27.43 130.5 124.7 12.01 11.80 0.818
November 115.3 71.14 25.55 107.2 102.0 9.95 9.78 0.826
December 124.1 65.91 24.90 112.3 106.4 10.41 10.24 0.825
Year 1927.2 922.28 28.18 1874.4 1799.7 170.11 166.91 0.806
References
1. Shalwala RA, Bleijs JAM (2009) Impact of grid-connected PV systems in residential areas
in Saudi Arabia. In: Proceedings of the 44th international universities power engineering
conference (UPEC) (IEEE 2009), pp 1–5
2. Tallab R, Malek A (2015) Predict system efficiency of 1 MWc photovoltaic power plant intercon-
nected to the distribution network using PVSYST software. In: 2015 3rd international renewable
and sustainable energy conference (IRSEC), Marrakech, Morocco, pp. 1–4. https://doi.org/10.
1109/IRSEC.2015.7454973
3. Yadav P, Kumar N, Chandel SS (2015) Simulation and performance analysis of a 1KWp Photo-
voltaic system using PVSYST. In: International conference on computation of power, energy,
information and communication
4. Sharma V, Chandel SS (2013) Performance analysis of a 190 kWp grid interactive solar
photovoltaic power plant in India. Energy 55:476–485
5. Lavanya A, Divya Navamani J, Vijaya Kumar K, Suman PD, Mishra S (2021) Selection of renew-
able energy materials for dual input DC–DC converter based hybrid energy system. Materials
Today: Proceedings 34:379–385
6. Yadav AK, Chandel SS (2013) Tilt angle optimization to maximize incident solar radiation: a
review. Renew Sustain Energy Rev 23:503–513
7. Mazuera MA, Posada J, Manrique P (2016) Analysis and emulation of a photovoltaic solar
system using a MATLAB/simulink model and the DS1104 Unit. IEEE, October 2016
Evaluation of Various Machine Learning
Algorithms for Detection of Attacks in 5G
C. Arul Stephen, R. Mathesh, L. Venkat, B. Ebenezer Abishek,

and A. Vijayalakshmi
Abstract Security plays an essential role in IoT as there are several components
connected to the network. An effective intrusion detection system is required to
detect all vulnerabilities in wireless networks using machine learning and also to
reduce the false alarm rate. The various machine learning algorithms were analyzed
in terms of various performance metrics and finally evaluated the best classifier
in terms of prediction time, training time, and accuracy. Using confusion matrix,
various classes of attacks were analyzed, and results revealed that various machine
learning algorithms were compared, And our proposed work was able to predict all
the five types of attacks effectively with an accuracy of 94% achieved using extra
tree classifier.
Keywords Machine learning · Security · Wireless attack detection · IoT · 5G

networks
1 Introduction
This introduction about the wireless attack and discussing IoT data with machine
learning is better security.
1.1 Wireless Attacks in Wireless Networks
Wireless networks are Internet-connected networks encompassed by dealing with

radio waves. A wireless network enables pursuits to bypass the expensive process
of offering cables into buildings or connecting different equipment locations.
However, one drawback of a wireless network [1] is a security issue. In a wire-
less network the information can be hacked. Hence, the research worker moves to a
C. Arul Stephen (B) · R. Mathesh · L. Venkat · B. Ebenezer Abishek · A. Vijayalakshmi

Vistas, Ch-117, Pallavaram, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_37
398 C. Arul Stephen et al.
sophisticated level like encrypting and decrypting the communications. Still, some-
times hackers may be viable to hack our data. If significant top-level companies lose
their information, this leads to millions of dollars which may be defeated. Therefore,
we required a secondary shield diagnosis system before black hat hack the data. We
can create more advanced when wireless attacks start in contact with a computer
device to suddenly identify by detecting system and automatically off the computer
before diminishing security. Accordingly, researchers investigate and develop more
detecting strategies.
1.2 Reason for IoT is the Need for Machine Learning

for Wireless Detection
With the increasing wireless attacks all day, people lose their security from smart-
phones and laptops, so hackers easily misuse our information. Other side increases
demand for the Internet of Things (IoT) automation system for detecting attacks
day by day, but technology is improving, IoT models are detectable, and the attacks
are more arduous. Progressively, this leads to researchers investigating, moreover,
machine learning-based applications besides IoT. Here the machine learning tech-
nique is to help the security additionally more robust detectable vulnerability. The
machine learning techniques advantage is that it works faster than other methodolo-
gies. The regular IoT intrusion detection system inside the data includes an unre-
lated data signal, so performs it wrongly detecting some attacks [2]. Still, in machine
learning, we collected the information directly from the IoT source, analyzed the data,
discarded unwanted data, and used the best features to secure the detection of attacks
more strongly. Making more robust data in the machine learning and predicting
styles that can overcome the problem raised from IoT cannot recognize threats. This
machine learning is able to identify the past threads and future unknown threads.
Hence, this paper is using machine learning to analyze the IoT data and then prepro-
cessing the data to make it more robust with the help of a machine learning algorithm
and to increase the detection rate. Thus improving the efficiency of detection.
2 Proposed Methodology
2.1 Dataset Collection and Description
The open-source dataset is collected from the Kaggle, and the dataset’s title is IoT
device [3] design of a WSN platform for long-term environmental monitoring for
IoT applications network logs. In these datasets, there are 477,426 samples, 13 input
features, and one target output name is normality.
Evaluation of Various Machine Learning Algorithms … 399
Feature Names
Frame_number, Frame_time, Frame_len,eth_src,eth_dst,ip_src,ip_dst,ip_
proto,ip_len,tcp_len,tcp_srcport,tcp_dstport, Value.
Target Six Class Names
Normal, man in the middle attack (MITM), distributed denial-of-service (DDos),
data-type probing, scan attack and wrong setup.
53,769 is normal data, 81,250 samples are the wrong setup, 17,507 samples are
DDOS, 78,958 samples are data-type probing, 73,087 samples are scan attack, 58,703
are MITM attack, remaining 73,790 samples are duplicate data, and 40,362 are
outliers.
One of this dataset’s benefits is that each attack becomes more numerous
weightage samples, so machine learning algorithms accumulate extensive data, so
prediction performs very strongly.
2.2 Preprocessing the Data
Preprocessing data means removing noisy data like empty records, duplicates rows
and extreme values. If we do not remove it, our machine learning algorithms read
this unwanted data and predict it wrong, so accuracy decreases.
2.2.1 Statistical Test
Utilizing the statistical test, we recognize any relationship between input and output
data. This step is imperative because these will decide to train the data to an algorithm
or not. If in an input there is no relationship, we must eliminate that input features.
Univariant Test
The univariant test is used to check if each feature has a relationship or not. If it is
below 0.05 of the p-value for each attribute, the association is genuine. If the p-value
is above 0.05, the relationship is worst. In numerical input feature, we use a t-test,
and categorical feature uses one proportion test.
Bivariant Test
The bivariant test is utilized to check feature and target having a relationship or
not. If numerical input (normal distribution) and output are numerical, use Pearson’s
correlation coefficient. Numerical input (non-normal distribution) and numerical
output use the Spearman rank correlation test. Input numerical (normal distribution)
and categorical output use ANOVA test. Input numerical (non-normal distribution)
and category target use Kendall’s rank correlation test. Input categorical (relation-
ship) and numerical output use ANOVA test. Input categorical (relationship) and
numerical output use Spearman rank correlation. When this bivariant analysis occu-
pies below 0.05 of the p-value, use that features for preprocessing. If above 0.05, do
not use that feature for preprocessing because there is no relationship.
Result Of Statistical Test For This Dataset
In the univariant test, we identify our input features numerical, and we perform a
t-test. Hence, all features are non-normal distribution that means above 0.05 of the p-
value. After that, we have numerical (non-normal distribution) input and categorical
output of bivariant test, so we use Kendall’s test, and all features of p-value have
below 0.05, which indicates a relationship. So we do not remove anything.
2.2.2 Identify the Same Feature
High correlation intends the same features of data repeated various times. In our
datasets, we discovered ip_proto, ip_len and frame_len having high correlation
features. Hence, we eliminated two feature parts and used only the frame_len feature
because the same information occupies all three feature parts. When we neglect two-
column data, it does not affect the accuracy, but if we do not remove it, our machine
learning algorithm learned this unwanted data and predicted it wrong.
2.2.3 Drop Duplicate Rows
Identical rows exist on another row place. Suppose we do not discard these duplicate
rows, our machine learning algorithm mug up. This mug is called overfitting. If our
machine mugs up the training data, it predicts well in training, but it does not predict
correctly testing, so obligation removes these repeated records. Our dataset consists
of 73,790 which is a duplicate sample, so we removed that.
2.2.4 Outlier Removal
Outlier removal is an extreme value. It is not always in a specific range. However, it

did not cast off outliers. The detection is predicted wrong. In our dataset observed,
the 40,362 outliers are then eliminated.
2.2.5 Standard Scalar
Standard scalar converted the data from 0 to 1. Many algorithms do not perform well
based on the higher values of data, so standard scalar is fundamental to transform
data from 0 to 1.
2.2.6 Split the Data
We split the data toward training and testing. Training is 70%, and testing is 30% out
of 100% of data. Here training data do not constitute any data in testing, so both are
different.
3 Algorithm Explanation
3.1 Extra Tree Classifier
Extra tree classifier is a classification tree method. It is an advance than a decision

tree [4]. The decision tree splits only two nodes in each feature, but the extra tree
classifier splits the best nodes based on the feature columns. Let us consider output
as a two target. Based on the condition, if extra tree splits with classifying one target
output in one leaf side node and second target in second leaf side correctly for one
feature, that type of feature column is most beneficial. By splitting the features which
is the root node into a number of trees. When both targets form both leaf nodes, then
that is the worst feature.
3.2 Logistic Regression
Logistic regression performs both binary and multiclass classifiers. Using the x-axis,
it predicts the y_axis. Let us consider two targets: logistic regression converts the
input data 0 to 1. Here, assume two targets in the y-axis and one input x-axis, so data
split two sides. If more number of one target is transferred to another side, the model
is defective. Above 0.5 refers to the first target, and below 0.5 refers to the second
target.
3.3 Gaussian Naive Bayes
Gaussian Naive Bayes is a probability measurements algorithm. It is mainly used for

input text classification and also works in numerical input problems. Many Naive
Bayes algorithms are available, like Gaussian Naive Bayes, multinomial Naive Bayes
and categorical Naive Bayes.
4 Performance Metrics
4.1 Accuracy
Accuracy is one metric for evaluating the classification models. It forms accuracy
for all the class that correctly predicted data from the total data. If the actual output
and algorithm expect the result are equivalent, the accuracy was rising.
4.2 Confusion Matrix
Confusion matrix is a specific table layout that allows visualization of an algorithm’s

performance to show the value count for how much data are correctly predicted and
not predicted for all output classes. TP and FN indicate correctly predicted data, and
FP and TN indicate wrongly predicted data.
4.3 Precision
Precision is a model evaluation metric. The accuracy is formed by how much correctly
predicted data are occupying from the actual results.
4.4 Recall
The recall is a model evaluation metric. This accuracy is formed by how precisely
the predicted data occupied by the machine’s expected results.
4.5 F1-Score
F1-score measures the performance of the model. F1-score accuracy has formed a
combination of both precision and recall.
see Table 1.
Comparative Analysis for all Classifiers:
Table 1 Comparison of various performance metrics

Trained Tested Training Time Precision Recall F1-Score
accuracy accuracy (Sec)
Logistic 0.90 0.90 13.71 0.91 0.91 0.91
regression
Extra tree 0.94 0.94 0.09 0.95 0.95 0.95
classifier
Neural 0.88 0.88 152.23 0.93 0.83 0.85
network
Gaussian 0.64 0.64 2.25 0.62 0.63 0.60
Naive Bayes
SGD classifier 0.90 0.90 2.55 0.91 0.91 0.91
Ridge 0.91 0.91 0.21 0.92 0.92 0.92
classifier
As the tested accuracy outcome, extra tree classifier is the Best classifier because
it scored 94% when corresponded to other classifiers. And Gaussian Naive Bayes is
the worst classifier because its score is 64% compared to all the classifiers.
As the analysis result of the training time, the extra tree classifier and ridge clas-
sifier catch 1 s for training to compare all the classifiers. And neural network is
the worst classifier in training time because it exerts 152 s. In final research anal-
ysis of this table, we determined that the extra tree classier is the most trustworthy
classifier to distinguish all the classifier precision, recall and F1-score concerning
91,91,91 scores. The Gaussian Naive Bayes is the worst classifier than all classifiers
for precision, recall and F1-score.
5.1 Confusion Matrix
In the confusion matrix, extra tree classifier is a good algorithm rival to all the
algorithms. Because it has low wrong predicted other attacks. Out of 16,248 test data
of normal class, machine’s 140 samples incorrectly predicted the wrong setup. One
sample wrongly predicted to data-type probing, ten classes wrongly expected as scan
attack, and 77 samples wrongly predicted to a MITM attack. Out of 24,420 sample
test data for the wrong setup, the machine perfectly founded all. Out of 5179 test
data of DDoS, the machine learning algorithm was all correctly established (Fig. 1).
Out of 21,775, machines wrongly predicted 3047 samples. Out of 17,650 test
samples of MITM attack, machine wrongly predicted 5683 samples of scan attack.
Compared to all the confusion matrix classifiers, the extra tree classifiers help less
wrongly predicted samples and more correctly predicted samples.
Fig. 1 a Logistic regression, b multilayer perceptron, c extra tree classifier, d ridge classifier,
e Gaussian Naive Bayes, f stochastic classifier
6 Conclusion
The dataset, which we have analyzed in multiclass attacks hence on evaluating the
various machine learning algorithms, extra tree classifier was able to exactly predict
the various attacks. Since extra tree classifier is much suitable for multiclass attacks
with an accuracy of 94%, with training time it is 1 s based on various parameter
analyses. At the complete analysis, finally we conclude that extra tree classifier is
best suited for IoT network log dataset.
7 Future Scope
In the future viable for interpolating machine learning in the hardware of analysis
signal after detecting the attack, the alarm is working. It is the secondary purposes
when the hacker hacks the encrypted data.
References
1. Pahl M-O, Aubet F-X (2018) All eyes are on you: distributed multi-purpose microservice
acquisition. In: At fourteen procedures international conference on network and performance
management (CNSM) (CNSM 2018), Rome, Italy
2. Liu X, Liu Y, Liu A, Yang L (2018) Protecting unwanted attacks using simple test messages on
smart sensors of industrial communication systems, IEEE
3. Lazarescu MT (2013) Design of a WSN platform for long-term environmental monitoring for
IoT applications, IEEE, pp 1–14
4. Deo RC (2015) Mechanical learning in medicine. Distribution 132(20):1920–1930
Performance Analysis
of Slope-Compensated Current
Controlled Universal PV Battery
Charger for Electric Vehicle Applications
S. Ramprasath, R. Abarna, G. Anjuka, K. Deva Priya, S. Iswarya,

and C. Krishnakumar
Abstract The main purpose of the proposed system is to design a low-cost universal
PV battery charger for electric vehicle application. The proposed system is integrated
with a slope-compensated current controller which controls the charging current that
corresponds to maximum power point of the PV module. As an interface converter,
the proposed system consists of a buck converter to control the flow of the charging
current and to find out the reference current Iref from the PV array at MPP. The
battery control circuit is implemented by measuring the state of charge (SOC) of the
battery, and an LCD display has been used to monitor the battery parameters. This
proposed system acts as a smart and efficient PV battery charger for e-vehicles.
Keywords Electric vehicle · Battery charger · State of charge · Slope compensated
1 Introduction
The current and future of energy options were renewable energies. This major
change toward green energy resources has culminated noticeable emission losses
to numerous environmental habitats due to the widespread use of fossil fuels as an
energy supply. From charging a Tesla vehicle to recharging a basic street lamp, the
energy generated from a PV panel can be used in numerous ways. Study in the field of
solar energy has shed light on its integration methods with different electronic power
converters. While the different energy sources are known, it is still a mandate to store
them in batteries for potential and effective use. What we call a renewable energy
reservoir is a PV panel combined with a power converter topology and lithium-ion
batteries.
A universal battery charger for lithium ion and lead acid is proposed. Figure 1
shows overall all block diagram of proposed battery charger, centered on a single
S. Ramprasath (B) · R. Abarna · G. Anjuka · K. Deva Priya · S. Iswarya · C. Krishnakumar

Saranathan College of Engineering, Trichy, Tamilnadu, India
https://doi.org/10.1007/978-981-16-4943-1_38
408 S. Ramprasath et al.
Fig. 1 Overall block diagram
control loop, the slope-compensated current mode control systems with a single-
stage buck DC–DC converter. The proposed system has a SOC measuring module
to avoid overcharge and isolate the battery if needed.
1.1 Methodology of Proposed System
A DC–DC converter is attached to the PV array and the power converter output is
connected to a battery that needs to be charged. It has a control input D to control the
operating point around MPP. As the batteries are the source, the voltage around the
battery terminals is fixed by the battery, then what the charge current would be, so
let’s say we need to put the maximum power from the panel into the battery; hence,
an MPPT that gives a current reference of the PV module is proposed [1, 2].
Generally, a PV panel and a battery cannot be connected directly for a battery
voltage of greater than 12v. So a charge controller is used in between the PV panel
and battery to steer the power flow from PV module between the battery and load.
The charge controller used here is a DC–DC converter (Buck converter). PV panel
are supposed to operate at maximum power point. In order to obtain the maximum
power output of the solar panel, MPPT (maximum power point tracking) technique is
Performance Analysis of Slope-Compensated Current Controlled Universal … 409
used. MPPT technique increases the efficiency of the solar panel, it compensates for
changing the V–I characteristics of the solar panel, and also maximum solar energy is
being utilized from the panel. For duty cycle (d < 0.5), the system is stable. But for (d
> 0.5), the system is unstable, so we need to provide slope compensation [6]. Current
control with slope compensation will give best results, and even if there are errors, the
errors will be removed within a cycle itself. Hence, this slope compensation method
is highly efficient. The battery is efficiently charged with the help of PV array and
Buck converter, and thus, the energy is stored and fed to the load. The charging level
can be sensed and displayed with the help of Arduino. Therefore, this methodology
seems to be more reliable and of low cost for implementation.
The control part of our proposed model will be done by programming the Arduino
for sensing the charging level of the battery. Once the battery has reached its charging
level, it cuts off from the PV array automatically. This can be achieved by coupling
the battery with Arduino also the charging level of the battery can also be displayed
with the help of LCD display.
1.2 Need for Slope Compensated Current Controller
The main purpose of a conventional battery charger is to charge the battery at an

ideal rate and break-off the charging at an appropriate time. On overcharging a
battery, stress is produced in the chemicals inside the battery. Thus, recombination
of chemicals takes place faster than they can. Undercharging a battery can lead to
problems like sulfation. Therefore, both overcharging and undercharging reduce the
lifetime of the battery [4, 5]. Hence, our project aims to design a current controlled
battery charger with an optimum charge profile.
When the duty cycle of the pulse width modulation (PWM) signal is more than
50%, then the current mode design may experience instability (sub-harmonic oscil-
lation). In order to overcome this instability, a technique called slope compensation
is employed to bring back the stable operation across the entire PWM duty cycle
range. Figure 2 shows slope compensated current waveform [8–10].
2 Design Specification
The proposed system is designed to charge a 12 V, 40 AH lead acid battery. The

charging system consists of three modules—battery module, converter module and
PV module.
Fig. 2 I-slope-compensated
current reference
2.1 Module 1: Battery
Table 1 shows the battery specification, the charging current required to charge the
battery is derived from C-rate as.
C-rate (From datasheet) is C10
Max charging current = 40 / 10 = 4A
Charging time = 10 h
Table 1 Lead acid battery

Battery parameters Value
specifications
Type Lead-acid (Solar Tubular Battery)
Nominal voltage 12 V
Capacity 40Ah
Internal resistance 0.5
Fully−charged voltage 14.5 V
C rate C10
2.2 Module 2: DC-DC Buck Converter
The conventional non-isolated Buck converter is used as power modulator for

charging system. The converter was designed for 80 W, with charging voltage as
13 V. From the converter design equations in [3, 7] for the specification show in
Table 2, we design the DC–DC converter parameters [11, 12].
Load current (I o )
Po 80
Io = = = 6.153 A
Vo 13
Duty cycle (D)
Vo 22.7
= = 0.572
Vin 13
Inductor (L)
Vo ∗ (1 − D)
= 0.009H
del I L ∗ f
Capacitor (C)
IL
= 1.183μ f
8 ∗ f ∗ delvo
Input Capacitor (C in )
10
(D − D 2 ) = 10.6μ f
Vin ∗ f
Table 2 Converter
Input voltage (V in ) 22.7 V
specifications
Output voltage (V o ) 13 V
Power (P) 80 W
Del IL 5% of ILavg
Del V o 5% of V o
Frequency 5000 Hz
Table 3 PV Parameters
S.no Parameters Values
1 Rated max power(Pmax ) 100 W
2 Current at Pmax (IMP ) 5.34A
3 Voltage at Pmax (VMP ) 18.78 V
4 Short-circuit current(SC) 5.71A
5 Open-circuit voltage(V oc ) 22.64 V
6 Operating temperature −40 to 85 °C
7 Weight 8.0 kg
8 Max system voltage DC1000
2.3 Module 3: PV Module
100 W PV panel is used for the proposed system, and Table 3 shows the parameters.
3 Simulation Results of Proposed Universal Charger
The proposed system is simulated in MATLAB for the design specification. Figure 3
shows the Simulink block of proposed system.
Figure 4 shows the simulation output for step change in reference current from
3 to 5A at 2 s. The change in battery voltage, output power, load current and duty
cycle was plotted.
Figure 5 shows the variation in charging current for variation in irradiance at 1 s
from 500 to 1000 W/m2.
Fig. 3 Simulink block for proposed system

Fig. 4 a Battery voltage (V), b Output power (W), c Charging current (A), d Duty cycle
Fig. 5 a Irradiation W/m2, b PV current (A), c PV voltage (V), d Power Ppv (W)
The current from PV module is maintained at MPP and the charging current varies
accordingly. Figure 6 shows the battery parameters for step change in irradiance and
charging current. The initial state of charge of the battery is assumed at 70% and the
variation is plotted.
The battery SOC and battery voltage increase as battery charges with constant
current. The battery charging current is shown as negative because the current enters
the battery is termed as negative and vice versa.
4 Conclusion
In this article, the slope-compensated current control technique is extended to

the universal constant current battery charger. The various building blocks of the
Fig. 6 a SOC (%), b Charging current (A), c Battery voltage (V)
proposed system were designed and simulated in MATLAB/Simulink environment.

For a step-change in irradiation and charging current, the converter and battery
parameters are plotted and satisfactory results were found. The work carried out
here is adequate for practical implementations. The proposed universal battery
charger would be manufactured with the requisite battery management scheme, and
a charging station for electric vehicles would be introduced in the future.
References
1. Biswas S, Huang L, Vaidya V, Ravichandran K, Mohan N, Dhople SV (2016) Universal current-

mode control schemes to charge Li-Ion batteries under DC/PV source. In: IEEE transactions
on circuits and systems I: regular papers, Sept. 2016, vol 63(9), pp 15311542. https://doi.org/
10.1109/TCSI.2016.2571218
2. Singh HK, Kumar N (2020) Current controlled charging scheme for off board electric vehicle
batteries from solar PV array. In: 2020 7th International conference on signal processing and
integrated networks (SPIN), Noida, India, pp 935–940. https://doi.org/10.1109/SPIN48934.
2020.9071160
3. Maity A, Patra A, Yamamura N, Knight J (2011) Design of a 20 MHz DC-DC buck converter
with 84 percent efficiency for portable applications. In: 2011 24th international conference on
VLSIdesign, Chennai, India, pp 316–321. https://doi.org/10.1109/VLSID.2011.37
4. Achaibou N, Haddadi M, Malek A (2012) Modeling of lead acid batteries in PV systems.
Energy Proc 18:538–544. ISSN 1876–6102. https://doi.org/10.1016/j.egypro.2012.05.065
5. Sant’Ana WC et al (2019) Implementation of automatic battery charging temperature compen-
sation on a peak-shaving energy storage equipment. In: 2019 IEEE 15th Brazilian power
electronics conference and 5th IEEE Southern power electronics conference (COBEP/SPEC),
Santos, Brazil, pp 1−7. https://doi.org/10.1109/COBEP/SPEC44138.2019.9065670
6. Kondrath N, Kazimierczuk MK (2013) Slope compensation and relative stability of peak
current-mode controlled PWM dc-dc converters in CCM. In: 2013 IEEE 56th interna-
tional midwest symposium on circuits and systems (MWSCAS), Columbus, OH, USA, pp
477−480.https://doi.org/10.1109/MWSCAS.2013.6674689R
7. Sivakumar R, Ramprasath S, Babu PR (2015) Efficency and power packing density improve-
ment for DC-DC boost converter by soft switching techniques. In: 2015 International confer-
ence on circuits, power and computing technologies [ICCPCT-2015], Nagercoil, India, pp 1–7.
https://doi.org/10.1109/ICCPCT.2015.7159504
8. Wiesner, Diez R, Perilla G (2013) Design and implementation of a buck converter with MPPT
for battery charge from solar module. In: 2013 Workshop on power electronics and power
quality applications (PEPQA),Bogota, pp 1–6. https://doi.org/10.1109/PEPQA.2013.6614938
9. Zhong J, Liu S (2011) Design of slope compensation circuit in peak-current controlled mode
converters. In: 2011 International conference on electric information and control engineering,
ICEICE 2011—Proceedings. IEEE, pp 1310–1313. https://doi.org/10.1109/ICEICE.2011.577
8040
10. Hu Y, Wei Y, Wang J, Sun M (2016) Design of slope compensation for a high-efficiency
high-current DC-DC converter. In: Proceedings 13th IEEE international conference solid-state
integration circuit technology (ICSICT), October 2016, pp 1306–1308
11. Matwankar CS, Alam A (2019) Solar powered closed-loop current controlled DC-DC buck
converter for battery charging application. In: 2019 International conference on vision towards
emerging trends in communication and networking (ViTECoN), Vellore, India, pp 1–5. https://
doi.org/10.1109/ViTECoN.2019.8899645
12. Radianto D, Dousoky GM, Shoyama M (2015) Design and implementation of fast PWM boost
converter based on low cost microcontroller for photovoltaic systems. IECON 2015–41st annual
conference of the IEEE industrial electronics society, Yokohama, Japan, pp 002324–002328.
https://doi.org/10.1109/IECON.2015.7392449
Comparative Analysis of Solar PV
Production in Durban to Other Cities
in South Africa
Sanjeeth Sewchurran, Innocent E. Davidson, and Elutunji Buraimoh
Abstract Solar irradiation levels in South Africa compare favourably with the rest
of the world. However, developers and installers in South Africa must know how the
irradiation levels affect solar PV production within different cities. An investigation
was carried out to compare the PV generation production in nine cities situated in each
of the nine provinces in South Africa to understand how solar PV generation varies
in each city. The city of Durban was used for the base case with its solar production
compared to that of other cities. Technical and non-technical factors affecting solar
PV generation production/installations in Durban were then investigated, and a case
study of simulated vs. actual plant production in Durban was compared.
Keywords Solar PV · Specific yield · PV generation production
1 Introduction
Solar PV projects are becoming a common sight in Durban and across South Africa
(SA). This has been driven by rising energy costs and reducing solar PV costs,
leading to very lucrative payback periods. Similarly, there is a need to reduce carbon
emissions in the face of impending carbon tax implementation, changes in regu-
lations, feed-in tariffs, energy shortage, etc. [1]. Studies have shown that SA and
Africa have some of the best Global Horizontal Irradiation (GHI) globally and is
well suited for solar PV projects [2]. Figure 1 shows the map of GHI across SA,
which indicates that the GHI levels vary across the different provinces indicating
solar energy production will vary across provinces. Thus, the best irradiation levels
S. Sewchurran (B)
eThekwini Electricity, eThekwini Municipality, Durban, South Africa
I. E. Davidson · E. Buraimoh
Department of Electrical Engineering, Durban University of Technology, Durban, South Africa
E. Buraimoh
https://doi.org/10.1007/978-981-16-4943-1_39
418 S. Sewchurran et al.
Fig. 1 Global horizontal irradiation across South Africa [4]
can be found in the Northern Cape, while Durban has the lowest irradiation levels.
Durban’s climate is humid, subtropical, with hot summers and mild winters. There
is an average of 2343 h of sunlight a year, with an average of 6.4 h of sunlight a day
[3]. This paper investigates the generation from a PV system in Durban compared to
the production from similar system in different cities. Also, the impacts of technical
and non-technical factors affecting PV are investigated, and a comparative analysis
is presented.
Comparative Analysis of Solar PV Production in Durban … 419
2 Investigation and Analysis
The PV production is simulated for a 1 MW rooftop solar PV system installed in nine

selected cities of Bloemfontein (BFN), Cape Town (CPT), Durban (DUR), Johannes-
burg (JHB), Nelspruit (NLP), Polokwane (PLW), Port Elizabeth (PEZ), Rustenburg
(RSB), and Springbok (SGK) in Provinces of Free State, Western Cape, KwaZulu-
Natal, Gauteng, Mpumalanga, Limpopo, Eastern Cape, North West, Northern Cape,
Free State, and Western Cape, respectively. The PV system consists of 2500 Cana-
dian Solar 400 W solar panels and 20 Huawei 50 kW inverters. The PV production
simulations were carried out for a flat roof, meaning zero degrees’ inclination angle.
In most PV installations, the panels are often installed to take the existing building
roof structure’s inclination and orientation angle. The panels were installed on the
profile of the existing flat roof of the building. The results from the simulations are
shown in Table 1 showing the monthly PV production for the cities.
Table 1 summarises the PV production for the cities, and it can be seen from the
case study that Durban has the lowest annual PV production, whilst Springbok has the
highest annual PV production. This ties up with Fig. 1, which shows that Springbok
has the highest irradiation levels and Durban the lowest. Subsequently, a comparative
analysis is carried out to indicate percentage annual production excess of Springbok
with respect to other cities. Table 1 also shows that the annual PV production in
Springbok is 36.01% more than Durban, 22.63% more than Port Elizabeth, 22.63%
more than Nelspruit, 14.09% more than Cape Town, 10.7% more than Johannesburg,
10.42% more than Rustenburg, 6.86% more than Polokwane, and 4.99% more than
Bloemfontein. Lower annual PV generation production will ultimately affect the
solar PV system’s payback period together with the city’s electricity tariff.
3 Factors Influencing PV Production
3.1 Inclination and Orientation Angle on PV Production
Most roof-mounted PV systems are installed at the angle of the roof inclination and
orientation, except mostly in the case of a solid decked flat concrete roof. In the case
of the flat concrete roof, the designer has flexibility with the PV panel inclination and
orientation; however, factors such as wind loading, parapet walls, shading from adja-
cent buildings, trees, air conditioners, chimneys, etc. have to be taken into account.
A study was then carried out to understand how the inclination and orientation angle
of a PV panel affects the annual PV production in Durban. Several simulations were
carried out for the annual PV production at different panel inclination angles and
orientations to compile Table 2. From the simulations carried out, the highest annual
PV production was then taken as 100%, or maximum generation output which was
for a PV system that was north-facing at an inclination angle of 30 degrees. After
420
Table 1 Monthly and annual PV production results for cities

City BFN CPT DUR JHB NLP PLW PEZ RSB SGK
January 177,510 194,940 139,330 158,430 127,580 164,020 169,710 158,490 202,500
February 147,110 157,550 121,130 135,050 113,170 136,700 138,190 135,080 166,070
March 142,290 146,460 121,010 136,940 128,090 141,360 125,480 138,070 157,280
April 118,580 102,100 96,995 113,500 124,130 121,060 96,726 114,090 121,590
May 104,130 74,229 83,106 106,850 123,330 116,140 77,552 107,750 101,610
June 91,310 60,517 73,820 95,042 123,730 103,070 65,318 95,248 84,760
July 102,480 68,312 80,184 105,460 129,430 111,230 73,406 105,600 94,723
August 124,310 88,792 94,405 124,760 128,480 132,080 92,630 124,910 117,130
September 143,570 118,580 101,850 139,900 127,960 144,140 111,480 140,990 141,120
October 166,630 154,970 118,540 153,870 132,650 153,960 145,240 153,620 175,780
November 172,190 180,490 124,400 155,370 127,200 161,870 162,060 155,760 193,640
December 189,000 198,180 141,410 167,160 126,160 163,990 179,760 166,860 206,570
Total 1,679,110 1,545,120 1,296,180 1,592,332 1,511,910 1,649,620 1,437,552 1,596,468 1,762,773
S. Sewchurran et al.
Table 2 Percentage of annual generation production for a rooftop PV system in Durban [5]
Orientation
South South East East North East North North West West South West South
-180 -165 -150 -135 -120 -105 -90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90 105 120 135 150 165 180
90° 22 24 28 33 40 46 52 57 60 62 64 64 64 64 64 62 60 57 52 46 40 33 28 24 22 90°
80° 26 28 32 39 46 53 59 64 68 71 73 74 75 74 73 71 68 64 59 53 46 39 32 28 26 80°
70° 32 33 37 44 51 59 66 71 76 79 82 83 84 83 82 79 76 71 66 59 51 44 37 33 32 70°
60° 38 39 43 50 57 65 71 77 82 86 89 91 91 91 89 86 82 77 71 65 57 50 43 39 38 60°
50° 45 46 50 56 63 70 76 82 87 91 94 96 96 96 94 91 87 82 76 70 63 56 50 46 45 50°
40° 53 54 57 63 69 75 80 86 90 94 97 99 99 99 97 94 90 86 80 75 69 63 57 54 53 40°
32° 60 61 64 68 73 78 83 88 92 95 98 99 100 99 98 95 92 88 83 78 73 68 64 61 60 32°
Inclination
Inclination
Comparative Analysis of Solar PV Production in Durban …
30° 62 63 66 69 74 79 84 88 92 95 98 99 100 99 98 95 92 88 84 79 74 69 66 63 62 30°

20° 72 72 74 76 79 83 86 89 92 95 97 98 98 98 97 95 92 89 86 83 79 76 74 72 72 20°
10° 80 81 81 82 84 86 87 89 91 92 93 94 94 94 93 92 91 89 87 86 84 82 81 81 80 10°
0° 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 0°
421
that, all other simulation results were then expressed as a percentage of the maximum
production.
In order to then understand how the annual PV production for the simulations done
in Table 1 for Durban change by changing the inclination angles and orientation
of the PV panels, we derived Table 3 and annual PV production at 0° inclination
angle versus 30° north-facing inclination. In Durban, 1296.2 MWh and 1491 MWh
of power are generated at 0° and 30° inclination angle, respectively. Based on the
inclination and orientation angle of the actual PV panels installed on a roof, we then
select the corresponding factor from the table and multiple it by the annual production
from Table 1 for the actual inclination angle and orientation of the system installed in
Durban. Table 3 shows the annual PV production when the solar system is inclined
at 30° north facing will give us the best overall annual PV production from a fixed
mounted solar system. This increases the annual PV production by 15%.
3.2 DC:AC Ratio of the Solar PV System
To further increase the annual production from an installed PV system, one can
increase the size of PV system DC rating compared to the system AC rating. This
typically means that we can add more panels, the sum of which will give us the system
DC rating. In comparison, the sum of the inverter ratings will give us the AC rating.
The ratio of the two will give us the DC:AC ratio. Table 4 shows the simulation results
from several installed rooftop solar PV systems in Durban, showing the DC rating,
AC rating, annual production, and the calculated specific yield. It must be noted
that these systems were installed to take the inclination angle and orientation of the
existing building roof structures, while many of the PV systems faced a number of
different orientations to avoid the effects of shading and take the shape of the building
roof. Specific yield is the ratio of the total annual energy generated from a PV system
divided by the DC installed plant’s size. Annual production is the total amount of
electricity generated from the PV system over a year. Installed capacity is the sum
of the number of solar panels installed for the PV system.
The production can be increased substantially by increasing the DC:AC ratio.
Thus, new PV panels are also designed to allow for higher inverter voltage input of
up to 1500 V, and newer inverters are also designed to allow for a DC/AC ratio of up
to 1.5. This means that for 1.5 MWp of PV panels, 1 MW of inverters can be installed
and the system will operate safely. However, it must be noted that factors such as the
panels’ inclination angle, orientation, shading effects from the building, cleanliness
of the panels, annual production degradation of the panels, etc. will influence the
DC/AC ratio when designing the rooftop solar PV system.
Table 3 Factor for annual generation production for a rooftop PV system in Durban (North–South)
Orientation
South South East East North East North North West West South West South
-180 -165 -150 -135 -120 -105 -90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90 105 120 135 150 165 180
90° 0.25 0.28 0.32 0.38 0.46 0.53 0.60 0.66 0.69 0.71 0.74 0.74 0.74 0.74 0.74 0.71 0.69 0.66 0.60 0.53 0.46 0.38 0.32 0.28 0.25
80° 0.30 0.32 0.37 0.45 0.53 0.61 0.68 0.74 0.78 0.82 0.84 0.85 0.86 0.85 0.84 0.82 0.78 0.74 0.68 0.61 0.53 0.45 0.37 0.32 0.30
70° 0.37 0.38 0.43 0.51 0.59 0.68 0.76 0.82 0.87 0.91 0.94 0.95 0.97 0.95 0.94 0.91 0.87 0.82 0.76 0.68 0.59 0.51 0.43 0.38 0.37
60° 0.44 0.45 0.49 0.57 0.66 0.75 0.82 0.89 0.94 0.99 1.02 1.05 1.05 1.05 1.02 0.99 0.94 0.89 0.82 0.75 0.66 0.57 0.49 0.45 0.44
50° 0.52 0.53 0.57 0.64 0.72 0.80 0.87 0.94 1.00 1.05 1.08 1.10 1.10 1.10 1.08 1.05 1.00 0.94 0.87 0.80 0.72 0.64 0.57 0.53 0.52
40° 0.61 0.62 0.66 0.72 0.79 0.86 0.92 0.99 1.03 1.08 1.11 1.14 1.14 1.14 1.11 1.08 1.03 0.99 0.92 0.86 0.79 0.72 0.66 0.62 0.61
30° 0.71 0.72 0.76 0.79 0.85 0.91 0.97 1.01 1.06 1.09 1.13 1.14 1.15 1.14 1.13 1.09 1.06 1.01 0.97 0.91 0.85 0.79 0.76 0.72 0.71
Inclination
Comparative Analysis of Solar PV Production in Durban …
20° 0.83 0.83 0.85 0.87 0.91 0.95 0.99 1.02 1.06 1.09 1.11 1.13 1.13 1.13 1.11 1.09 1.06 1.02 0.99 0.95 0.91 0.87 0.85 0.83 0.83
10° 0.92 0.93 0.93 0.94 0.97 0.99 1.00 1.02 1.05 1.06 1.07 1.08 1.08 1.08 1.07 1.06 1.05 1.02 1.00 0.99 0.97 0.94 0.93 0.93 0.92
0° 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
423
Table 4 Simulation results

kWp DC kW AC Annual production Specific yield
of various projects in Durban
(MWh) (kWh/kWp)
1041.15 948 1,344,200 1291.1
1199.55 1000 1,494,778 1246.1
1201.2 960 1,500,521 1249.2
1155 966 1,524,466 1319.9
1149.75 986 1,550,117 1348.2
1227.6 988 1,580,295 1287.3
1211.76 1000 1,622,300 1338.8
1313.4 1000 1,699,563 1294.0
1344 1000 1,788,460 1330.7
1368 1000 1,808,797 1322.2
1428.48 1000 1,900,000 1330.1
3.3 PV Panel Performance Degradation
A solar panel performance degrades every year, and an example of a manufacturer

performance warranty for Longi LR4-60GBD solar panel is explained. As per the
performance warranty offered by Longi, a 98% performance is guaranteed in year
1, 93.95% in year 10, 91.7% in year 15, 89.45% in year 20, 87.2% in year 25, and
84.95% in year 30. For an actual solar PV installation, this will be the maximum
degradation that can be expected using these panels.
3.4 Shading of Panels
When installing solar panels, an assessment needs to be taken to avoid the shading of
panels. Shading can be either temporary, such as the case of trees that may grow and
cast shadows on the panels. Trees and their branches can periodically be trimmed to
prevent shading, or permanent such as high-rise buildings where the effects cannot
be amended. Shading of the panels results in a reduction in the annual generation
production. If one panel on a string is shaded, it reduces generation on the entire
string. Solar panels come with built-in bypass diodes that can assist with bypassing
partial shading on a panel. The methods to deal with shading will be to use an inverter
with multiple maximum power point trackers (MPPT) where strings which may be
shaded can be installed on a separate MPPT or the use of an inverter which uses
power optimisers installed per panel such as solar edge. In this case, only the shaded
panel production will be affected and not the rest of the panels on the string.
(a) (b)
Fig. 2 a Effect of cloud cover on a 1 MW PV generation system b Effect of rain on a PV generation
profile
3.5 Site Climatic Conditions
Several climatic conditions will affect the PV generation; such as cloud cover in the
area. When clouds cover the sky, they reduce the amount of the sun’s irradiation
reaching the solar panel resulting in a reduction in the generation. An example of the
effects of cloud cover on PV production is shown in Fig. 2a for an installed 1 MW PV
system in Durban, which shows how cloud covers affect solar system generation. This
results in rapid decreases and increases in generation production. The PV production
on a rainy day is also drastically reduced since it will be cloudy when it is rainy. An
example of the PV production from a 1 MW PV system performance on a rainy day
in Durban is shown in Fig. 2b.
The operating temperature of the solar cell also affects its power output. Solar
panels are rated at standard test conditions, which is 25 °C and 1000 W/m2 . The
effects of temperature on the panel operating voltage and current has been shown
in the literature; thus, the panel power output is equal to the sum of the voltage
and current. Hence, the reduction of the panel voltage results in a reduction of the
panel power output. Similarly, the effects of irradiation levels on the panel power
output; as the irradiation levels reduce, so does the PV panel power output. The
irradiation levels are affected by several factors, including clouds, dust, shading, etc.
Figure 3a shows how the irradiance vs. PV generation production for a 1 MW solar
PV installation in Durban on a clear day whilst Fig. 3b shows the irradiance versus
PV generation for the same system on a cloudy day in Durban.
3.6 Cleanliness of Panels
The cleanliness of the PV panels affects the generation production from the panel.
The amount of PV production produced by the PV cell depends on the amount of
irradiance that reaches the PV cell. Like the dirt, soil, dust, leaves, salt sprays (in
Fig. 3 a Clear day irradiance vs. generation output for a 1 MW PV plant b Cloudy day irradiance
vs. generation output for a 1 MW PV plant in Durban
coastal installations) builds upon the PV panel’s glass surface, the panels’ production
starts to reduce due to lower irradiance levels reaching the PV cell. In the case of
panels mounted on an inclined roof, rain assists with washing away dirt and dust from
the panels. Most inverters have software that monitors the strings’ PV production or
even some at the individual panel level, which then allows the owner to see when the
panel production starts reducing and the panels are in need of washing.
3.7 System Losses
In a PV system, there a number of losses that are experienced which includes the
following:
1. DC cable losses: depending on the length of the DC cable run to the inverter
will affect the losses that incur.
2. Inverter losses: Most inverters are highly efficient and generally exceed 98%
efficiency under STC, but the efficiency reduces when the inverter temperature
increases. This can be mitigated by placing the inverter in a substation that
is cooled by an air conditioner. Figure 4 shows the efficiency curve for an
SMA Sunny Tripower Core 1 STP 50–40 inverter. Figure 4a indicates that the
efficiency is also a function of the power output, with lower efficiency at below
20% power output compared to greater than 20%.
3. Clipping losses: This occurs in systems with much larger DC:AC ratios and
occurs when the DC power fed into the inverter is greater than the inverter’s AC
rating. Figure 4b shows an example of a larger DC panel installation compared
to the 1 MW inverter rating and the result of clipping of generation due to the
inverter rating. Most PV systems will generate maximum energy during the
summer months, and depending on the DC:AC ratio, and clipping may most
likely only occur during the summer months. Installers also need to consider
that the production from the PV system degrades over time coupled with the
cleanliness of the panels, etc. Therefore, clipping may only occur in the first
Fig. 4 a Efficiency curve for a SMA Sunny Tri Power STP 50–40 inverter b Graph of clipped
energy for a 1 MW PV system [6]
few years after installation. Most modern inverters can allow up to 1.5:1 DC:AC
ratio and will clip off the energy exceeding its rating safely with no damage to
the inverter.
4. PV panel mismatch: This is the difference in performance between panels on
the string. This results in mismatch losses on the string.
5. Reflection: Solar radiation reflected by the surface/collector/module will not
hit the cell and therefore cannot contribute to the generation of electricity.
6. Efficiency of the PV panel: The efficiency of a solar panel refers to the amount
of energy it produces from the maximum produced from the amount of irra-
diance that strikes the panel. There have been numerous solar PV technology
developments in recent years, such as bi-facial solar panels, passivated emitter
and rear contact cells, half-cut cells, and multi-junction concentrator solar cells.
This has improved solar panels’ efficiency currently on the market, offering
efficiencies of up to 23.81%, such as Trina Solar panels, whilst labs such as the
National Renewable Energy Lab have obtained higher efficiency up to 47.1%
using multi-junction concentrator solar cells. However, these are not currently
commercially available.
3.8 Grid Availability
Grid-tied solar PV systems need the grid as a reference in order to generate elec-
tricity. However, network faults, outages, and load shedding all negatively affect the
availability of the grid. This then in turn, results in no PV generation production
when the grid is not available, resulting in loss of production. Figure 5 shows the
effect of 2-h load shedding on a grid-tied solar PV system in Durban. The effects
of grid unavailability for pure grid-tied PV can be mitigated with the addition of
battery storage or configuring the PV system to operate in parallel with a suitably
sized standby generator.
Fig. 5 Loss in PV production during 2 h’ load shedding
4 Simulation versus Practical Examples of Solar PV

Generation
Table 1 shows the simulation results from a 1:1 DC:AC ratio flat roof solar PV
installation. The monthly generation varies and is dependent on the seasons, with
the highest generation in summer and the lowest generation in winter. Subsequently,
a practical example of a 1 MW PV system installed in Durban was examined and
compare the simulation results of Table 1 to the actual installed solar PV system
monthly production. The DC size of the system was 1,199,550 kW. The AC system
size was 1000 kW. This resulted in a 1.2:1 DC:AC ratio. The panels were inclined at
3° to take the roof profile, which is very close to the flat roof used in the simulation.
Figure 6 shows that the simulation results versus the actual 1 MW installed solar
PV system percentage of the total monthly production. The results reflect the simu-
lation, and the actual results are very close to each other. The largest difference is 1.
53%, which occurred in March 2020. Since the actual plant installed on site is 20%
bigger on the DC side than the simulated plant, we see a 20% higher production in
the installed plant compared to the simulated plant.
S I M UL AT IO N VS A CT UAL P LA NT P R OD UCT I ON
% OF TOTAL GENERATION/MONTH (SIMULATION) % OF TOTAL GENERATION/MONTH (ACTUAL)
10.91
10.87
10.75
10.60
9.97
9.60
9.59
9.35
9.34
9.15
8.91
8.38
7.86
7.81
7.61
7.48
7.39
7.28
7.00
6.41
6.19
6.04
5.83
5.70
O CT-2 0
NO V-2 0
JUN-2 0
SEP -2 0
MAY- 2 0
DEC-2 0
AP R - 2 0
FEB-2 0
MAR - 2 0
JUL-2 0
JAN- 2 0
AUG- 2 0
Fig. 6 Monthly PV Production from a 1 MW PV System
5 Conclusion
The studies have shown that Durban’s PV production is the worst compared to
the other provinces in South Africa. This, however, still compares well with other
countries in the world, such as Germany. A higher DC:AC ratio of the PV system
can be utilised to compensate for additional energy production. When designing and
installing a PV system, it is crucial to consider numerous sources’ potential for losses.
Some could be avoided whilst others were permanent and cannot be avoided. The
effects of panels’ inclination angles and orientation, shading, etc. can be taken into
account, and these have major impacts on the amount of energy produced from the PV
system. Many additional factors need to be considered when deciding to install a PV
system, including technical and financial analysis. A comparative analysis between
the software simulation results and an existing plant in Durban reveals similar results.
This provides us with confidence in the use of software simulation to provide accurate
PV production data.
References
1. National Energy Regulator of South Africa (2015) Small-scale embedded generation: regulatory
rules consultation paper, 25 February 2015
2. Pegels A, (2010) Renewable energy in South Africa: Potentials, barriers and options for support.
Energy Policy 38(9):4945–4954. ISSN 0301–4215
3. Zawilska E, Brooks MJ (2011) An assessment of the solar resource for Durban, South Africa.
Renew Energy 36(12):3433–3438
4. http://solargis.info/doc/free-solar-radiation-maps-GHI. Last Accessed 28 February 2021
5. Sewchurran S (2016) Modelling and performance analysis of the ethekwini electricity distribu-
tion grid with increased embedded generation sources. PhD Thesis, UKZN 2016
6. Serdio Fernández F, Muñoz-García MA, Saminger-Platz S (2016) Detecting clipping in photo-
voltaic solar plants using fuzzy systems on the feature space. Solar Energy 132:345–356. ISSN
0038–092X
An Experimental Analysis of the Impact
of a Grid-Tied Photovoltaic System
on Harmonic Distortion
Abayomi A. Adebiyi, Ian J. Lazarus, Akshay K. Saha, and Evans E. Ojo
Abstract This study presents analyses of the results of selected experimental power
quality (PQ) data obtained on a grid-tied photovoltaic (PV) system composed of
different PV modules technologies and four PV inverters integrating the grid at
the point of common coupling. A PQ analyzer is utilized to collect the data from
the 110 kW PV system for two weeks. The measured PQ data was obtained by
following South Africa and international power quality measurement standards.
The data measured were RMS voltage, RMS current, voltage unbalance, frequency,
voltage total harmonic distortion (THDV ), current total harmonic distortion (THDI ),
active and reactive power. Voltage unbalance, frequency, voltage total harmonic
distortion (THDV ), current total harmonic distortion (THDI ) have been appraised
and juxtaposed with the requirements of the South Africa Renewable Energy Grid
Code. These parameters were all found within the regulated limits, except for the
current total harmonic distortion (THDI ) that exceeded the grid code limit due to
sudden voltage variations. This occurrence is observed primarily at dawn and dusk
when the PV inverter output active power is at the lowest level. This study’s perfor-
mance analysis contributes to evaluating large-scale PV systems inverters’ behaviour
and their impact on the distribution grid power quality at different power generation
levels.
A. A. Adebiyi (B)
Department of Electrical Power Engineering, Durban University of Technology, Durban, South
Africa
I. J. Lazarus
Department. of Physics, Durban University of Technology, Durban, South Africa
A. K. Saha
School of Engineering, Howard College, University of KwaZulu-Natal, Durban, South Africa
E. E. Ojo
Department of Electrical Power Engineering, Durban University of Technology, Durban, South
Africa
https://doi.org/10.1007/978-981-16-4943-1_40
432 A. A. Adebiyi et al.
Keywords Photovoltaic system · Total harmonic distortion · Voltage unbalance
1 Introduction
Electricity generation from photovoltaics (PV) systems, a climate-friendly and clean

energy source, has evolved significantly over the years. It is projected that the PV
technology will provide a significant proportion of energy demand as time goes on
[1]. The Photovoltaic Power System Program of the International Energy Agency
indicated that 75 gigawatts of PV power plants were installed globally in 2016, there-
fore taking the total global PV installed capacity to about 305.5 gigawatts (GW). As
shown in Fig. 1, the global cumulative installed capacity stands at 627.2 GW at the
end of 2019, 82% over the previous year [2]. The rapid increase in this technology’s
deployment is motivated by increasing global warming, implementing renewable
energy standards, government incentives and cost reduction of PV systems. The
intermittency of the power injected by the grid-tied PV system due to the varia-
tion of solar irradiation and temperature on the distribution system could result in
major power quality issues. Thus, power utilities are concerned about these systems’
impacts on the power system [3, 4]. Different research studies have been conducted
to evaluate PV systems’ impact on the low-voltage distribution grid, particularly their
impact on distortions in power quality, voltage control and system security [5].
Conventionally, in the event of a disturbance in the distribution network,
distributed generation must disconnect within 1 s [6, 7]. But with the high-level pene-
tration of the local generation on the distribution grid, a sudden disconnection of a
large portion of the grid will trigger transient disturbances in the network, therefore
degrade the grid stability. Consequently, different countries have come up with new
700
627.2
600
512.3
500
409.1
Gigawatt
400
305.5
300
228.3
200 178.2
138.2
100.8
100 70.9
39.8
8 14.5 22.7
0
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Year
Fig. 1 PV power global cumulative installed capacity

An Experimental Analysis of the Impact of a Grid-Tied … 433
standards to address this issue, like the South Africa Renewable Energy Grid Code
(SAREGC), which provides that the local generation must remain connected to the
grid under specific variations in voltage and frequency [8–10]. The grid code provided
for the concept of Low-Voltage Ride-Through (LVRT) or Fault Ride-Through (FRT)
capability for the PV inverters to ride-through voltage dip up to 0 pu at the point of
common coupling (PCC) for a maximum period of 0.15 s in the event of grid distur-
bance with a variation within the upper and lower limits of 1.1–0.9 pu as depicted in
Fig. 2.
Significant attention has been drawn to voltage and current harmonics’ contribu-
tion by PV systems in the distribution grid’s power quality in many studies [11–14].
It is established from the literature that the inverter technology, solar irradiation level,
temperature and loads determine the harmonic generation of a PV system [14]. The
PV system harmonics have intrinsic and extrinsic effects in that the distortions are
inherently attributable to nonlinearities of the inverter control loop and limited reso-
lution of Pulse Width Modulation. The external impact of the harmonic distortion
is due to PV penetration on an unbalanced and weak grid. In this case, the inverter
control system senses the distorted voltage as a disturbance, distorting the current
waveform curve that the inverter generates [14, 15]. Thus, since South Africa has
abundant solar energy resources, this paper analyzes the power quality impact of the
low-voltage grid-tied 110 kW PV power plant in Durban. The results are compared
to the South Africa Renewable Energy Grid Code required limits.
Fig. 2 The SAREGC voltage ride-through capacity

2 Methodology
2.1 The PV Power Plant Layout
The photovoltaic system analyzed in this work is a 110 kW PV power plant with
different PV panels and inverters. The plant was installed by the Durban Univer-
sity of Technology energy research centre called the KZN Industrial Energy Effi-
cient Training and Resource Centre at the university’s Steve Biko campus. It is the
highest installed PV system capacity currently in operation in the university power
system. The average solar irradiation on the campus is 4.6 kWh/m2 /day; the average
ambient temperature is 26 °C. The wind speed generally varies between 0.83 m/s
and 1.4 m/s. Figure 3 depicts the layout of the grid-tied PV system. The PV array
is installed on a flat roof with a tilt angle of 30° and oriented towards the South.
The panels’ correct tilt helps the PV array generate maximum energy when the solar
irradiation hits it directly to receive the Global Horizontal Irradiance (GHI) [16, 17].
The system has 328 polycrystalline silicon (330 and 340 watts) PV panels connected
in series into eleven strings to form the PV array that is, in turn, connected in parallel
into four different DC-AC inverters. The PV system penetrates the grid via the trans-
formerless inverters. The PV panel’s electrical characteristic is presented in Table 1,
while Table 2 presents the inverter specification.
Fig. 3 The grid-tied PV array

Table 1 Characteristics of
Description Value
the PV panels
(ARTsolar 330–72)
Maximum power (Pmax ) 330 W
Maximum voltage at peak power (V max ) 37.4 V
Maximum current at peak power (I max ) 8.83 A
Open circuit voltage (V oc ) 46.1 V
Short-circuit current (I SC ) 9.3 A
(ARTsolar 340–72)
Maximum power (Pmax ) 340 W
Maximum voltage at peak power (V max ) 37.8 V
Maximum current at peak power (I max ) 9.0 A
Open circuit voltage (V oc ) 46.3 V
Short-circuit current (I SC ) 9.41 A
Table 2 Solar Edge

Description Value
three-phase inverter
SE-27.6 K specification Output
Rated AC power output 27 600 VA
Maximum power output 27 600 VA
Nominal output voltage: line to line/line to 400/230 Vac
neutral
Maximum output current (per phase) 40 A
Frequency 50 ± 5 Hz
Three-phase grid support WYE with neutral
Input
Maximum DC power (STC) 37 250 W
Maximum input voltage 900 Vdc
Nominal DC input voltage 750 Vdc
Maximum input current 40 Adc
Transformerless Yes
3 Methodology
3.1 The PV Power Plant Layout
The photovoltaic system analyzed in this work is a 110 kW PV power plant with
different PV panels and inverters. The plant was installed by the Durban Univer-
sity of Technology energy research centre called the KZN Industrial Energy Effi-
cient Training and Resource Centre at the university’s Steve Biko campus. It is the
highest installed PV system capacity currently in operation in the university power

system. The average solar irradiation on the campus is 4.6 kWh/m2 /day; the average
ambient temperature is 26 °C. The wind speed generally varies between 0.83 m/s
and 1.4 m/s. Figure 3 depicts the layout of the grid-tied PV system. The PV array
is installed on a flat roof with a tilt angle of 30° and oriented towards the South.
The panels’ correct tilt helps the PV array generate maximum energy when the solar
irradiation hits it directly to receive the Global Horizontal Irradiance (GHI) [16]. The
system has 328 polycrystalline silicon (330 and 340 watts) PV panels connected in
series into eleven strings to form the PV array that is, in turn, connected in parallel
into four different DC-AC inverters. The PV system penetrates the grid via the trans-
formerless inverters. The PV panel’s electrical characteristic is presented in Table 1,
while Table 2 presents the inverter specification.
3.2 Standards and Measurement of the Selected Power

Quality Indicators
This study considered the requirement of the standards IEC 61000-4-30, IEC 61000-
3-6:2.0, IEEE 519: 2014, Eskom quality of supply NRS 048-2:2003 and the South
African Renewable Energy Grid Code. The different parameters to identify for quality
of supply analysis include RMS Voltage, RMS current, voltage unbalance, flickers,
voltage total harmonic distortion (THDV ), current total harmonic distortion (THDI ),
frequency, power factor, active and reactive power. The power quality data of the PV
installation was measured with the ELSPEC Blackbox PQ analyzer, in line with the
IEC 61000-4-30 standard; it measures and calculates all the power quality indicators
at the point of common coupling (PCC), as illustrated in Fig. 4. The measured PQ data
was downloaded and extracted using ELSPEC Sapphire software on a computer. The
software was used for data analysis. The technical requirements for power quality
by the South Africa grid code for renewable energy are presented in Table 3 are
juxtaposed with this study’s results.
Total harmonic distortion, the ratio of the sum of all harmonic components’
powers to the fundamental frequency, measures the harmonic distortion contained
in a voltage or current signal. Equations (1) and (2) express the voltage and current
THD.
Vn_RMS
THD = (1)
VFundamental_ RMS

N 2
h=2 In_RMS
ITHD = × 100% (2)
I1,RMS
where: V n_RMS is the summation of harmonic signal frequencies from the 2nd
harmonic and the V Fundamental_RMS is the RMS voltage of the fundamental frequency.
Fig. 4 The schematic diagram of the installation
Table 3 Selected grid code

Description Value
limits for power quality
indicators Voltage THD 8%
Current THD 5%
Voltage unbalance 2%
Frequency 50 ± 5 Hz
Voltage unbalance in a three-phase system is one of the major concerns of distribu-

tion system operators. It is a state in which a poly-phase voltage magnitude fluctuates
or is not within the standard 120° phase angle or both. The extent of this unbalance
is generally expressed as the ratio of the voltage negative sequence component to the
positive sequence component. This measured disparity in phase voltages degrades the
performance and significantly reduces loads’ lifespan on the distribution network. All
the power quality standards reviewed for this study set 2% [18, 19] as the allowable
limit for poly-phase system voltage unbalance, calculated as expressed in Eq. (3).
Maximum Voltage Deviation

VU = × 100
Average Voltage
(3)
V2
Voltage Unbalance (%) = × 100%
V1
4 Measured PQ Data Analysis
As measured on the DUT Low-Voltage switchboard in steady-state on different

operating days, the power quality data is presented in this section. This study choice
to evaluate the PV system impact on the grid power quality using the following
dataset:
• Total Voltage Harmonics (THDV [%])
• Total Current Harmonics (THDI [%])
• Voltage unbalance (U 2 /U 1 [%])
• Frequency (F [Hz])
The data analyzed in Table 4 indicates that the three-phase system voltage total
harmonic distortion content, voltage unbalance and frequency did not violate the grid
code required maximum limit, also depicted in Figs. 5, 6 and 7. On the other hand,
there were unusual values beyond the required limits measured for the current total
harmonic distortion on Phase B and C, as depicted in Figs. 9 and 10, indicating that
some loads draw current with total harmonic distortion above the limit. This occurs
during the PQ event of voltage dip during load shedding and when the PV system
was starting up, this causes the current harmonic distortion to increase exponentially
[20].
Likewise, another factor that could cause the unusual current THD is the level
of active power generated by the PV system, depending on the solar irradiation
levels. During cloudy days, sunrise and sunset, current THD values tend to increase
significantly at low solar irradiation levels since the PV inverter’s active output power
Table 4 Analysis of
Parameter Maximum Minimum
measured PQ data
THDVa (%) 2.04 1.48
THDVb (%) 1.88 1.35
THDVc (%) 2.15 1.55
THDIa (%) 2.74 1.48
THDIb (%) 17.61 1.43
THDIc (%) 22.14 2.85
Voltage unbalance (%) 1.1 0.16
Frequency (Hz) 50.24 49.77
Fig. 5 Voltage total harmonic distortion content
Fig. 6 Measured voltage unbalance

Fig. 7 Measured frequency
is proportional to the inverse of the total harmonic distortion factor. The curves
depicted in Figs. 8 and 9 characterize the RMS current and current THD generated
by the PV inverter as a function of solar irradiation behaviour [20]. However, the
Fig. 8 RMS current on a sunny and cloudy day

Fig. 9 Measure three-phase current total harmonic distortion
current THD significantly decreased towards regulatory limit value with increased
PV system active power output, as shown in Fig. 10.
Fig. 10 Sharp decay of current THD during increased solar irradiation

Fig. 11 The current waveform distortion at dawn
5 Discussion
Figure 10 presents the sharp decay of the current THD obtained after the PV inverter
has tracked the PV output power as the solar irradiation level rises towards the mid-
day. The distortion level decreased to 1.4% to maintain the required limit. Remark-
ably, the harmonic current is higher for low power values. It is visible that the high
content of current THD is obtained during the periods that the current generated is
low, as shown in Fig. 11.
Figure 11 depicts the three-phase current waveform distortion during the period
that the current THD rose above the limit at dawn when the PV inverter begins to
track power generation under low solar irradiation.
6 Conclusion
This study presented the power quality measurement results obtained on a 110 kW
PV system, which contributes extensively to evaluating PV systems’ impact on the
distribution network. For two weeks, under different generation conditions, the PV
inverter’s performance and impact on power quality (PQ) were measured and evalu-
ated. For the analysis, four PQ indicators were selected and juxtaposed with the South
Africa Renewable Energy Grid Code requirements. The voltage total harmonic distor-
tion, voltage unbalance and frequency was found to be within the required limits.
The current total harmonic distortion (THDI ) measured data was observed to have
exceeded the grid code limits. This occurrence is observed primarily at dawn and
dusk when the PV inverter output active power is at the lowest level when the PV
system does not generate power. Although the current THD obtained was signifi-
cantly higher than the standard limits during low power generation, the value does
not pose a problem to the grid due to the grid robustness as the PV output power
is less than the rated output at that period. It is essential to study further the impact
of a large-scale PV system on a weak distribution grid that reflects new inverter
technology with improved response during low power generation.
Acknowledgements The authors are grateful to PQ Flow Solutions, a power quality risk manager,
for providing the ELSPEC power quality analyser used for this study.
References
1. IEA (2019) Solar energy mapping the road ahead, France

2. Kaizuka I, Jäger-Waldau A, Donoso J, Detollenaere A, Wetter JV, Masson G (2019) Trends in
photovoltaic (PV) applications. International Energy Agency
3. Adebiyi AA, Saha AK, Lazarus IJ, Ojo EE (2019) Investigation to determine the impacts of
SPV penetration on an unbalanced distribution grid. In: IEEE PES/IAS PowerAfrica. IEEE,
pp 116–120
4. Adebiyi AA, Lazaru I, Saha AK, Ojo EE (2019) High PV penetration impact on an unbalanced
distribution network. In: IEEE PES/IAS PowerAfrica. pp 706–710
5. Mahela OP, Shaik AG, Gupta N (2015) A critical review of detection and classification of
power quality events. Renew Sustain Energy Rev 41:495–505
6. Antonova G, Nardi M, Scott A, Pesin M (2012) Distributed generation and its impact on power
grids and microgrids protection. In: 65th Annual Conference for Protective Relay Engineers.
pp 152–161
7. Seyedi Y, Karimi H, Grijalva S (2016) Distributed generation monitoring for hierarchical
control applications in smart microgrids. IEEE Trans Power Syst 32:2305–2314
8. Eskom (2014) Grid connection code for renewable power plants (RPPs) connected to the
electricity transmission system (TS) or the distribution system (DS) in South Africa. In: Grid
Code Version 2.8 ed. National Energy Regulator of South Africa (NERSA), South Africa, p 64
9. Marinopoulos A, Papandrea F, Reza M, Norrga S, Spertino F, Napoli R (2011)Grid integration
aspects of large solar PV installations: LVRT capability and reactive power/voltage support
requirements. In: 2011 IEEE Trondheim PowerTech. Trondheim, Norway, pp 1–8
10. Sewchurran S, Davidson IE (2016) Guiding principles for grid code compliance of large
utility scale renewable power plant intergration onto South Africa’s transmission/distribution
networks. In: IEEE international conference on renewable energy research and applications
(ICRERA). pp 528–537
11. Barbu V, Chicco G, Corona F, Golovanov N, Spertino F (2013) Impact of a photovoltaic plant
connected to the MV network on harmonic distortion: an experimental assessment. Univer
Bucharest Electr Eng Comput Sci Ser C 75:179–194
12. Etier I, Abderrazzaq M, Al Tarabsheh A, Saraereh O, Albdour M (2016) The impact of large
scale photovoltaic systems on the harmonic increase in distribution networks. Jordan J Mech
Ind Eng 10:239–244
13. Granja AV, De Souza TM, Sobrinho PM, Santos DFA (2018) Study of power quality at the
point of common coupling of a low voltage grid and a distributed generation system of 7.8
kWp in a tropical region. Energies 11:1–19
14. Block PA, Salamanca HL, Teixeira MD, Dahlke DB, Shiono OM, Donadon AR et al (2014)
Power quality analyses of a large scale photovoltaic system. In: 5th international renewable
energy congress (IREC). pp 1–6
15. Abeyasekera T, Johnson CM, Atkinson DJ, Armstrong M (2005) Suppression of line voltage
related distortion in current controlled grid connected inverters. IEEE Trans Power Electron
20:1393–1401
16. Adebiyi A, Lazarus I, Saha A, Ojo E (2018) Performance analysis of PV panels connected in
various orientations under different climatic conditions. In: Proceedings of the 5th Southern
African Solar Energy Conference (SASEC 2018). pp 46–51
17. Adebiyi AA, Lazarus IJ, Saha AK, Ojo EE (2021) Performance analysis of grid-tied
photovoltaic system under varying weather condition and load. Int J Electr Comput Eng
2088–8708(11):94–106
18. Zolkifri NI, Gan CK, Khamis A, Baharin KA, Lada MY (2017) Impacts of residential solar
photovoltaic systems on voltage unbalance and network losses. In: TENCON 2017–2017 IEEE
Region 10 Conference. pp 2150–2155
19. Liao H, Milanović JV (2017) Methodology for the analysis of voltage unbalance in networks
with single-phase distributed generation. IET Gener Transm Distrib 11:550–559
20. Niitsoo J, Jarkovoi M, Taklaja P, Klüss J, Palu I (2015) Power quality issues concerning
photovoltaic generation in distribution grids. Smart Grid Renew Energy 6:148–164
Genetic Algorithm Based Energy
Management in Microgrid
M. Sadees, Aishwarya Raju, Utsahan Mukherjee, and K. Vijayakumar
Abstract Microgrid gives a viable way to advance sustainable power source use
through conveying multiple distributed generators (DGs) with energy storage systems
(ESSs), loads, control gadgets, and ensure gadgets, which can work in either islanded
mode or lattice associated mode. So as to organize the yield of various DGs and under-
stand the capability of sustainable power source, vitality of the executives, and finan-
cial dispatch of the microgrid is required. Both distributed energy resources (DERs)
and client stacks in microgrids have vulnerability qualities; so, the irregularity of
the breeze speed and sunlight based radiation force is displayed by interim science,
and the interim yield of the breeze turbine and photovoltaic (PV) age framework are
obtained. At that point, a microgrid monetary enhancement model dependent on an
interim advancement technique is proposed.
Keywords Microgrid · DG placement · DG sizing · Genetic algorithm · Cost

reduction
1 Introduction
The lattice associates homes, organizations, and different structures to focal force
sources, which permit us to utilize machines, warming/cooling frameworks, and
gadgets. In any case, this interconnectedness implies that when part of the matrix
should be fixed, everybody is influenced. This is the place a microgrid can help.
A microgrid by and large works while associated with the matrix, yet critically, it
can serve and work on its own utilizing neighborhood vitality age in the midst of
emergency like tempests or force blackouts, or for different reasons. A microgrid
can be fueled by circulated generators, batteries, as well as sustainable assets like
M. Sadees (B) · A. Raju · U. Mukherjee · K. Vijayakumar

Kattankulathur, Chengalpattu, India
A. Raju
https://doi.org/10.1007/978-981-16-4943-1_41
446 M. Sadees et al.
sun powered boards. In [1, 7] the model for calculating the optimal size and place-
ment of DGs in a distribution network is investigated. The modeling of the wind and
solar DGs has been mainly done as illustrated by Kutaiba et al. in [2]. Load flow in
distributed networks by distributed generators as elaborated in [3, 4]. In [5] there has
been explained, a way to integrate solar generators into the planned hybrid network
model efficiently. The increasing penetration of wind energy in the network in [6]
has helped to improve the wind generator model. The methods of improvement of
distributed networks are discussed in [8]. Crucial information on placing multiple
DGs is obtained from [9]. Maximization of benefits from DG placements in distribu-
tion as modeled in [10] has been used. The placement of DGs in a radial distribution
system is elaborated in [11] which helped us in choosing a 33-bus radial system.
1.1 Checking Problems of Dg Fixing
The progress towards microgrid prompts new necessities and difficulties for assur-
ance that should be versatile for varieties in the system topology along with the asso-
ciation of dynamic assets. The assurance challenges, related to microgrids, develop
from adding DG sources to the matrix. The impacts of including DG assets to systems
are abundant, including changing force stream and expanded or diminished shortly
out levels among other harmful impacts. The impacts of DG on the insurance addi-
tionally depend on the sort of DG source utilized. The simultaneous generator can
continue higher deficiency current for longer time frames while the deficiency current
created by an acceptance generator rots quickly. Converter-interfaced DG sources
have short-circuited current creating abilities restricted to roughly 2—multiple times
their appraised current.
2 Proposed Objectives
(a) To design using MATLAB a functional microgrid, powered by solar and wind
power,
(b) To transform a standard IEEE 33 bus system into a microgrid and run load flow
analysis,
(c) To integrate into the microgrid, equations for reduction of losses and emission
reduction.
Genetic Algorithm Based Energy Management in Microgrid 447
Fig. 1 Basic layout of

Microgrid
3 Mathematical Design of Distributed Generators
3.1 System Architecture Modeling and Simulation
The two DG frameworks are connected in AC settings. Separately, every DG produces

its own excellent AC power. The DG units are then associated together to shape a
neighborhood AC miniaturized scale framework.
This is a secluded structure is shown in Fig. 1 Pretty much DG units could be added
to fulfill the nearby need. The interest is constantly provided through the built smaller
scale lattice. Such design needs subordinate help, especially VAR prerequisites. Exact
coordination between the DG units is basic for the coherence of the force stream
into the network. Force trade with existing utility is conceivable. In the following
subsections, the mathematical modeling of the micro grid’s DG systems is presented.
3.2 Photovoltaic (PV) System Mathematical Modeling
In the glassy silicon PV module of Fig. 2, the intricate material science of the PV
cell can be spoken to through the comparable electrical circuit that appeared in the
diagram. For the given specific comparable circuit, many conditions have already
Fig. 2 Equivalent electrical circuit of a PV module
Table 1 Parameters and

Parameters used Value
specifications of solar panel
used Short circuit current 3.92 A
Open circuit voltage (at 25C) 0.585 V
Maximum power (at 25C) 60 W
been inferred, in light of standard hypothesis which permits the activity of one sun
oriented cell which was to be recreated utilizing information from producers or field
tests. The proportional circuit of the diagram, the present conveyed to the outer
weight approaches the present I l produced by the light, lessen the semiconductor
diode current Id as well as the ground thrust present I sh .
The expression is given by:
I = Il − Id − Vo /Rsh (1)
The cell can be represented by a voltage current formula as follows:
V = Vo − R s I (2)
Vo = Voltage on the semiconductor diode and the thrust struggle (V);

Id = DiodeCurrent(A), V = Cell yield voltage (V),
I = Weight (cell) output current (A), I L = Photocurrent(A) (Table 1);
3.3 Modeling of Wind Turbine
The FSWECS is separated into two sections; the electrical part speaking to the
acceptance generator, and the mechanical part speaking to the breeze turbine activity.
The electrical portion of the machine is spoken to by a fourth-request state-space
model and the mechanical part by a second-request framework. Every single electrical
variable and parameter has alluded to the stator. This is demonstrated by the prime
Fig. 3 Induction generator dq equivalent circuit referred to the stator
signs in the machine conditions given underneath. All stator and rotor amounts in
the subjective two-hub reference outline (dq-outline) have appeared and are shown
in the following Fig. 3.
The electrical mathematical model of the presented circuits from the above Fig. 3
is as follows and given as per the satisfying conditions (Table 2):
Vqs = Rs i qs + dϕ qs /dt + ωϕds (3)
Vds = Rs i ds + dϕds/dt − ωϕqs (4)
Table 2 Parameters and

Parameters used Specific value
specifications used
Rotor diameter 6m
Rated output 5KW
Cut in speed 4 m/s
Rated wind velocity 15 m/s
Cut out velocity 37 m/s
4 Problem Formulation
Consideration ought to stay assumed to fixing and measuring distributed generator

assets when arranging a dissemination framework as well as its effect. The improper
area and scope of DG can bring about high genuine force misfortune as well as
diminished dependability. It isn’t prudent to introduce exceptionally high-limit DG
in the system. In choosing the ideal size of DG for least framework misfortune, the
system design and the heaps on the conveyance framework assume a significant job.
So as to satisfy the previously mentioned objectives, the issue is planned to meet two
targets. In the first place, to choose a proper area at which the framework misfortune
gets least; second, to register the ideal DG size for least framework misfortune. We
will be using an IEEE 33 bus Radial Distribution system to perform the tests.
4.1 Selection of Location
The ideal amount of areas is the amount of transports that give the greatest advantages
(reserve funds) for ideal size DGs set at chosen areas. The adjustment in a functioning
control loss of the organization because of the adjustment in dynamic power infusion
at a hub is communicated as:
⎛ ⎞
n
∂ PL /∂ Pi = 2⎝ αi j Pi − βi j Q i ⎠ (5)
j=1

n
n

PL = αi j Pi P j + Q i Q j + βi j Q i P j − Pi Q j (6)
i=1 j=1

LN
Ng
To maximize, F = Ce PLwithout − Ce PLK Tk + Cdg PDGi (7)
k=1 i
Vimin ≤ Vi ≤ Vimax , (8)
Ii j ≤ Ii j sheduled
(9)
At lowest possible losses, energy lessening cost and fixing, procedure, and
preservation costs of DGs are used for assessment of investments.
Fig. 4 Position of Dg In 33 bus radial distribution system after placement of Dg
4.2 Optimal Sizing of Distributed Generators Using Genetic

Algorithm
The GA recreates the organic procedures that permit sequential ages trendy a popu-
lace to adjust to their condition. GAs are unrestrained enhancement techniques that
are typical the developmental adjustment in nature. The benefits of utilizing GAs are
that they want no evidence on angle data around the reaction superficial, they are
impervious to getting caught in neighborhood goals, and they could be utilized for a
varied assortment of enhancement issues (Fig. 4).
4.3 Genetic Algorithm Implementation
Step 1: Decide competitor areas. Position the buses in downward order of their
PLS, and choose the bus with maximum PLS after running the load flow algorithm
to calculate the entire actual control loss and loss compassion factors of different
buses.
Step 2: Input GA control parameter data.
Step 3: Initialize population with random strings and copy into mating pool.
Step 4: Do while generation number is not exactly the most extreme number of
generations picked. Do while populace number is not as much as populace size.
Step 5: Get key for ideal DG extent, least system damage, and reserves.
Step 6: Recurrence Step 1 to Step 5 with DG at the bus found former.
Step 7: Quit.
4.4 After Placement of Dg
The proposed methodology is tested on two widely used 33-bus–32- branches at

different loading conditions. The active power (sum of total connected load and base
system losses) at the source node at the beginning of the period is about 3936 kW.
The base case loss sensitivity results of the six most sensitive buses arranged in
descending order are shown in the following Table 3.
As bus 6, is the most sensitive node, it is selected as the candidate location for DG
placement in the 33-bus system. The optimal DG size of 2380 kW was obtained by
the above mentioned method. A distributed generator combining the solar and wind
output was brought to 2380 kW. A combination of solar panels and wind turbines
gave us the output required to be integrated. On integration of said output at bus 6,
the power loss was found to be 132.64 kW. The power loss was reduced from 216 kW
to 132.64 which is a reduction of 83.36 kW.
5 Energy Output of the Matlab Code
In the Fig. 5. we have obtained two graphs, namely the I-V characteristics graph and
the PV characteristics graph for solar energy power at an ideal temperature of 25
°C. From the I-V graph, we can say that at 3.9 Amperes when there was an ideal
temperature of 25 °C the voltage obtained was approximately around 22 V. When
different temperature data were fit into the system then at the same current of 3.9 A
the different data corresponding to the different temperatures showed a gradual and
drastic decrease in the amount of voltage that was generated.
From this graph, we can conclude that the maximum voltage for solar power was
obtained at 25C when the current was around 3.9 Amperes and for the rest of the data
is kept on decreasing gradually as obtained. From the PV graph we can say that the
maximum power of 60 Watts was obtained at around 18 V approximately and then as
the voltage increased, the power decreased gradually to zero. The maximum power
output was obtained for 25C, whereas for other data corresponding to the required
temperatures we got less power outputs and there was a gradual decrease of power
as the voltage increased as shown.
Table 3 Sensitivity values

Systems bus no Sensitivity values
6 0.06779
2 0.0670792
5 0.0664193
20 0.0573545
8 0.0537697
7 0.0476795
Fig. 5 Output of the code representing solar energy power
5.1 Wind Turbine Output of the Matlab Code
The above obtained graph Fig. 6 is a Power (kW) vs Speed (ms) graph where the
pink line represents the power in an ideal wind machine whereas the light blue line
represents the typical characteristics of a wind generator. The respective starting
points are called cut in wind while the end points are called as cut out wind speed
points. For both the cases as we can see the cut in wind speed is 0 m/s while the cut
out wind speed for ideal case is 40 m/s and that for the typical case is around 37 m/s
approximately. The rated wind speed is between 20—30 m/s. For the ideal case, the
maximum power obtained is 55,000 Watts and that for the typical wind machine case
it is coming around to be 4000 Watts. In both cases, after reaching the maximum
power output the graph flattens as shown and then after a particular point they start
decreasing to their respective cut out wind speeds as shown. The area under the blue
line is called idealized power output. The area between the pink and the blue line is
the area of inefficiencies.
Fig. 6 Output of the code representing wind energy power
6 Conclusion
In this paper, we aimed to reduce the costs and emissions of a microgrid by using
intensive programming techniques. As node 6, is the most sensitive bus, it is chosen
as the optimal area for DG arrangement in the 33-transport RDS. The ideal DG size
of 2380 kW was acquired by the previously mentioned technique. A disseminated
generator joining the sunlight based and wind yield was brought to 2380 kW. A
blend of sunlight based boards and wind turbines gave us the yield required to be
incorporated. On the coordination of said yield at transport 6, the power misfortune
was seen as 132.64 kW. The power misfortune was diminished from 216 kW to
132.64 which is a decrease of 83.36 kW. This spares energy and along these lines
energy cost is advanced. As we have utilized the ideal size of DG we have utilized
sustainable assets to the most extreme potential in the microgrid. In this manner,
emanation costs have likewise been advanced. If we assume 83.36 kW of energy is
saved per hour per day, by integrating DGs into the microgrid and the cost of one
unit is Rs. 5/unit
Energy consumed in one day = 83.36 ∗ 24 = 2000.64 kWh
Energy consumed in a year = 2000.64 ∗ 365 = 7, 30, 233.6 kWh
Cost of 7,30,233.6kWh = 5 ∗ 7, 30, 233.6 = 36.5 Lakhs

References
1. Hung DQ, Mithulananthan N, Bansal RC (2013) Multiple distributed generators placement in

primary distribution networks for loss reduction. IEEE Trans Ind Electr
2. Kutaiba S, El-Bidairia, Nguyena HD, Jayasinghe SDG, Mahmoud TSB, Penesis I (2005) A
hybrid energy management and battery size optimization for standalone microgrids: a case
study for Flinders Island, Australia. IEEE Trans Power Syst 20:856–791
3. Ackermann T, Anderson G, Soder L (2001) Distributed generation: definition. Elect Power
Syst Res. 57(3):195–204
4. Jóos G, McGillis BT, Ga D, Liana FD, Marceau R (2000) The potential of distributed generation
to provide ancillary services. In: Proceedings of the IEEE power engineering society summer
meeting, seattle. WA
5. Hoff T, Wenger HJ, Farmer BK (1995) The value of grid-support photovoltaic in providing
distribution system voltage support. In: Clean power research conference. Napa, CA, 1995.
6. Rau NS, Wan YH (1994) Optimum location of resources in distributed planning. IEEE Trans
Power Syst 9(4):2014–2020
7. El-Khattam W, Hegazy YG, Salama MMA (2006) Investigating distributed generation systems
performance using Monte Carlo simulation. IEEE Trans Power Syst 21(2):524–532
8. Wang C, Nehrir MH (2004) Analytical approaches for optimal placement of distributed
generation sources in power systems. IEEE Trans Power Syst 19:2068–2076
9. Naik SNG, Khatod DK, Sharma MP (2015) Analytical approach for optimal siting and sizing
of distributed generation in radial distribution networks. IET Gener, Trans Distrib
10. Shukla TN, Singh SP, Srinivas Rao V, Naik KB (2010) Optimal sizing of distributed generation
placed on radial distribution systems. Electr Power Comp Syst
11. Shaaban MF, Atwa YM, El-Saadany EF (2013) DG Allocation for benefit maximization in
distribution networks. IEEE Trans Power Syst
Improvement of Voltage Stability
in Micro Grid System Using Hybrid
Power Flow Controller
R. Suganya and M. Anitha
Abstract Hybrid Power Flow Controller (HPFC) discovers habitation amongst

weak buses in multi-bus systems to condense the properties of voltage sag. This
effort deals with power quality improvement of Four Bus System (FBS) using HPFC.
This effort mainly deals with the simulation of FBS with and without HPFC (Hybrid
Power Flow Controller). The outcomes are assessed in terms of real power, reactive
power, and voltage. The simulation consequences characterize the aptitude of the
HPFC in enlightening the power quality.
Keywords Hybrid Power Flow Controller (HPFC) · FACTS controller · Unified

Power Flow Controller (UPFC) · Four Bus System (FBS) · Transient stability
1 Introduction
Simulation readings for HPFC-based controller for SMIB system were presented
by Swati Bhasin [1]. This effort covered the reviews of HPFC for developing the
accomplishment of power system. Here we used one topology of HPFC which results
in by- 2VSC’s in shunt and series passive component between the tie-lines.
Optimal placement of UPFC and HPFC using optimization technique was
presented by Anjali Atul Bhandakkar [2]. Optimal placement of UPFC and HPFC
was obtained using Genetic Algorithm (GA) based approach. A power flow model of
UPFC and HPFC was developed using constraint equations and objective functions.
A hybrid power flow controller for flexible operation of multi-terminal DC grids
was presented by Kumars Rouzbehi [3]. The idea for proposing HPFC was inspired
by the successful operation of FACTS devices, the FACTS includes static power
electronics-based elements to provide voltage regulation, power control, and load
flow control in the multi-terminal DC (MTDC) grids.
R. Suganya · M. Anitha
Department of Electrical Engineering, Annamalai University, Chidambaram, India
R. Suganya (B)
Department of Electrical and Electronics Engineering, IFET College of Engineering, Villupuram,
India
https://doi.org/10.1007/978-981-16-4943-1_42
458 R. Suganya and M. Anitha
Displaying and use of HPFC in distribution frameworks was introduced by

Behnam Tamimi [4]. A nitty gritty unique model of the HPFC was created and
actualized in PSCAD, portraying and proposing control techniques to appropriately
work this regulator in dispersion framework applications, for example, a compelling
and basic method for firing up the gadget.
Demonstrating and simulation of HPFC actualized on Multi-Machine frame-
work was introduced by Lini Mathew [5]. Novel and financially savvy converter
based FACTS geographies had been proposed which utilize probably existing aloof
parts likewise and in this manner viewed as cross breed in nature. The distinctive
HPFC setups concocted by the examiner had been actualized on a Multi-machine
framework.
Commonsense utilization of the HPFC was introduced by Behnam Tamimi [6].
An outline of the HPFC and its controls was familiarized, and its solicitation to the
Ontario-Canada network was examined. The outcomes indicated the achievability
of utilizing this regulator for blockage alleviation in commonsense lattices, at lower
costs than those of a UPFC.
HPFC consistent state demonstrating, control, and viable application was intro-
duced by Behnam Tamimi [7]. A procedure for control mode exchanging and limit
dealing with in power stream computations was proposed. The OPF model of the
HPFC spoke to all the gadget control and actual cutoff points as limitations in
the numerical definition, so the HPFC can be ideally dispatched as a piece of the
transmission framework control resources.
Near Simulation reads for HPFC and UPFC based regulator for SMIB-framework
was introduced by Swati Bhasin [8]. This exertion characterized the investigation of
execution act of spontaneity in the force framework by looking at the HPFC and
UPFC. Here we utilized 1topology of HPFC which was shaped by 2VSC’s were
associated in arrangement and SVC is associated in shunt through the tie-lines.
Demonstrating and simulation of HPFC actualized on SMIB framework was intro-
duced by Lini Mathew [9]. Novel and financially savvy FACTS geographies had been
proposed which were based on existing gear which utilized static converters. The
diverse HPFC designs concocted by the examiner had been executed on a SMIB
framework and mimicked utilizing MATLAB/SIMULINK.
Examination of HPFC utilizing static burden model under possibility screening
was introduced by Sai Ram Seshapalli [10]. This exertion introduced a correla-
tion of HPFC utilizing static burden displaying on the transmission framework.
Burden stream was brought out on each chose strategy for the base case-load and
pre-possibility conditions. The detailing of a question was set to be poorly adapted
whenever assessed values were entirely helpless to minute changes in info conditions.
MATLAB/Simulink based simulation of an HPFC was introduced by Garima
Aggarwal [11]. Here, HPFC model was planned utilizing programming and inves-
tigation of the force framework security upgrade by executing the HPFC in SMIB
System power framework. The framework likewise had the arrangement of a similar
investigation of the exhibitions of UPFC and HPFC with respect to control framework
security improvement of the framework.
Improvement of Voltage Stability in Micro Grid System … 459
Transformer-less UPFC utilizing the course MLI was recommended by Liu.

Consistent state models of the HPFC for power stream and OPF studies were intro-
duced, considering the various control methods of the gadget. A system for control
mode exchanging and limit dealing with in force stream estimations was proposed.
The OPF model of the HPFC spoke to all the gadget control and actual cutoff points
as requirements in the numerical detailing, with the goal that the HPFC can be ideally
dispatched as a piece of the transmission framework control assets [12]
With the expanding significance of BESS in miniature frameworks, exact demon-
strating assumes a vital part in understanding their conduct. This work researched and
thought about the exhibition of BESS models with various profundities of detail. In
particular, a few models were analyzed: a normal model spoke to by voltage sources.
2 Research Gap
The above literature does not deal with the comparison of performance of 4 bus-MG
systems with and without HPFC. This work proposes HPFC to augment PQ in 4 bus
microgrid system. The above exertions don’t deal with voltage stability enhancement
of 4 bus systems using FOPID/PR/SM controllers.
The single line-diagram of the 4 bus MGS with HPFC is revealed in Fig. 1. The
HPFC is sited amongst buses 3 and 4. The Single line-diagram of the 4 bus MGS
without HPFC is revealed in Fig. 2.
Fig. 1 Single line-diagram of the 4 bus MGS

Fig. 2 Single line-diagram of the 4 bus MGS with HPFC without HPFC
Figure 3 outlines the simulink diagram of 4 bus systems without HPFC. Figure 4
outlines the circuit model of 4 bus system. Figure 5 outlines the -Voltage at bus3
without HPFC and its significance is 0.75*104 V. Figure 6 outlines the current at-
bus-3 without HPFC and its significance is 70A. Figure 7 outlines the Real power
at bus3 without HPFC and its significance is 2.3*105 MW. Figure 8 outlines the
reactive power at bus3 without HPFC and its significance is 5.6*104 MVAR.
Figure 9 outlines the simulink diagram of 4 bus systems with HPFC. Figure 10
outlines the Circuit model of 4 bus system with HPFC. Figure 11 outlines the Voltage
Fig. 3 Simulink diagram of 4 bus systems without HPFC

Fig. 4 Circuit model of 4 bus system
Fig. 5 -Voltage at-bus-3 without HPFC
Fig. 6 -Current at-bus3 without HPFC
at bus3 without HPFC and its significance is 0.8*104 V. Figure 12 outlines the Current
at-bus-3 without HPFC and its significance is 80A. Figure 13outlines the Real power
at bus3 without HPFC and its significance is 3.4*105 MW. Figure 14 outlines the
Reactive power at bus3 without HPFC and its significance is 7.9*104 MVAR.
Fig. 7 Real power at-bus-3 without HPFC
Fig. 8 Reactive-power at bus3 without HPFC
Fig. 9 Circuit diagram of 4 bus system with HPFC
Assessment of voltage, real power, and reactive power with and without HPFC is
outlined in Table1. By using HPFC, Real power is enhanced from 0.240 to 0.336 MW.
Reactive power is enhanced from 0.055 MVAR to 0.076 MVAR; Voltage is enhanced
from 7500 to 8700 V. Hence, the outcome represents that the 4 bus system with HPFC
is superior to 4 bus system without HPFC.
Fig. 10 Simulink diagram of 4 bus system with HPFC
Fig. 11 Voltage at bus3 with HPFC
Fig. 12 Current at bus3 with HPFC
Fig. 13 Real power at bus3 with HPFC

Fig. 14 Reactive power at bus3 with HPFC
Table 1 Assessment of
Case Voltage Real power Reactive power
voltage, real power, and
reactive power with and Without HPFC 7500 0.240 0.055
without HPFC With HPFC 8700 0.336 0.076
5 Conclusion
The 4 bus systems with and without HPFC (Hybrid Power Flow Controller) are
modeled, analyzed, and simulated. The outcomes are assessed in terms of voltage,
real power, and reactive power. By using HPFC, Real power is enhanced from 0.240
to 0.336 MW. Reactive power is enhanced from 0.055 to 0.076 MVAR. Voltage is
enhanced from 7500 to 8700 V. Hence, the outcome represents that the 4 bus system
with HPFC is superior to 4 bus system without HPFC. The present work deals with
the simulation of four bus systems with and without HPFC. The closed loop HPFC
4 bus system with FOPI can be done in Future.
References
1. Bhasin S, Bhargava A, Chaudhary V, Verma S (2019) Simulation studies for hybrid power flow
controller based controller for SMIB system. In: 2019–2nd international conference on power
energy, environment and intelligent control (PE-EIC)
2. Bhandakkar AA, Mathew L (2018) Optimal placement of unified power flow controller and
hybrid power flow controller using optimization technique. In: 2018 IEEE/IAS 54th industrial
and commercial power systems technical conference (I&-CPS)
3. Rouzbehi K, Miranian A, Candela JI, Luna A, Rodriguez P (2014) A hybrid power flow
controller for flexible operation of multi-terminal DC grids. In: 2014 international conference
on renewable energy research and application (IC-RERA)
4. Tamimi B, Cañizares CA (2018) Modeling and application of hybrid power flow controller in
distribution systems. IEEE Trans Power Delivery 33(6)
5. Mathew L, Chatterji S (2015) Modeling and simulation of hybrid power flow controller imple-
mented on Multi-Machine system. In: 2015 2nd international conference on recent advances
in engineering & computational sciences (RA-ECS).
6. Tamimi B, Cãnizares C (2016) Practical application of the hybrid power flow controller. In:
2016 IEEE conference on power and energy society general meeting
7. Bhasin S, Bhargava A, Verma S, Chaudhary V (2019) Simulation studies for hybrid power flow
controller and UPFC based controller for SMIB system. In: 2019 2nd international conference
on power energy, environment and intelligent control (PE-EIC).
8. Mathew L, Chatterji S (2014) Modeling and simulation of hybrid power flow controller
implemented on SMIB system. In: 2014 North American power symposium (N-APS).
Year2014 |-ConferPaper
9. Seshapalli SR (2019) Analysis of hybrid power flow controller using Static load model under
Contingency Screening. In: 2019 IEEE international conference on clean energy and energy
efficient electronics circuit for sustainable development (IN-CCES)
10. Aggarwal G, Mathew L, Chatterji S (2014) MATLAB/-simulink based simulation of a hybrid
power flow controller. In: 2014 fourth international conference on advanced computing &
communication technologies
11. Peng FZ, Liu Y, Yang S, Zhang S, Gunasekaran D, Karki U (2016) Transformer-less unified
power flow controller using the cascade multi-level inverter. IEEE Trans Power Electron
31(8):1077–1084
12. Tamimi CA, Canizares, Battistelli C (2017) Hybrid power flow controller steady state modeling
control and practical application. IEEE Trans Power Syst 32(2):1483–1492
An Intelligent Fuzzy Controlled
Microgrid Fed by Distribution
Generation Sources
Md. Aijaz , T. Muthamizhan , and T. Venkateswarlu
Abstract In this paper, intelligent fuzzy based microgrid fed distribution generating
sources are presented. Optimal power electronic converters (PEC) are employed to
produce and control the performance of PV-wind power. Double fed IG is employed
as wind generator to obtain maximum power wind. Peak power is absorbed from
solar with help of proposed fuzzy MPPT algorithm. The simulation results for steady
state variation of wind speed and dynamic behavior of load variations are presented.
Real/reactive power energy management is managed efficiently with fuzzy control
algorithm. The proposed fuzzy controller scheme provided improved performance
results and solved power quality issues at load side.
Keywords Micro grid · Fuzzy controller · MPPT · PV-Wind system · DFIG
1 Introduction
For rapid growth of population providing electricity to the society is the biggest
challenge in front of electrical engineering. Electricity is acts as a backbone to run
any industry, so implementation of generations growing very fast in every ware. Now
the absorption of power assets has multiplied, so efforts were made to production
of power from RES sources that include wind, solar. Wind strength has emerged
as certainly one of the most important and promising assets of renewable strength.
This needs extra transmission capability and a better approach to preserving system
reliability. Non-conventional energy sources (NCES) are greener and ecofriendly
nature, Production of power throughout the year is possible. It doesn’t need any fossil
fuels to generate power. CES is hydro and thermal. In thermal power generation, raw
material used is a fossil fuel is almost ended by next two decades. For hydro Power
plants, the source is available seasonally so power generation is not throughout the
year. These two power generations inject various fuel gases into the environment
Md. Aijaz · T. Venkateswarlu

Kodada Institute of Technology and Science for Women, Nalgonda, India
T. Muthamizhan (B)
Sri Sai Ram Institute of Technology, Chennai, 44, India
https://doi.org/10.1007/978-981-16-4943-1_43
468 Md. Aijaz et al.
and pollute the atmosphere. These problems are overcome by using RES, which is
greener and ecofriendly nature, and power can be generated throughout the year.
In this paper a Hybrid PV-wind energy power generation with grid connected
converters topology and various control schemes [1–3], to enhance the power quality
is discussed. Basically, the presence of nonlinear loads always prohibits disturbances
and uncertainties in the PS network [4, 5]. If these nonlinear loads are available longer
time period in the PS network, it is very difficult to provide smooth power supply to
the load. When power feeds to this nonlinear load (PEC converters, laptop/mobile
chargers, etc.) it creates harmonics into the power system network, fault conditions,
sudden insertion or removal of load also create disturbance [6, 7] in the power system
network very severe extent. If duration these unstable issues are many that may lead
to damage the load. The major objective of this article is (i) to implement more
groups of hybrid power generating systems in various locations and all of them are
connected to main AC bus/grid through Fuzzy decision-making control system, (ii)
Enhance power quality and reliability with fuzzy control-based demand management
systems to meet the desired power demand, (iii) Extract maximum power from solar
[8, 9], wind and utilized with optimum pricing and (iv) Solve PQ issues of micro
grid when it is fed with nonlinear loads.
2 System Configuration
The system configuration of isolated microgrid fed by RES sources is shown in

Fig. 1. Different topologies are available to get maximum/lower power from PV
[10–12] such as Ripple Correlation Technique (RCT), Perturb & Observe (P&O),
Fig. 1 Structure of microgrid fed by RES sources

An Intelligent Fuzzy Controlled Microgrid Fed … 469
and incremental conductance (IC). These techniques sense the amplitude of the load
current and make changes in input current quickly. The main drawback of this MPPT
technique is its transient response of variation in current is sluggish. This is overcome
by fuzzy MPPT intelligent control algorithm. To maintain constant voltage across
capacitor (Cdc) and maximum utility DC-DC converters are operated with fuzzy
controller.
2.1 Design of PV Cell
Generally, the design of solar cell is characterized by a current source in shunt with
a diode, and two resistors are connected anti parallel to each other. These resistors
control the power generation of solar cells. To get desired output voltage and output
current from PV panel, N number of PV panels are interfaced in series-parallel
combinations, the voltage and current is expressed in terms of mathematically as
given below

n
Vseries = +V j = V1 + V2 + · · · + Vn f (1)
j=1

n
Vseriesoc = V j = Voc1 + Voc2 + · · · + Vocn + forI = 0 (2)
j=1

n
Iparallel = I j = I1 + I2 + · · · + In (3)
j=1
Vparallel = V1 = V2 = · · · = Vn (4)
By default, we are placing the bypass diodes in solar panels to diminish the over
voltages in the system. But it causes an increase in the system cost (Fig. 2).
Fig. 2 Schematic diagram

of a PV Cell
Fig. 3 Structure of fuzzy logic based boost converter
2.2 DC-DC Converters
In generally controlling and switching of series VSC converters by duty ratio this
will expose for satisfactory performance, switches with some delay this causes stress
on the switches as well as life time of the converters in dc-dc converter and getting
the lower power from PV panels. Nowadays controlled switching pattern is done by
fuzzy MPPT algorithm to extract the peak power from the PV module. Fuzzy based
operated DC-DC converters provide This fuzzy based DC converter provides less
oscillatory voltage to the series VSCs improves dynamic performance and overall
efficiency of the PS network. The structure of fuzzy based DC-DC converter is shown
in Fig. 3.
2.3 Modeling of DFIG Based WECS:
In WECS systems, Wind turbine (WT) converts the movement wind speed (kinetic
energy) into mechanical energy, WT coupled With DFIG generator converts mechan-
ical energy into electrical energy. Produced electrical energy from generators is trans-
mitted to a power station (grid). Wind power is abundant in some states, with the
largest wind farms located in Texas. Wind is unique because it carries incentives
for farmers to give parcels of land for building wind turbines, and has the most
potential as far as widespread adoption due to the large areas of land with consistent
wind available to harness. The stator of the DFIG is directly associated with the
grid or load, and rotor is connected to the VSC based converters via a slip rings.
Power converters (A.C to D.C and D.C to A.C) are effectively control rotor speed
and frequency of rotor. The rotor linked converter is compensating the torque ripple
as well as compensating the reactive power. Rotor side converter and stator side
converter are connected via dc link. It controls the real and reactive power of WECS
system, by varying the terminal voltages of rotor circuit of DFIG. Hence it controls
the power factor of whole WECS. The DFIG is a wound rotor induction generator
Table 1 Fuzzy ruleset

Error (e) NB NS Z PS PB
Change in error (e)
NB NB NB NS NS Z
NS NB NS NS Z PS
Z NS NS Z PS PS
PS NS Z PS PS PB
PB Z PS PS PB PB
so it is simple and robust construction. Low cost compared to PMSG generator.

This feature may leads higher advantages such as lower switching losses, harmonics
inserted in grid are less, reduced power rating means is volume of power converters
is less so it more economical and require filtering section also less. By controlling
this dc link to enhance the converter performance, the dc link voltage may be done by
considering handiest percent voltage. In V ab is line voltage at CPI: m is modulation
index is selected as 1 for linearity.
3 Control Scheme
Fuzzy based VSI controller implementation is similar to the fuzzy MPPT algorithm.
In this error is treats as a set of fuzzy rules. These fuzzy sets are provide PI control
parameters by selecting rules, shape of membership, and de-fuzzification. The fuzzy
rules set mentioned in Table 1. The fuzzy logic rationale contrasts with both idea and
substance from conventional multi = valve legitime frame works such as negative
big (NB), negative small (NS), zero (Z), positive big (PB), and positive small (PS).
The actual value of voltage across (Vdcact) CPI point is in contrast with reference
DC voltage (Vdc) that error is optimized with fuzzification then error is rectified send
to the system after de-fuzzification. The structure DFIG control scheme is shown in
Fig. 4.
4.1 Performance of the System with Wind Energy Variation
A 10 KW and 5.5 KVAR constant load is considered as output solar providing fixed
power and wind energy production is varying. Where wind at t = 0.6 s, the speed
is 7 m/sec and at t = 0.6 s as to t=1 s WT speed is varied from 7 m/sec to 8 m/sec
that observed in Fig. 5. WT power variations occurred in between 0.6 to 1 s. The
variation of performance characteristics of simulation results is represented in Fig. 6,
Fig. 4 Control scheme for DFIG connected to wind turbine
Fig. 5 Wind speed profile of the wind turbine
Fig. 6 Wind turbine velocity
7, 8 and 9 respectively. Those are WT velocity, performance of WT, load voltage,

voltage across DC link capacitor, and load power.
Fig. 7 Voltage generated by the DFIG
Fig. 8 Voltage available at the DC link
Fig. 9 Power measured at the load
4.2 Performance of the System with Solar Energy Variation
A 10 KW and 5.5 KVAR constant load is considered as where WE CS system

providing fixed power and PV energy production is varying. Where At t = 0.4 s, the
irradiance of PV cell is 0.8 Kw/m2, and at t = 0.4secas to t = 0.6secs irradiance is
consider is 0.9 Kw/m2, illustrate in Fig. 10. The variation of performance character-
istics simulation results is represented in Fig. 11, 12 and 13 respectively. Those are
irradiance level, power production in PV array, load voltage, and load power.
Fig. 10 Irradiance level to the PV array
Fig. 11 Power generated using solar PV array
Fig. 12 Voltage available at the load
Fig. 13 Power available at the load

4.3 Performance of System with Unbalanced Loads
The performance characteristics of the PV—Wind battery based microgrid system

with nonlinear loads is mentioned below Fig. 14, 15, 16, 17 and 18 respectively. Those
are load voltage, unbalanced phase current (ABC), and load power. The connected
load consists of 2 kW linear load and 8 kW nonlinear load. At t = 3.25 S, the load
of a-phase is disconnected from the network followed by b-phase load at t = 3.46 S.
Even unbalanced loads also by employing proposed technique energy management
is done by solar and wind.
Fig. 14 Three phase load voltage (for unbalanced load condition)
Fig. 15 Phase current-C (measured with unbalanced load condition)
Fig. 16 Phase current-B (measured with unbalanced load condition)

Fig. 17 Phase current-A (measured with unbalanced load condition)
Fig. 18 Power measured at the load side (PL )
5 Conclusion
The three phase PV-wind-battery integrated hybrid power generation with enhanced
reliability and continuity power is modeled with fuzzy logic controllers. The simu-
lation results are more evident that the proposed control schemes work well with
less number of Power Electronic converters and optimal cost. Proposed fuzzy logic
control scheme, proves best management of active and reactive power to load demand
over conventional control approaches. Satisfied performance is obtained, when the
microgrid is operated with constant load and solar/wind energy generation, and
microgrid is connected to the nonlinear loads.
References
1. Muthamizhan T, Ramesh R (2014) Experimental investigations on PV powered SVM-DTC

induction motor without AC phase current sensor. Int J Eng Technol 6(2):842–857
2. Basaran K, Cetin NS, Borekci S (2017) Energy management for on-grid and off-grid wind/PV
and battery hybrid systems. IET Renew Power Gener 11(5):642–649
3. Chen YM, Cheng CS, Wu HC (2006) Grid-connected hybrid PV/wind power generation system
with improved DC bus voltage regulation strategy. In: Twenty-first annual IEEE applied power
electronics conference and exposition, 2006. APEC ‘06., Dallas, TX
4. IEEE guide for selecting, charging, testing, and evaluating lead-acid batteries used in stand-
alone photovoltaic (PV) systems. IEEE P1361/D5, February 2014
5. Javed K, Ashfaq H, Singh R (2019)Application of supercapacitor as hybrid energy storage
device in stand-alone PV system. In: International conference on power electronics, control
and automation (ICPECA). New Delhi, India, pp 1–4
6. Mohanty S, Subudhi B, Ray PK (2016) A new MPPT design using grey wolf optimization
technique for photovoltaic system under partial shading conditions. IEEE Trans Sustain Energy
7(1):181–188
7. Abinaya I, Sundarsingh Jebaseelan SD, Ravi CN (2016) Reactive power compensation in grid
system using STATCOM with closed loop control. Indian J Sci Technol 9(6):1–4
8. Marlin S, Sundarsingh Jebaseelan SD, Padmanabhan B, Nagarajan G (2014) Power quality
improvement for thirty bus system using UPFC and TCSC. Indian J Sci Technol 7(9):1316–
1320
9. Muthamizhan T, Ramesh R (2014) PV powered fuzzy based direct torque controlled induction
motor without AC phase current sensor. Int Rev Electr Eng 9(2):307–314
10. Kollimalla SK, Mishra MK, Narasamma NL (2014) Design and analysis of novel control
strategy for battery and supercapacitor storage system. IEEE Trans Sustain Energy 5(4):1137–
1144
11. Merabet A, Tawfique Ahmed K, Ibrahim H, Beguenane R, Ghias AMYM (2017) Energy
management and control system for laboratory scale microgrid based wind-PV-battery. IEEE
Trans Sustain Ener 8(1):145–154
12. Ramyar A, Iman-Eini H, Farhangi S (2017) Global maximum power point tracking method for
photovoltaic arrays under partial shading conditions. IEEE Trans Ind Elect 64(4):2855–2864
Multiresolution Representation
of SONAR Pipeline Image Using
Pyramidal Transforms
R. Kumudham, P. Sathish Kumar, V. Rajendran, M. S. Jagan Mugesh,

and U. Charan Raj
Abstract This research work is anxious with the issues of identification of objects
placed on seabed captured on underwater sonar images. Image segmentation methods
play a very important responsibility in partitioning dividing the image into segments.
In this paper, we are applying watershed segmentation algorithm for detecting
pipelines in underwater seabed using sonar images. Initially, we applied Gaussian
and Laplacian pyramid for image resolution enhancement, fixing threshold auto-
matically using Otsu threshold determination. Also, morphological operations were
performed for identifying the foreground and background of the image. Followed by
that for marking the regions on the image, we used watershed segmentation algorithm
to detect the pipeline in underwater acoustic images.
Keywords Image processing · Gaussian pyramid · Laplacian pyramid · Otsu

threshold · Watershed algorithm
1 Introduction
Side Scan sonar image analysis is utilized in several underwater applications,

including object localization and also object identification to seabed categorization
and 3-D reconstruction. There are two main approaches in side scan image analysis:
categorization using consistency features and supervised learning and unsupervised
classification and segmentation. Many of the researchers analyzed object detection
in underwater sea and classifying objects into rocks or mines specifies that rocks are
abnormal whereas mines are normal. But the proposed work insists that identifying
pipelines [1, 2] which is buried in underwater sea level that has to be taken out and
replaced by some others pipeline if any cracks or damages were available. The main
objective of this work is
R. Kumudham (B) · P. Sathish Kumar · V. Rajendran · M. S. Jagan Mugesh · U. Charan Raj

Vels Institute of Science, Technology & Advanced Studies (VISTAS), Chennai, India
P. Sathish Kumar
https://doi.org/10.1007/978-981-16-4943-1_44
480 R. Kumudham et al.
• To gather sonar underwater images from specified sources.

• To find the Gaussian pyramid and Laplacian pyramid for image resolution
enhancement.
• To perform Otsu threshold determination method that fixes the threshold value
automatically to detect the regions of sonar images.
• To apply segmentation technique [3, 4] namely, watershed algorithm partitions
the images into foreground and background portions to mark the regions of sonar
images.
• To detect the pipeline in underwater acoustics using sonar images.
Anastasios Stamoulakatos et al. [1] presented an automatic image marginal note
construction that classifies key events of interest on video footage. Here, the inves-
tigator applied transfer learning with Deep Convolutional Neural Network (ResNet-
50), fine-tuned on real-life data from challenging sub-sea environment, with low
lighting conditions, sea-life, vegetation, and sand agitation. The CNN is configured
to execute multi-label image classifications for critical events. The marginal note
performance fluctuates between 95.1 and 99.7% by means of accuracy as 90.4 and
99.4% for F1-Score depending on types of events. Synthetic Aperture Sonar segmen-
tation method by Imen Mandhouj et al. [5] is based on statistical attributes of the
sonar images, decorated by the Mean-Standard deviation (MSD) plane. This segmen-
tation method is automated by using an entropy criterion by Maussang et al. [6]. Huo
et al. [7] found that categorization of objects in underwater on side scan sonar images
using deep learning based transfer algorithm. Here several CNN models were tried
for better performance in finding objects as well as perform classification too.
2 Methodology
In this section, our research work describes how we are segmenting the sonar images
as pipeline using image processing techniques. The proposed workflow is mentioned
in Fig. 1.
2.1 Image Acquisition
The sonar images were collected from some specified resources as Indomer coastal
hydraulics ltd and Edgetech websites to detect the pipeline in underwater submarine
system. From those images, we used only two images to perform pipeline identifi-
cation using image processing techniques. Initially, we have taken the images with
size along with their dimensions as (1131, 693).
Multiresolution Representation of SONAR Pipeline Image … 481
Fig. 1 Proposed framework for pipeline detection
2.2 Pyramid Technique
Then we apply pyramid technique which is a kind of multi scale signal representation
enlarged by image processing as well as signal processing society in that the images
are repeatedly smoothening and sub-sampling. This representation mainly focused
on two categories namely down sampling of images and pyramid up especially to
enhance the images for high resolution.
2.2.1 Down Sampling
The down sampling is otherwise called Pyramid down in which the succession of
low-pass, down-sampled images, [l0 , l 1 … l N ]. Typically assemble with a discrete 1
dimensional kernel h = [h1; h2; h3; h4; h5], (Fig. 2) and a down sampling feature
of two.
The main impetuses of image pyramids are redundancy diminution as well as
image modeling for
• Efficient and effective coding.
• Image enhancement/restoration.
• Image analysis/synthesis.
Fig. 2 Representation of features as down sampling
Fig. 3 Level of Gaussian pyramid represented as down sample procedure
The sonar images were sampled using down sampling procedure with three iter-
ations for enhancing the property of the image. The levels of Gaussian pyramid
expanded to the size of the original image as shown in Fig. 3.
2.2.2 Up Sampling
Similar to down sampling procedure, here the images are decomposing into low pass
as well as high pass bands recursively. Each band of the Laplacian pyramid is the
divergence among 2-adjacent low-pass images of Gaussian pyramid, [I 0 , I 1 , … I N ].
The constructions of Laplacian pyramid along with two levels are illustrated in
Fig. 4.
The Laplacian pyramid with two levels is shown in Fig. 5.
2.3 OTSU Threshold Determination
Otsu is an automatic threshold selection region based segmentation method applied

on images. This approach is a form of global thresholding which depends only on
Fig. 4 Representation of Laplacian pyramid
Fig. 5 Laplacian pyramid with two levels

gray scale value of the image. In 1979 Scholar Otsu proposed Otsu method which
is broadly utilized since it is trouble-free and efficient. Hence, we applied this Otsu
method on sonar images intended to mark the regions for identifying the pipelines
by fixing the threshold automatically.
2.4 Morphological Operations
The word morphology refers to the study of the outward appearance of images
including the composition of objects. The morphological functions can be utilized to
perform some activities in image processing like image enhancement, noise removal,
smoothening the images, thinning, and segmentation. Therefore, we are utilizing
these morphological operations such as open and closed to find the background
and foreground of the image to find the pipeline underwater as a parameter. The
foreground method is intended to identify the deviations if any occurs in image
sequences whereas the background technique that consents foreground image to be
mined for object detection on images especially pipeline detection.
2.5 Watershed Algorithm
The intention of morphological watershed segmentation [8] technique being envi-

sioned input gray level image into its contour depiction sort it out into minimum
intensity of image, catchment basin and watershed boundaries. Therefore, in our
research work, we utilized this algorithm to segment the input sonar image, dense
region of image specified as fewer altitudes also transparent regions have to be taken
as greater altitudes which seems to be contour façade. Hence, this algorithm is well
suited for image segmentation which recognizes affected areas very easily.
2.6 Segmentation Technique
Ivan Aleksi et al. [9] proposed a novel Robust A-Search Image Segmentation (RASIS)
scheme. RASIS scheme signifies a hybrid combination of both region based and
contour based image segmentation methods. Priyadharshini et al. [10] edge based
segmentation method so as to utilizes the morphological operations for categorizing
the edges followed by an object tracing algorithm is proposed. Here they utilized
real time images which were captured through Edgetech 4125 SSS device. The
segmentation technique is a process of partitioning the given input image into several
regions which are called segments. An image is a combination of different pixels.
We cluster the pixels together that have similar attributes using image segmentation.
Fig. 6 Input images for

pipeline recognition in
underwater
Using this technique, we can identify the pipeline in underwater sonar images via
region based segmentation to detect the specified regions on images.
3.1 Input Images
The original images shapes are specified in dimensions as (1131, 693). These images
were utilized for performing underwater pipeline detection are depicted in Fig. 6.
3.2 Original Image Representation in Matrix Form
Primarily the images are represented in pixels since large number of pixels together
formed a digital image. To process the image, we necessitate to represents it in pixels.
Hence, the pipeline input images are mentioned in matrix resolution to make familiar
about image pixels as well as resolution. The input images are specified in a matrix
format which is depicted in Table 1.
Table 1 Input image represented in matrix resolution

255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
255 255 255 255 255 255 255 255 255 255
Table 2 Output image

1 0 0
representation
1 0 0
1 0 0
1 0 0
1 0 0
1 0 0
1 0 0
1 0 0
1 0 0
1 0 0
3.3 Output Image Representation in Matrix Form
Here we are representing the pipeline image output as matrix resolution as 1 and 0.
Note that “1” mentioned that the pipeline is detected in underwater seabed whereas
“0” represents the pipeline is not found in underwater aquatic environment. The
output image resolution is depicted in Table 2.
3.4 Resolution Enhancement and Segmentation Process
Generally, SONAR images are affected by ambient noise, hence pixels carrying
valued data are missing in image. Hence the image is implemented with Resolution
enhancement Pyramidal Transform techniques. First Lower Pyramid Image shape
dimensions as (566, 347). The restored images having only slight variations while
comparing with original images. The dimension of resorted image with high reso-
lution is specified as (1132, 694). Here, we are identifying the pipeline [11] which
is buried in underwater seabed using segmentation techniques [12] from images
Fig. 7 Finding pipeline after segmentation process
Fig. 8 Marking pipeline after Watershed Technique
which will be helpful in finding the cracks and replacing if any damages are avail-
able. After segmentation process using morphology, the pipeline edges are marked
utilizing Watershed Technique illustrated in Figs. 7 and 8 respectively.
3.5 Restored Image Representation in Matrix Resolution
Image restoration is the process on the way to recover an image from corrupted
version frequently vague impression image or unclear or noisy image. After
performing segmentation procedure the pipeline which was buried underwater is
Table 3 Restored image representation in matrix resolution

92 73 43 29 24 22 22 92 73 43
91 73 43 28 23 21 21 91 73 43
91 73 43 28 22 20 20 91 73 43
92 74 44 29 24 22 23 92 74 44
94 75 45 31 26 25 26 94 75 45
95 77 47 32 27 26 27 95 77 47
96 78 48 34 29 27 26 96 78 48
97 79 51 37 30 26 23 97 79 51
97 80 53 38 30 24 20 97 80 53
95 78 49 35 28 23 20 95 78 49
94 75 45 30 25 21 19 94 75 45
92 73 42 27 22 19 18 92 73 42
identified to remove the cracks that make helpful in avoiding eroded underwater soil
(Table 3).
4 Conclusions
In this paper, we detect the pipelines in underwater seabed using image processing
techniques. We applied Gaussian and Laplacian pyramid which performed down
sampling and up sampling procedures to sample the images by different pyramid
levels for enhancing sonar image. Next, we utilized Otsu threshold determination
for marking the regions based on fixing threshold automatically. In addition to that,
we found foreground and background of the image by performing morphological
operations. Finally, we applied segmentation procedure to segment the images to
detect the pipeline in underwater sonar images. In future, based on features we will
detect the buried pipeline on underwater images.
References
1. Stamoulakatos A, Cardona J, McCaig C, Murray D, Filius H, Atkinson R, Bellekens X, Michie

C, Andonovic I, Lazaridis P, Andrew Hamilton Md, Hossain M, Di Caterina G, Tachtatzis C
(2019) Automatic annotation of subsea pipelines using deep learning. Sensors 20:674. https://
doi.org/10.3390/s20030674
2. Peng X-L, Hao H (2012) A numerical study of damage detection of underwater pipeline using
vibration-based method. Int J Struct Stab Dyn 12(03):1250021. https://doi.org/10.1142/s02
19455412500216
3. Tian Y, Lan L, Guo H (2020) A review on the wavelet methods for sonar image segmentation.
Int J Adv Rob Syst 17(4):172988142093609. https://doi.org/10.1177/1729881420936091
4. Wang X, Guo L, Yin J, Liu Z, Han X (2016) Narrowband Chan-Vese model of sonar image
segmentation: a adaptive ladder initialization approach. Appl Acoust 113:238–254. https://
doi.org/10.1016/j.apacoust.2016.06.028
5. Mandhouj I, Amiri H, Maussang F, Solaiman B (2012) Sonar image processing for underwater
object detection based on high resolution system. In: Signal and document processing
6. Maussang F, Chanussot J, Hetet A (2003) Automated segmentation of SAS images using
the mean—standard deviation plane for the detection of underwater mines. In: Oceans 2003.
Celebrating the Past ... teaming toward the future (IEEE Cat. No.03CH37492). https://doi.org/
10.1109/oceans.2003.178236
7. Huo G, Wu Z, Li J (2020) Underwater object classification in sidescan sonar images using
deep transfer learning and semisynthetic training data. IEEE Access 8:47407–47418. https://
doi.org/10.1109/access.2020.2978880
8. Liu F, Fang M (2020) Semantic segmentation of underwater images based on improved deeplab.
J Mar Sci Eng 8:188. https://doi.org/10.3390/jmse8030188
9. Aleksi I, Matić T, Lehmann B, Kraus D (2020) Robust A*-search image segmentation algorithm
for mine-like objects segmentation in SONAR images 11(2). https://doi.org/10.32985/ijeces.
11.2.1
10. Priyadharsini R, Sree Sharmila T (2019) Object Detection in underwater acoustics images using
edge based segmentation method. In: International conference on recent trends in advanced
computing 2019, ICRTAC 2019
11. Jacobi M, Karimanzira D (2014) Multi sensor underwater pipeline tracking with AUVs. In:
2014 Oceans—St. John’s. https://doi.org/10.1109/oceans.2014.7003013
12. Burguera A, Bonin-Fon F (2020) On-line multi-class segmentation of side-scan sonar imagery
using an autonomous underwater vehicle. J Mar Sci Eng
Dual Motor Power Management Strategy
for Plug-in Hybrid Electric Vehicle
Vinoth Kumar Balan and P. Avirajamanjula
Abstract In order to increase the driving range of plug-in hybrid electric vehicle,
dual motor control strategy was proposed. This method will offer various mode of
control and less energy consumption. This dynamic performance of system improved
by using the rule-based algorithm of fuzzy logic. Power demand for vehicle propul-
sion and energy storage system parameter is estimated. The effect of this value based
on the torque requirement for vehicle propulsion is calculated. Based on the total
torque requirement and SoC of the battery, the corresponding mode of operation is
identified by using the rule-based algorithm. The main controller will monitor the
propulsion control and dynamic performance of the system. The power train configu-
ration for dual motor with engine control is designed using Simulink. Most of period
of control, torque requirement for the vehicle propulsion met by the single electric
motor, i.e., vehicle operating in electric mode. The result shows the extension of
driving range and reduced energy consumption.
Keywords Motor · Fuzzy · Plug-in hybrid vehicle · Power management · Driving

cycle · Power split hybrid vehicle
1 Introduction
Rising oil prices is causing a setback in the transport sector. Further increase in fossil
fuel consumption causes air pollution such as global warming and fuel shortages, etc.
Increasing the use of electric vehicles is the best solution for solving these problems
[1–3]. Traveling long distances, refilling fuel is still a problem in fully battery-
operated vehicle. Using more battery to increase battery capacity also increases the
capital cost of the electric vehicle. That is why people are reluctant to buy electric
vehicles.
To address these issues, the electric motor and engine combination propulsion
hybrid architecture was explored. In hybrid electric vehicle, improved control method
enables multi-mode operation and high performance. Based on the condition, this
V. K. Balan (B) · P. Avirajamanjula

Department of Electrical and Electronics Engineering, PRIST University, Thanjavur, India
https://doi.org/10.1007/978-981-16-4943-1_45
492 V. K. Balan and P. Avirajamanjula
system will operate in both electric and conventional mode, i.e., engine mode. So,
through this, extended range of driving and high efficiency can be obtained [4–7].
The hybrid vehicle not having option of battery charging using grid. This limits the
electric mode of operation for long driving condition. The plug-in hybrid vehicle has
attractive feature of grid charging, and also, charging period can be reduced by fast
charging method. The advanced power electronics technology, high performance
electric motors, and better control technology are main key for developing power
train configuration of PHEV.
2 Power Train Configuration
This system comprises dual motor, engine, generator, and CVT-based power train
which was designed for plug-in hybrid vehicle as shown in Fig. 1. The main objective
is to achieve best drive ability and extended driving range. The sizing of the compo-
nents for the power train configuration, control strategy will play major role for
controlling vehicle dynamic and improving the overall performance of the PHEV
[7–9]. The various type of electric traction motor is reviewed for electric vehicle
such as induction motor, brushless DC motor, permanent magnet synchronous motor
and switched reluctance motor, etc. The high rated motors is preferred as perma-
nent magnet synchronous motor for this system. These motor drive has feature of
high-power density, small size, low weight, and high performance [10–12].
Pt = Pice + Pem1 + Pem2 (1)
min
Pem1 ≤ Pem1 ≤ Pem1
max
(2)
Fig. 1 Powertrain configurations for plug-in hybrid electric vehicle

Dual Motor Power Management Strategy for Plug … 493
Table 1 Specification of
Name Value
power train element
Electric motor A power 89 KW
Electric motor B power 60 KW
Engine 160 KW
Battery capacity 180 Ah
Nominal voltage 380 V
Vehicle loaded mass 1400 kg
Gear box 3.52,2.04,1.40,1
Air resistance coefficient 0.01
Rolling resistance coefficient 0.012
Frontal area, Af 2.33 m2
Drag coefficient, Cd 0.26
Air density, ρ air 1.22 kg/ m3
Wheel radius, Rwh 0.32 m
min
Pem2 ≤ Pem2 ≤ Pem2
max
(3)
min
Pice ≤ Pice ≤ Pice
max
(4)
SoCmin ≤ SoC(t) ≤ SoCmax (5)
min
Pdem ≤ Pt ≤ Pdem
max
(6)
where the Pdem is the power demand of the vehicle, Pt is the total traction power,
andPice is the power produced by the ICE engine. Pem1 is the power produced the
min
Motor A, Pem2 is the maximum power of the Motor B, Pem1 is the minimum power
max min
of the Motor A, Pem1 is the maximum power of the Motor A, Pem2 is the minimum
max
power of the Motor B, and Pem2 is the maximum power of the Motor B. SoCmin is
minimum state of the charge of the battery. SoCmax is maximum state of the charge
of the battery. Here the specification of power train elements mention in Table 1.
3 Modes of Hybrid Drive System
The goal of the control strategies is operating vehicle maximum in electric mode and
extending the driving range of the vehicle. Based on the enviroment condition and
vehicle operator control, the vehicle power demand will change. The environmental
condition includes the wind speed, road slope, traffic condition, etc.
Mode
Mode
Mode
Fig. 2 Driving modes of plug-in hybrid electric vehicle
Based on the torque requirement, vehicle operation mode generally classified as

start/stop, acceleration, cruising, de-acceleration, and Idle are shown in Fig. 2 so the
power requirement of each mode will various accordingly [13–15]. Mode of operation
initialized based on the SoC value of the battery and traveling distance of the vehicle.
If SoC ≥ D, the electric mode of operation is decided by the fuzzy controller. The
dual motor power train configuration has the different transmission pattern based
on power demand of the vehicle. The acceleration condition power demand is so
high. In this mode, based on the SoC of the battery and power demand, vehicle
power train will be configured either series or parallel hybrid mode. Acceleration
mode further classified in to three type. In case of heavy load (A), the Medium SoC
level and high-power demand parameter estimated by the controller, so hybrid dual
power transmission is configured; i.e., in this case, motor A and engine work. In
heavy load (C) case, the high SoC and medium power demand required, So Motor A
and Motor B will meet the power demand of the vehicle. The cruising mode further
classified into two types such as Medium Load (A) and Medium Load (B). In the
case of Medium Load (A) SoC is high and power demand less the dual motor power,
then the power train is configured in dual motor operation. When motor operate
in Medium Load (B), the power demand is greater than dual motor power, so the
Motor A and engine operating togeather to meet the power demand. For Start/stop
mode, the vehicle power demand of the vehicle is very low, so the motor A exclusive
power the vehicle. The fuzzy controller will decide the Motor B and Engine rating
Table 2 Operating mode of PHEV

Load condition Motor A Motor B Engine Vehicle mode
Start/stop Off On Off Parking mode
Heavy load (A) On Off On Acceleration
Heavy load (B) On On On Acceleration
Heavy load (C) On On Off Acceleration
Breaking (A) On On Off Regenerative mode
Breaking (B) On Off Off Regenerative mode
Medium load (A) On On Off Cruising
Medium load (A) On Off On Cruising
Light load Off Off On Idle or traffic mode
Table 3 Fuzzy control tables

ESS P dem Engine Motor B
H VL VL VL
H L L L
H M M M
H H H H
M VL VL VL
M L L L
M M M M
M H H H
L VL L VL
L L M VL
L M H VL
L H H VL
of operation. In de-acceleration mode, regenerative power is stored in battery. When

vehicle in idle mode, the Motor B will be engaged (Tables 2 and 3).
4 Simulation
The rule-based system using fuzzy logic controller is designed for various mode
of operation. The input of the controller is vehicle power demand, and SoC of the
battery using this parameter desired mode of operation is identified, and propulsion
element will be controlled for desired power train configuration. The fuzzy control
membership function as shown in Fig. 3.
The variable speed control of the vehicle condition is simulated using Simulink
software. Figure 4a shows vehicle body response for controller dynamic response of
Power SOC
Engine Motor b
Fig. 3 Fuzzy logic for various of driving modes
various load. Figure 4b Engine, Motor A and B shaft rpm response. Figure 4c, d is the
battery loss and temperature result for the various load. The Fig. 4e shows the result
of load power supply by both battery and engine. This pattern of the power train
drive system is decided by the fuzzy rule controller. The time period of 0–5, light
load demand is occurring, the power demand is met by motor B drive system, and
the fuzzy controller enables the dual motor power train after 5 s of the time period, so
battery loss also increasing certain level. Due to high acceleration, the power demand
of the vehicle was increased. So the power split mode was activated and this mode
power train configuration meet the maximum power demand of vehicle. The battery
losses are comparing to conventional vehicle 20% optimally reduced in this power
train configuration. The de-acceleration starts at 11 ms time period. So, the power
demand, i.e., required power of the vehicle for propulsion is calculated, and based
on the fuzzy rule, the controller is optimizing the motor drive. Here engine operate
at minimum torque. So, this vehicle mostly operating linear mode of control. The
result of this optimizing control the SoC of the battery is varying linearly.
5 Conclusion
In both acceleration mode and cruising mode, better dynamic control was achieved
with dual motor power train. This configuration is supporting maximum electric mode
of operation. This system was designed through Simulink and tested for various
dynamic condition. The optimized method of rule-based fuzzy logic controller is
effectively control the power train at various dynamic condition. The dual motor
a b
75
70
Speed km/hr
65
60
55
0 5 10 15 20 Time in Sec
c d
e
30
20
Power (kW)
10
-10
-20
-30
0 5 10 15 20 Time in Sec
Fig. 4 a Vehicle command. b Shaft speed of power train elements. c Battery loss for different load.
d Battery temperature condition. e Load sharing of battery and engine
configured hybrid plug-in vehicle satisfies the main objective of extended driving
range and better drivability.
References
1. Urbina Coronado PD, Ahuett-Garza H (2015) Control strategy for power distribution in dual
motor propulsion system for electric vehicles, vol 2015. Hindawi Publishing Corporation
2. Wang Y, Sun D (2014) Powertrain matching and optimization of dual-motor hybrid driving
system for electric vehicle based on quantum genetic intelligent algorithm, vol 2014. Hindawi
Publishing Corporation
3. Nguyen CT, Walker PD, Zhang N, Ruan J (2020) Efficiency improvement of a novel dual motor
powertrain for plug-in hybrid electric buses. Proc Instit Mech Eng, Part D: J Automobile Eng
234(7):1869–1882
4. Zhang L, Liu W, Qi BN (2020) Energy optimization of multi-mode coupling drive plug-in
hybrid electric vehicles based on speed prediction. Energy 206:118126
5. Wang Y, Zeng X, Song D (2020) Hierarchical optimal intelligent energy management strategy
for a power-split hybrid electric bus based on driving information. Energy 199:117499
6. Jin-Hua X, Guo JF, Peng B, Nie H, Kemp R (2020) Energy growth sources and future energy-
saving potentials in passenger transportation sector in China. Energy 206:118142
7. Goetz M (2005) Integrated.: powertrain control for twin clutch transmissions (Leeds, U.K.:
Univ. Leeds)
8. Wang W, Li Y, Shi J, Lin C (2018) Vibration control method for an electric city bus driven
by a dual motor coaxial series drive system based on model predictive control. IEEE Access
6:4118841200
9. Wang W, Zhang Z, Shi J, Lin C, Gao Y (2018) Optimization of a dual-motor coupled powertrain
energy management strategy for a battery electric bus based on dynamic programming method.
IEEE Access 6:3289932909
10. Ruan J, Walker PD, Zhang N, Wu J (2017) An investigation of hybrid energy storage system
in multi-speed electric vehicle. Energy 140:291306
11. Morozov A, Humphries K, Zou T, Martins S, Angeles J (2014) Design and optimization of
a drivetrain with two-speed transmission for electric delivery step van.In: Proceeding IEEE
international electric vehicle conference (IEVC). Florence, Italy, pp 18
12. Kim H, Kim J, Lee H (2011) Mode transition control using disturbance compensation for a
parallel hybrid electric vehicle. Proc IMechE, Part D: J Automobile Eng 225:150–166
13. Enang W, Bannister C (2017) Modelling and control of hybrid electric vehicles (A compre-
hensive review). Renew Sust Energ Rev 74:1210–1239
14. Naunheimer H, Bertsche B, Ryborz J et al (2010) Automotive transmissions: fundamentals,
selection, design and application. Springer Science & Business Media, Dordrecht
15. Enang W, Bannister C (2016) Robust proportional ECMS control of a parallel hybrid electric
vehicle. Proc IMechE, Part D: J Automobile Eng 231:99–119
Investigation on Power System Stability
Improvement Using Facts Controllers
Gajana Penchalaiah and R. Ramya
Abstract Due to rising demand and constraints on the construction of new lines,
modern power transmission networks are becoming more complex. Following a
disruption, the loss of stability is one of the key issues with such a modern power
system. Transient stability control is a vital principle that ensures the power system’s
stable operation during faults and significant instabilities. FACTS tools have proven
to be very effective in improving controllability and increasing power transfer capa-
bility while retaining the necessary stability margin in a power system transmission
network. The improvement of transient stability of a three-machine, nine-bus power
system using FACTS controllers (SSSC, SVC, and UPFC) is investigated in this
paper. In the MATLAB/SIMULINK environment, the simulation results demonstrate
the effectiveness and solidity of the FACTS devices in improving system transient
stability.
Keywords SVC · SSSC · UPFC · FACTS · Transient stability
1 Introduction
The combination of numerous generators, transmission lines, and different types of

loads makes the power system network complex. As demand for power increases,
some of the transmission lines are overloaded. If a failure occurs during overloading
of a transmission line, the issue of transient stability arises [1]. Transient stability in
the system, even after disruptions, should be managed to keep the system in stable
service. The switching operations and faults in the transmission line are considered
to be disturbances [2]. Due to heavy deviations between the stator and rotor, FACTS
controllers are the latest developed power electronic controllers that are used in the
G. Penchalaiah (B) · R. Ramya

Kattankulathur 603203, India
R. Ramya
https://doi.org/10.1007/978-981-16-4943-1_46
500 G. Penchalaiah and R. Ramya
Fig. 1 Present day solutions for improving power system stability [7]
power system to improve voltage profile, compensation of reactive power thereby

controlling, transient and steady-state stability of the desired power system. Static
VAR compensator (SVC), series compensators (SSSC), combination of series and
shunt controller (UPFC), and others are examples of FACTS controllers [3–8]. Static
VAR Compensator is a FACTS system that flattens the voltage profile by controlling
the voltage at the desired end. This is possible through the compensation of reactive
power [9]. SVSs are preferred over traditional shunt compensators for improving the
performance during transient and steady-state conditions, and it is also used to damp
out the oscillations in power, reducing losses thereof improving transient stability.
SSSC is a FACTS device, which plays a vital role in controlling the power systems.
The role of SSSC includes controlling the load flow, reducing system losses, and
providing good voltage regulation [10]. It also helps in limiting short-circuit currents,
damping the power swings, and thereby regulating transient stability [11].
Unified power flow controller is the most effective one among all the controllers. It
improves the control attributes such as transient, dynamic, and steady-state stability
of the electrical system. It is the product of combining SSSC and STATCOM.
It can control many parameters like phase angle, voltage magnitude, etc. [12]. A
MATLAB/Simulink model is created for a three-machine, nine-bus power system
with a UPFC. The UPFC exhibition is contrasted, and various FACTS devices such
as SVC and SSSC are used individually. According to the simulation results, UPFC
is a strong FACTS device for transient stability studies (Fig. 1).
2 Test System-WSCC Three-Machine Nine-Bus Power

Systems
For transient stability studies, the WSCC system is widely used. The synchronous
machines have voltage regulators, an exciter, and a detailed model of a steam turbine
and governors. A hydraulic turbine and governor (HTG), an excitation mechanism,
and a power system stabilizer are mounted on each of the three devices (PSS). As
Investigation on Power System Stability Improvement Using Facts … 501
shown in Figs. 2 and 3, a single-circuit long transmission line connects a WSCC

three-machine nine-bus power system with series and shunt FACTS units.
Here are the data for the nine buses, three generator systems. The system depicted
in a single line diagram in Table 1 include information about the actual values of the
transmission line parameters. We must assume the system ratings of MVA BASE
100 MVA, f = 60 Hz, bus nominal voltage = 230 kV, maximum voltage = 1.1pu,
minimum voltage = 0.9 p.u, transformer 1 rating 16.5/230 kV, transformer 2 rating
18/230 kV, and transformer 3 rating 13.8/230 kV [4]. When measuring the actual
Fig. 2 WSCC three-machine nine-bus power system
Fig. 3 WSCC’s three-machine nine-bus power system simulation diagram

Table 1 Three-machine nine-bus power system at WSCC: data from the transmission line
parameters
Line Length(km) R1(/km) R0(/km) L1(H/km) L0(H/km) C1(F/km) C0(F/km)
1–4 – – – – – – –
2–7 – – – – – – –
3–9 – – – – – – –
4–5 97.38 0.054 0.162 0.0012 0.0036 9.06×10–9 2.72×10–8
4–6 95.99 0.093 0.281 0.0013 0.004 8.257×10–9 2.477×10–8
7–5 176.71 0.095 0.287 0.0012 0.0038 8.687×10–9 2.606×10–8
9–8 115.56 0.054 0.163 0.0012 0.0036 9.073×10–9 2.721×10–8
9–6 196.41 0.105 0.315 0.0012 0.0036 9.144×10–9 2.743×10–8
7–8 82.5 0.054 0.163 0.0012 0.0036 9.073×10–9 2.721×10–8
values of transmission system using per unit parameters, these equations can be
used to measure transmission length as well as positive, zero sequence resistance,
inductance, and capacitance.
√
X×B
Length = × Velocity of Light in Km per Sec,
2 × 3.14 × f
Wherev = 3 × 108 m/Sec
R1
Transmission line Calculations
R0
Ractual = Rp.u × Rbase
(Vbase )2
Rbase = X base =
Sbase
R1is a Positive Sequence and
R0 = R1 × 3 is a Z er o Sequence Component
SimilarlyL1andL0&C1andC0should be calculated.
As shown in Fig. 3, the simulation focuses on 230 kV, three-machine nine-bus

power systems which are made up of three generators linked together in a transmis-
sion line network. The G1 generator has a limit of 247.5 MVA, the G2 generator
has a limit of 192 MVA, and the G3 generator has a limit of 128 MVA. The corre-
sponding wave forms for the system without FACTS controllers are shown below.
Fig. 4 a WSCC three-machine nine-bus power system load angle of Generator 1. b WSCC three-
machine nine-bus power system load angle of Generator 2. c WSCC three-machine nine-bus power
system load angle of Generator 3
Table 2 Settling time for

Machines Settling time for load angle in sec
load angle in sec for three
machines 247.5 MVA Generator 8.8
192 MVA Generator 7.8
128 MVA Generator 9.2
A three-phase fault is created at t = 6 s for a duration of 100 ms. Fig. 4a–c reveals a
transient behaviour of system at the time of fault created.
As shown in Table 2, a relation is established between the above systems for
improving the stability of the three-machine nine-bus system, based on simulation
results shown in Fig. 4a–c.
3 Performance Analysis of WSCC System with FACTS

Controller
To control the power flow in the lines and to maintain the voltage levels at buses,
the SVC, SSSC, and UPFC are installed separately in the centre of bus 4 and bus 5.
The WSCC three-machine nine-bus systems making use of series and shunt FACTS
devices. Here, UPFC which is the combination of (STATCOM and SSSC); SVC is
Fig. 5 SSSC delta1-2 and

delta1-3 load angle
comparison
equipped with 4 bus and SSSC and UPFC which are equipped with 4 and 5 bus
system.
• Case-i: Inclusion of SSSC
• Case-ii: Inclusion of SVC
• Case-iii: Inclusion of UPFC
Case-i: Inclusion of SSSC
The ‘SSSC’ is a FACTS unit that is used in conjunction with VSI and a transformer
on the transmission line. The SSSC unit, on the other hand, injects voltage into the
line in sequence. Inductive or capacitive reactance may also be used with the voltage
injected. The electric power in the line can be changed by using variable reactance in
series with the line. As a result, the SSSC system is used to dampen the oscillations
that occur after a three-phase fault (Fig. 5).
The above simulation results for synchronous machine of three-machine nine-
bus system shows the response of load angle. At the time of 6.1 s, Fig. 5 reveals a
symmetric three-phase fault for 100 ms. As compared to earlier traditional types that
did not employ a controller, the ‘SSSC’ system is used to suppress power oscillations
in the system.
Case-ii: Inclusion of SVC
As SVC is used to stabilize the network, the value of SVC in fixed susceptance
mode Bref = 0 is used to observe the extreme contingency conditions. Open the
simulation phase and select voltage regulation mode from the SVC block menu.
When the voltage drops below the reference voltage, the SVC will try to compensate
by combining open force on the line (1.0 p.u). If the SVC is unavailable, the selected
SVC reference voltage is equal to the system voltage. The SVC will now be ‘flick
through’ and keeping the voltage together for voltage compensation as it departs
from its reference set point in steady express (Fig. 6).
The load angle plot for synchronized machines in a three-machine nine-bus system
is shown in the diagram above. At a rate of 6.1 s, a three-phase symmetric 100 ms
problem is applied. The power curve exhibits oscillations after a deficiency occur-
rence, which evolves into a stable state. As compared to when the device does not
have a regulator, oscillations are damping prior to applying SVC.
Fig. 6 SVC delta1-2 and

delta1-3 load angle
comparison
Case-iii: Inclusion of UPFC

In a three-machine nine-bus configuration, the simulation results for load angles
are shown in the diagram above. A three-phase 100-ms short-circuit fault is intro-
duced at a rate of 6.1 s. The power curve shows an oscillation after the issue occurs,
which settle to a stable state. When comparing movements with and without a regu-
lator in the system, it is observed that oscillations are damping prior to the application
of UPFC. In order to improve the stability of the nine-bus system, a relationship is
established between the above FACTS devices, as shown in Table 3. Since the UPFC
post-fault time is less than that of the SSSC and SVC, it is presumed that UPFC is the
superior FACT devices for stability improvement over the SSSC and SVC (Fig. 7).
The system without FACTS controller takes 2.8 s to settle to normal oper-
ating conditions after fault occurrence. Whereas, the system installed with FACTS
controllers settles to steady state operating conditions within 1.4 s. This shows the
Table3 Comparison between

FACTS devices Time for delta1_2 to Time for delta1_3 to
FACTS devices
settle (in sec.) settle (in sec.)
SSSC 7.4 7.8
SVC 7.6 7.9
UPFC 6.8 7.2
Fig. 7 UPFC delta1-2 and

delta1-3 load angle
comparison
effectiveness of installing FACTS controllers at the appropriate place in the test

system.
4 Conclusion
The power system stability of a WSCC three-machine nine-bus system with SSSC,
SVC, and UPFC has been explored for the test system of a WSCC three-machine
nine-bus system. The comparative study is made among the SSSC, SVC, and UPFC
controllers to test the dynamic behaviour of the power system in the event of a major
disturbance such as three phase fault. The simulation results show that using the
UPFC device, which has a shorter settling time in post-fault conditions than the
other controllers and it, improves system efficiency significantly.
References
1. Yu S, Chau TK, Fernando T, Savkin AV, Iu HH-C (June 2019) Novel Quasi-decentralized SMC-
based frequency and voltage stability enhancement strategies using valve position control and
FACTS device. IEEE Access
2. Liu T, Liu Y, Liu J, Yang Y, Taylor GA, Huang Z (Sept 2016) Multi- indicator Inference Scheme
for fuzzy assessment of power system transient stability. IEEE 2(3)
3. Liu Y, Wu QH, Kang H, Zhou X (June 2016) Switching power system stabilizer and its
coordination for enhancement of multi-machine power system stability. IEEE 2(2)
4. Chen B-S, Hsu Y-Y (Jan 2007) An analytical approach to harmonic analysis and controller
design of a STATCOM. IEEE 22(1)
5. Liu S, Li G, Zhou M (June 2016) Power system transient stability analysis with integration of
DFIGs based on center of inertia. IEEE 2(2)
6. Hingorani NG, Gyugyi L (2000) Understanding FACTS: concepts and technology of flexible
AC transmission systems. In: El-Hawary M (ed) vol 1. IEEE press, New York
7. Sharma B, Sharma A (May 2015) Power system stability enhancement using FACTS.
IJARCSSE 5(5)
8. Prakash D, Tripathi EVK Enhancing stability of multi-machine IEEE 9 bus power system
network using PSS. IJAREEIE 4(5)
9. Kundar P (1994) Power system stability and control. McGraw-Hill, NewYork
10. Nallagalva SK, Kirar MK, Agnihotri G (July 2012) Transient stability analysis of the IEEE
9-bus electric power system. IJSET 1(3)
11. Sethy SK, Moharana JK (Nov 2012) Design, analysis and simulation of linear model of a
STATCOM for reactive power compensation with variation of DC-link voltage. IJEIT 2(5)
12. Murthy PSR (2007) Power system analysis. BS Publications, Hyderabad
Power Quality Enhancement Using
Interline Dynamic Voltage Restorer
in Renewable Energy System
T. Ahilan, P. Suresh, S. Elam Cheren, and G. Ramya
Abstract Nowadays, renewable energy is attractive due to the depletion of fossil

fuels. The energy requirement in most developed countries is met by solar and
wind energies. In this paper, solar and wind energies are used in a safer way for
power utilization and power quality improvement. Solar energy and wind energies
are utilized by employing custom power devices (CPD). Interline dynamic voltage
restorer (IDVR) is the most advanced CPD device for wind and solar energy utiliza-
tion. IDVR is the combination of two DVRs which protect the sensitive loads in
various distribution feeders. The solar energy source is connected with feeder 1 and
wind energy is connected with feeder 2. If the power requirement in feeder 1 is
more, the feeder 2 aid feeder 1 by compensating reactive power using IDVR and it is
vice versa. IDVR helps to maintain the power continuity and power quality in both
feeder 1 and 2. Simulation of two bus system using solar and wind energy with and
without IDVR is simulated using MATLAB/Simulink and analyzed in terms of real
and reactive power.
Keywords Interline dynamic voltage restorer · Power quality · Photovoltaic ·

Renewable energy
T. Ahilan (B)
Department of Mechanical Engineering, St. Joseph College of Engineering, Sriperumbudur, India
P. Suresh
Department of Electrical and Electronics Engineering, St. Joseph College of Engineering,
Sriperumbudur, India
S. E. Cheren
Department of Electrical and Electronics Engineering, Sri Krishna College of Engineering and
Technology, Coimbatore, India
G. Ramya
Ramapuram, India
https://doi.org/10.1007/978-981-16-4943-1_47
508 T. Ahilan et al.
1 Introduction
Recently, a renewable source of energy plays a vital role in compensating the

depleting of conventional fuel resources. One of the renewable sources of energy
is the photovoltaic system. Using this technique, PV arrays convert solar energy into
electrical energy. Based on the development of PV systems, PV arrays efficiencies
have been increased. So, the PV system studies have been at the forefront [1–3]. In
remote areas, standalone PV generation is an attractive essential source of energy.
The PV system has two disadvantages 1. Weather conditions vary the PV power
generation and 2. The energy conversion range is minimal during low irradiation
conditions. Though the PV system has demerits, two different techniques can be
employed for effective utilization. Firstly, through space heating applications, solar
thermal energy is obtained using captured heat. Secondly, incident solar radiation
is converted into useful electrical energy [4–6]. MPPT technique is employed in all
power inverters to extract maximum power. Incremental conductance (INC), pertur-
bation and observation (P&O) and fuzzy logic controller are the different techniques
of MPPT [7–9]. P&O algorithm-based MPPT technique has modest structure and
limited parameters. It measures PV voltage using a voltage sensor, and the cost of
the sensor is very less, so it is easy to implement [10]. The power from the PV panel
determines the efficiency of the PV system. The position of the panel determines the
maximum extraction of power from the panel. MPPT algorithm includes software
coding at a reasonable price to supplement the energy throughout. The algorithm
ensures that the PV panel works at full efficiency with PV characteristics. P&O and
INC are the different MPPT techniques applied to extract maximum power. The
phase angle jump in the voltage can initiate transient current in the capacitors, trans-
formers and motors. It can also disturb the operation of converters and may lead
to glitch in the performance of thyristor-based loads. In order to protect sensitive
loads from grid voltage sags, custom power devices (such as, SVC, D-STAT COM,
DVR and UPQC) are being widely used. Among these devices, IDVR has most cost
effective and comprehensive solution. The above literature survey does not deal with
a comparison of with and without IDVR in renewable energy system for the power
quality enhancement. Integrating two renewable sources in a grid with compensating
the power quality was not done in the above papers. In this paper, the main objec-
tives are to improve the power quality, and fully utilize the wind and solar energy
was mainly considered. In Sect. 2, the conventional method of solar and wind energy
system connects to the grid were given. The proposed method of integrating the solar
and wind energy system is connected in a grid with compensation using IDVR were
described. The operation of IDVR, flow chart of IDVR control system was explained.
In Sect. 3, the simulation results of the proposed system with and without IDVR were
presented. In Sect. 4, the conclusion of this work.
Power Quality Enhancement Using Interline Dynamic Voltage … 509
Fig. 1 Block diagram of solar power generation system
2 Proposed Research Method
In this section, described the proposed method of power quality enhancement using
IDVR in renewable energy system.
2.1 Solar Energy Source
PV system is used to convert the incident sunlight on the PV panel into electrical
energy. Maximum PV output is extracted from the PV system automatically by using
the MPPT concept which varies the load conditions of PV array. The utilization factor
of the existing PV inverter is very less, as all PV arrays are provided with MPPT,
which is an inefficient method for tracking maximum power. In recent times, MPPT is
applied to the individual PV array since radiation is weak during night time. Figure 1
represents the block diagram of an existing solar power generation system. It consists
of a boost converter, DC energy storage device, injecting boost transformer and PWM
control strategy. Three phase AC voltage is obtained from DC voltage using VSI in
synchronization with the line voltage. Performance of 6.6 kV 3 phase 2 bus system
is analyzed using Sinusoidal Stepped Pulse Width Modulation (SSPWM).
2.2 Wind Energy Source
The wind energy conversion system (WECS) is used to convert mechanical energy
into electrical energy. Generally in WECS, two types of generators are used to
produce electricity. One of the generators is PMSG and another one is DFIG. Both the
generators offer an AC voltage at a frequency of 50 Hz. The converters are needed to
interface the power flow from turbine to grid. Figure 2 represents the block diagram
of an existing wind power generation system. It consists of a permanent magnet
synchronous generator (PMSG), boost converter, DC energy storage device, injecting
boost transformer and PWM control strategy. Three phase AC voltage is obtained
from DC voltage using VSI in synchronization with the line voltage. Performance
Fig. 2 Block diagram of wind power generation system
of 6.6 kV 3 phase 2 bus system is analyzed using Sinusoidal Stepped Pulse Width
Modulation (SSPWM).
2.3 Integrating of Solar and Wind Energy Using IDVR
Figure 3 depicts the block diagram representation of the proposed system. The inter-
connection of the utility grid with a renewable source of energy creates power quality
issues. IDVR system consists of two DVR connected back to back with a common
DC link. Two DVRs of IDVR are connected to two various feeders which are fed
from two grids with solar and wind sources. These two feeder voltage levels could
be equal or different.
Whenever the power quality issues occur in any one of the load, one of the DVR
mitigates the voltage quality, the other DVR in IDVR system is operated in power
flow control mode to restore the DC link capacitor. The poor voltage quality in the
power system occurs due to a fault, which depends on many factors like fault current,
different voltage level and transformer connecting arrangement. To overcome the
above factors, the IDVR system should be connected in two feeders and each feeder
is connected to two different grid sources. Figure 4 shows the flowchart of IDVR
Fig. 3 Block diagram representation of proposed system

Fig. 4 Flowchart of the

IDVR control system
control system. In this control strategy, first the amplitude of voltage sag is calculated
and compared with 2 − 2(ϕ). If the voltage sag amplitude is greater than this value,
then the shunt reactance are similar to the loads to decrease the load power factor.
Next, with respect to equivalent power factor which is seen by the source voltage, the
DVR voltages are resolved. This control system needs a quick and exact estimation
framework for figuring of magnitude and phase of relating waveform.
To investigate the system performance in voltage compensation, several simulations

have been done in the MATLAB/Simulink software on a single-phase IDVR.
Single line diagram of the proposed model with IDVR that is simulated is shown in
Fig. 5. Figure 5: Simulink diagram of two bus system is connected in solar and wind
energy source with IDVR. The output voltage of solar and wind generating stations
are depicted in Figs. 6 and 7, and their values are 2000 V and 4000 V, respectively.
Figure 8 shows the real and reactive power.
Figure 9 shows the voltage across the load in feeder 2 without compensation.
Figure 10 shows the injecting voltage from IDVR in feeder 2. Voltage across the
load in feeder 2 with compensation using IDVR is shown in Fig. 11. Frequency
Fig. 5 Simulink diagram of two bus system is connected in solar and wind energy source with
IDVR
Fig. 6 Output voltage of solar generating station
Fig. 7 Output voltage of wind generator

Fig. 8 Real and reactive power across the load in feeder 2. a Linear condition. b Nonlinear condition.
c Nonlinear with compensation using IDVR
Fig. 9 Voltage across the load in feeder 2 without compensation
Fig. 10 Injecting voltage from IDVR in feeder 2
Fig. 11 Voltage across the load in feeder 2 with compensation using IDVR
spectrum analysis with IDVR system is 4.85% and it is Fig. 12. Table 1 shows the
summary of real and reactive power loss with DVR and without DVR. Summary of
real and reactive power across the load is shown in Table 2, and their values with
nonlinear load using IDVR are 0.992 MW and 0.05043 MVAR.
Fig. 12 Frequency spectrum

analysis with IDVR system
Table 1 Summary of real and reactive power loss with DVR and without DVR
Feeder Real power loss Real power loss Reactive Reactive power loss
Without IDVR With IDVR power loss With IDVR
(MW) (MW) Without IDVR (MVAR)
(MVAR)
Feeder 1 0.0754 0.0521 0.0351 0.0278
(solar energy source)
Feeder 2 0.3671 0.2184 0.3210 0.2047
(Wind energy
source)
Total loss 0.4425 0.2705 0.3561 0.2325
% loss reduced using 35% 25%
IDVR
Table 2 Summary of real

Feeder 2 Real power Reactive power
and reactive power across the
(MW) (MVAR)
load
Normal condition 0.564 0.03753
Voltage sag without IDVR 0.394 0.01750
Voltage sag with IDVR 0.992 0.05043
4 Conclusion
IDVR is the advanced CPD device used for wind and solar energies utilization.
Simulation of two bus system using solar and wind energy with and without IDVR
is simulated using MATLAB/SIMULINK and analyzed in terms of real and reac-
tive power. The real power and reactive power of feeder 1 (solar energy source) is
improved with IDVR when compared without IDVR and its values are found to be
0.0521 MW and 0.0278 MVAR, respectively. Feeder 2 (wind energy source) has
improved real power and reactive power with IDVR and its values are 0.992 MW
and 0.05043, respectively. The total real power and reactive power losses are found
to be decreased by 17% and 12%, respectively, by employing IDVR.
References
1. Ghiasi M, Esmaeilnamazi S, Ghiasi R, Fathi M (2020) Role of renewable energy sources in

evaluating technical and economic efficiency of power quality. Technol Econom Smart Grids
Sustain Energy 5(1):1
2. Jain SK, Soni A (2020) Mitigation of power quality for wind energy using transmission
line based on D-STATCOM. In: Intelligent computing techniques for smart energy systems,
Springer, Singapore, pp 927–935
3. Ola SR, Saraswat A, Goyal SK, Sharma V, Khan B, Mahela OP, Alhelou HH, Siano P (2020)
Alienation coefficient and wigner distribution function based protection scheme for hybrid
power system network with renewable energy penetration. Energies 13(5):1–25
4. Khadem SK, Basu M, Conlon M (2010) Power quality in grid connected renewable energy
systems: role of custom power devices
5. Gowrishankar A, Ramasamy M (2020) Experimental validation of solar photovoltaic-based
unified power quality conditioner with modified power angle control scheme. J Test Eval 48(1)
6. Mallesham G, Kumar CS (2020) Power quality improvement of weak hybrid PEMFC and
SCIG grid using UPQC. In: Advances in decision sciences, image processing, security and
computer vision, Springer, Cham, pp 406–413
7. Sarker K, Chatterjee D, Goswami SK (2020) A modified PV-wind-PEMFCS-based hybrid
UPQC system with combined DVR/STATCOM operation by harmonic compensation. In: Int
J Model Simul pp 1–13
8. Suresh P, Baskaran B (2017) Power quality improvement of nine bus system during line
interruption using IVDFC. Int J Appl Eng Res 12(10):2527–2537
9. Suresh P, Baskaran B (2017) Voltage sag compensation in multiline distribution system using
closed loop controlled IDVR. Int J Appl Eng Res 12(8):1576–1583
10. Suresh MP, Ramya MG, Vijayalakshmi MK, Babu KS, Nithya MS (2018) Performance of
DVR and IDVR For voltage sag compensation in power distribution network. Int J Pure Appl
Math 119(16):4225–4239
Enhanced Hybrid Touch Screen Display
for Industrial Applications
A. Vignesh Babu, J. Ajay Daniel, V. N. Ganesh, S. Balaji, and G. Ramya
Abstract A Hybrid model with resistive touch and laser grid system is the solution
proposed to rectify the problems faced by the existing industrial touch screen displays.
Placing laser grid above the touch screen has improved the touch recognition and also
reduces the possibility of erroneous inputs. Lasers and phototransistors are placed
above the bezel of the display to form the laser grid. Laser grid forms a 2-dimensional
co-ordinate system for navigation of input location on the display. In order to avoid
unnecessary inputs caused by non-human entities like dust, flies, etc., inputs should
be taken by both laser grid and resistive touch panel simultaneously. Touch input
is taken into account only when the co-ordinates from both resistive touch screen
and laser grid are the same. If not, co-ordinates from laser grid alone are taken to
calibrate the resistive touch screen. If the calibration is not successful, then the touch
functionality is fully taken over by laser grid system.
Keywords Resistive touch screen · Laser grid system · Hybrid touch screen
1 Introduction
Nowadays, touch screen display plays a major role in every industry, and they are
widely implemented to enable features that allow complex touch inputs. Touch
Present Address:
A. Vignesh Babu (B) · J. Ajay Daniel · V. N. Ganesh · S. Balaji · G. Ramya
Ramapuram, India
J. Ajay Daniel
V. N. Ganesh
S. Balaji
G. Ramya
https://doi.org/10.1007/978-981-16-4943-1_48
518 A. Vignesh Babu et al.
display gives many features such as multi-touch operations, gesture control. Devices
with touch screen displays also have a longer lifetime than the older devices with
mechanical keys.
The equipment in an industry is controlled from small computers called Human
Machine Interfaces (HMI). These HMIs are equipped with resistive touch screens
which results in reduced clarity and they are also easily prone to scratches/cracks.
Since any malfunction of HMI directly affects the production, it is necessary to
provide a reliable solution to improve durability and precision of touch sensing in
HMI touch screens.
Finger touch can be recognized by various methods such as measurement of
pressure, variation in electrical characteristics and optical sensing [6]. Resistive touch
panel depends on both pressure and variation in electrical characteristics. Touch
panels which depend only on electrical characteristics are capacitive touch panels.
Optical touch panels include infrared touch screens which detects touch inputs with
photodetectors.
For industrial applications, resistive touch screens are mostly preferred over capac-
itive touch screens. The touch screen devices are placed near manufacturing lines
on-field where the operators are supposed to always wear protective gloves and capac-
itive touch screens won’t respond when used with a glove which makes resistive touch
screen the obvious choice in industrial applications. Since resistive touch screens are
used, features such as multi-touch and gesture controls cannot be implemented in
industrial displays.
Another such technology which can be implemented to improve precision and
durability is optical touch screen [1]. Optical touch screens also have the ability
to incorporate features such as multi-touch, gestures controls and sensitivity but
implementing optical touch screens alone will not be a proper solution as the optical
sensing could misidentify any dust particle or insects as a touch input.
The proposed solution is a hybrid that brings together the features of resistive and
optical touch screen technologies and also compliments each other to eradicate their
individual disadvantages.
2 Existing Technologies
Currently, most industries use resistive touch screens for almost all Human Machine
Interfaces. The few applications that use capacitive touch screens are located in places
like control rooms where there is no need for usage of gloves. Surface Acoustic touch
screen is another new technology which is sparsely implemented.
The major disadvantage of resistive touch screens is that multi-touch cannot be
implemented and it is easily affected by temperature changes. Since touch sensing
in resistive touch screen depends on resistance measurement, the display might take
erroneous inputs under non-ideal temperatures and such conditions require frequent
calibration of the touch screen. Implementation of resistive touch screen also reduces
the clarity of the display because of a thin resistive layer present on top of the display.
Enhanced Hybrid Touch Screen Display for Industrial Applications 519
It is more prone to scratches and cracks. If there are any cracks, it might stop working
or take improper inputs. Since these resistive touch screens are under repeated stress
because every touch is done with a small amount of pressure to make the two resistive
layers within it to meet resulting in a change in electrical characteristics. On the
other hand, capacitive touch screens are more sensitive than resistive touch screens,
it enables multi-touch capabilities which in turn provides multiple gesture controls.
Unlike resistive touch screen, capacitive touch screen does not reduce the clarity of
the display. It works even with small crack or scratches. Hardened glasses can be
placed above the capacitive touch layer in order to prevent cracks or scratches. These
displays cannot be used while wearing gloves. Since it does not need any pressure
for sensing, its lifetime is comparatively higher than the resistive touch screen.
The optical touch screens include infrared and laser-based touch screens. This
technology is not as reliable as the previous technologies as it can easily misjudge
any kind of dust particle/flying insect to be a touch input.
Surface Acoustical Wave is another similar technology which makes use of sound
wave reflectors to detect touch input. The major issue with this technology is its
vulnerability to external interferences and very low lifetime.
3 Proposed Solution
The proposed solution is a combination of both resistive touch screen and optical
touch screen technologies. The optical touch screen based on laser grid system has
the major advantage of being precise and resistant to damages done by repeated usage
but it can misread any kind of interruption between laser diodes and phototransistors
as an input which can lead to unintended operations. The resistive touch screens are
already widely used in industries but they lack features like multi-touch and gesture
control. These resistive touch screens are also vulnerable to damages like cracks due
to repeated stress on its surface. This technology also has the drawback of being
defective under non-ideal heat conditions. These drawbacks of optical and resistive
touch screens can be overcome by combining them both and making a hybrid system
with them. This solution is feasible as this can be implemented on existing HMIs
with minimal cost and labor. When a touch input is given, Both the laser grid system
and resistive touch screen system recognizes the input and the co-ordinates from
both these systems are taken in by microcontroller unit and it processes the data to
give out the corresponding output on the display (refer Fig. 1).
4 Simulation
The simulation for the proposed solution is done using proteus simulation software.
The model of laser grid is created using switches which represent the sensing points
on the touch screen (refer Fig. 2). Each time these switches are turned on, it simulates
Fig. 1 Block diagram of proposed hybrid solution
Fig. 2 Proteus simulation model
the process done when a touch input is given at that point on the touch screen. The
inputs from this laser grid model are taken to the Arduino mega microcontroller unit
for processing. The display is shown separately in the model but in reality, the laser
grid will be placed above the screen itself.
The simulations show that the touch response from the laser grid is properly
recognized by the microcontroller unit and being displayed on the display.
5 Construction
The simulation for the proposed solution is done using proteus simulation software.
The model of laser grid is created using switches which represent the sensing points
on the touch screen (refer Fig. 2). Each time these switches are turned on, it simulates
the process done when a touch input is given at that point on the touch screen. The
inputs from this laser grid model are taken to the Arduino mega microcontroller unit
for processing. The display is shown separately in the model but in reality, the laser
grid will be placed above the screen itself.
5.1 Laser Grid Formation
Optical touch system is created by forming a laser grid array using laser diodes
and phototransistors. Laser diodes are placed on one horizontal and one vertical side
serially and phototransistors are placed on opposite sides to the laser diodes. Figure 3
shows the arrangement of laser diodes and phototransistors.
These laser diodes and phototransistors are placed in a custom designed 3-d model
which houses these components and is capable of being placed above the resistive
touch screen.
Each intersection point represents laser light intersections. Touch is recognized
only on these intersections when a finger blocks the laser beam from hitting its corre-
sponding phototransistor. This prototype model has lower resolution because there
are only 11 phototransistors and 11 laser diodes which are arranged as 5 in horizontal
and 6 in vertical positions. To increase the resolution, number of laser diodes and
number of phototransistors should be increased which will result in increase in the
number of touch points.
This laser grid structure is placed above the resistive touch screen display and
they are both connected to an Arduino mega for processing the operations.
Fig. 3 Laser grid

arrangement
5.2 Resistive Touch Panel
The prototype is made with a 2.4-inch TFT LCD resistive touch display module
which is compatible with Arduino. It is a 4-wire resistive touch screen with a display
resolution of 240×320 pixels. A 4-wire resistive touch screen has two conductive
layers which are placed one on top of the other spaced in-between by spacer dots to
prevent sagging. The 4 wires are on the 4 edges of the screen and they are responsible
to detect the co-ordinates. Whenever a pressure (touch) is given on a point on the
screen, the top layer comes in contact with the bottom layer. To obtain the X co-
ordinate, a voltage is applied from left wire to right wire and the voltage drop at
the sensing point is captured. To obtain the Y co-ordinate, a voltage is applied from
top wire to bottom wire and the voltage drop at the sensing point is captured. These
captured voltages drop values are processed by the processing unit to identify the
point of touch on the screen.
5.3 Design Considerations
Laser grid is placed slightly above the display (around 5 mm above the surface) to
avoid any foreign particles such as dust or insects to disrupt the laser beam. Figure 4
shows the arrangement. This positioning is extremely necessary for displays which
are placed horizontally, due to gravity some foreign particles might fall on the display
which could interrupt the laser grid. Such foreign particles would normally lie well
below the laser beams or they might not have enough pressure to be recognized by
resistive touch screen, and since the hybrid system needs both inputs, there erroneous
inputs can be prevented. 5 mm is the threshold height of laser beam above screen
surface that has been implemented in the prototype. This height is more than enough
to overcome the issues which will be faced by these foreign particles.
Fig. 4 Design consideration (Height)

5.4 Processing Unit
A program is written on Arduino to integrate both the resistive and laser grid system to
get input co-ordinates. Additionally, multi-touch feature is also coded. The Arduino
mega is used as the processing unit. The inputs from both the systems are taken
and compared by the Arduino. When both the systems show same co-ordinates, the
Arduino does a corresponding processing to perform intended operation. When the
co-ordinates from the two systems are not same, then the Arduino tries to calibrate
the resistive screen with the laser grid inputs as reference. If the resistive screen is not
calibrated properly, then the input sensing is totally taken over by laser grid alone.
5.5 CAD Model
A 3-D printing model had to be created for housing the laser diodes, phototransistors
and resistive touch screen together. The 3-D model was designed using Autodesk 360
Fusion (as shown in Fig. 5). The number of lasers and phototransistors used in the
Fig. 5 CAD model for prototype housing

Fig. 6 Main screen, menu screen, lights and motors control page
prototype are few due to size and budget constraints but with custom manufactured
components, this solution could be easily implemented with lesser size and also at a
reduced cost (Fig. 5).
6 Working
The flow chart in Fig. 7 shows the actual algorithmic workflow when a touch is
sensed in this system. The system is actively waiting to sense any kind of touch
input. When it identifies any input that blocks laser beam and also provides pressure
on resistive touch screen, it is validated as a proper touch. The co-ordinates are
obtained from both resistive panel and laser grid and compared to check if they
are equal. The corresponding operation is carried on when both the co-ordinates are
equal. If they are not equal, the system tries to calibrate the resistive touch screen with
laser co-ordinates as reference. The values are compared again to check whether the
calibration removed the error. If they are still not same, the calibration and checking
process happens again for a maximum of three trials. After three trials, the resistive
panel is totally removed off the loop and only the co-ordinates of laser grid are
considered for operation.
6.1 Operations with Proposed Model
For demonstration purpose, the prototype is built to operate lights and motors. The
prototype is supposed to be an HMI in an industry which is connected to 4 loads
which are two lights and two motors. The demonstration is to showcase the multi-
touch feature of this system. Figure 6 shows the main screen and the menu screen of
the prototype. The light control screen is shown in Fig. 6. The light control screen
helps to control lighting loads. The motor control screen has motor switching buttons
(refer Fig. 6). This screen allows the user to simultaneously operate both the motors.
Fig. 7 Flow chart
The proposed model also has the ability to self-calibrate its resistive touch screen
when it gets erroneous due to temperature changes. The working model that was
created proved to overcome the challenges faced by the older isolated technologies.
7 Result
The simulation and prototype were developed and tested. The working prototype
has been satisfactory in operation. Multi-touch functionality was tested by taking a
case where 4 loads are operated simultaneously by turning on/off on the display by
multi-touch. To perform multi-touch 4 loads such as 2 DC motors and 2 LEDs are
connected to the Microcontroller, in which two motors can be turned on or off by
simultaneously touching on/off button for the corresponding motor in the display.
Similarly, 2 LEDs can be controlled simultaneously by providing 2 touch points.
The hybrid solution is working as intended and also overcomes the drawbacks of
previous technologies.
References
1. Liang J, Lin K, Wang W (2018) Turn any display into a touch screen using infrared optical
technique. vol 6, pp 13033–13040
Reliability Study on the Distribution
System Integrated with Wind Generator
S. B. Aruna and D. Suchitra
Abstract Electricity should be provided to the customers with good quality and
continuity of power supply. To predict the future of distribution system performance,
reliability assessment is one of the major factors that have to be considered. Reliability
assessment helps in designing and planning of distribution system and also helps in
analyzing the severity of the system failures. The system performance indices for
reliability help in analyzing the severity of the system failures and thereby help in
predicting the reliability of the system in future. In this paper, the system performance
analysis has been simulated in ETAP on the Bus 2 of Roybillinton Test system (RBTS)
for Feeder 1, 3 and 4. The system performance indices such as SAIFI, SAIDI, CAIFI,
ASAI, ASUI, AENS and EENS under different scenarios of operations with and
without distributed generator have also been analyzed.
Keywords Reliability Assessment · Distribution generation · Load oriented

indices · RBTS
1 Introduction
Power systems are subjected to random faults due to component failures, transmission
faults, climatic conditions, etc. Regarding the modeling and operation of the power
network, maintaining a reliable power supply becomes a crucial factor that needs
to be addressed first for the better operation of the system. The concept of power
system reliableness is extraordinarily broad and covers all the aspects associated
with the flexibility of the power system to produce adequate supply of electricity
thereby satisfying the user’s necessities. Electric power utilities also face increasing
uncertainty due to the political, economic, societal and environmental constraints,
as they have to effectively operate the existing system and also should plan for the
S. B. Aruna (B) · D. Suchitra

D. Suchitra
https://doi.org/10.1007/978-981-16-4943-1_49
528 S. B. Aruna and D. Suchitra
enhancement to manage the ever increasing demands. All these conditions have
created an environment for the utility to plan for the new facilities, optimal use of the
existing system configuration for the improvement in system reliability and reduced
operation costs [1]. The Distributed Generators (DG) are connected near to the load
centers in the distribution network for obtaining a better reliability of the system.
By connecting DG at the appropriate locations in the distribution system, the power
demand of the system can be satisfied which in turn improves the reliability. When
properly designed, a regional power grid that combines both large central plants
and distributed generator can provide higher capacity factor on all the assets and
reliability enhancement.
Many research has been carried out until now, considering the various aspects
of reliability with respect to power system. More number of techniques for relia-
bility evaluation of distribution system and significance of reliability indices were
summarized in [2]. The impact of distributed generators with respect to the reli-
ability was described and analyzed with various ratings at different locations on
distribution system [3]. The significance of integration of distribution generation
with the various reliability indices was analyzed, with the different case studies [4].
Reliability analysis of distribution system under the impact of different operating
configurations were presented in [5]. Reliability Index for the distribution system
with distribution generation and its effect on the power losses were explained in [6].
Reliability assessment of the medium voltage network with respect to automation,
aging and load growth were presented in [7]. Reliability assessment of the overall
power system with various hierarchical levels and analysis of reliability indices with
actual customer levels of service was explained in [8]. Hence, this work analyzes the
impact of DG on the reliability, based on the various indices, by varying the locations
of the DG. Also, it brings out the best location of the DG for the considered feeder.
2 Reliability Indices of Distribution System
Providing an uninterrupted and quality power supply to its consumers becomes the
main objective of the distribution system. Normally distribution networks are radially
configured, so failure of single section may lead to whole supply interruption. Hence,
reliability of the system has a major impact on such system, and therefore, system
performance has to be completely analyzed. Failure rate, repair time and annual
unavailability of the system are the primary reliability indices used to envisage the
reliability of a distribution system. The overall System performance indices such as
SAIDI, SAIFI, CAIDI, ASAI, ASUI, EENS and AENS were also evaluated to assess
the operating conditions of the distribution systems. The system performance indices
of the utility with respect to distribution level are as follows:
System Average Interruption Frequency Index (SAIFI) [int/yr]

Total Number of customer interruptions λi N i
= = (1)
Total number of customers served Ni
Reliability Study on the Distribution System Integrated with Wind … 529
where N i represents the number of customers at load point ‘i’, λi represents failure
rate.
System Average Interruption Duration Index (SAIDI) [hr/yr]

Total Number of customer interruption duration Ui N i
= = (2)
Total number of customers served Ni
Customer Average Interruption Duration Index (CAIDI)[hr/int]

Total Number of customer interruption Durations Ui N i
= = (3)
Total number of customers interruptions λi N i
Average Service Availability Index (ASAI)

Customer hours available in service 8760N i − Ui N i
= = (4)
Customer hours demanded 8760N i
where U i represents annual unavailability.

Average System Unavailability Index (ASUI)

Customer hours not available in service Ui N i
= = = 1 − ASAI (5)
Customer hours demanded 8760N i

n
Expected Energy Not Supplied (EENS) [hr/yr] = E i (6)where E i represents
i=1
the average of interruption energy per load point.

n
Ei
Average Energy Not Supplied (AENS) [kWh/yr] = i=1
Ni
(7)
3 Test System Description
Roy Billinton Test System (RBTS) is a benchmark system, which is considered for
the reliability study to estimate the overall reliability indices. RBTS was developed
by university of Saskatewan, Canada for carrying research activities in 1996. RBTS
is a system with five load buses and eleven-generator bus. The installed capacity is
240 MW and peak load of the system is 185 MW. This system has five voltage levels
230 kV, 138 kV, 33 kV, 11 kV, 415 V. Figure 1 depicts the RBTS Bus-2 system that
is considered for the reliability analysis in this work [9, 10].
This system has 22 load points with four feeders at a working voltage of 11 kV.
A Breaker is positioned at the source end of the 11 kV main feeder. It is calculable
that the operation of the breaker is 100% and the breaker is opened if any fault
happens within the system. Also, it separates the faulty load area in order that the
service is sustained for the remaining customers with the healthy load points. A
Fig. 1 Roybillinton bus-2 test system
radial distribution feeder along with main and lateral feeder sections are considered.
Load points are connected to the main feeder through lateral distributors. Fuses are
used in distributors to eliminate failures on the lateral section from the main feeder.
Disconnects or isolators are used in main feeder section to detach faulted sections.
Furthermore, an alternate power supply is provided to the healthy load points with the
aim continuing the service throughout the failure. Table 1 is listed with the input data
of main feeder and lateral distributors. The reliability parameters for the different
components connected to the system is given in Table 2. The input data related to
peak load with respect to load point connected is specified in Table 3.
Table 1 Main feeder and

Length in km Load points in feeders
lateral distributor load point
and distance data 0.60 4, 6, 9, 14, 15, 18, 24, 29, 31, 32
0.75 1, 2, 3, 5, 7, 10, 12, 13, 20, 25, 27, 30, 35
0.80 8, 11, 16, 17, 19.21, 22, 23, 26, 28, 33, 34, 36
Table 2 Reliability
Components Failure rate Repair rate Switching time
parameters of components
connected (f/hrs) (hrs) (hrs)
connected to the system
Transformers
33 kV/11 kV 0.015 15 1
11 kV/0.415 V 0.015 10 1
Breakers
33 0.002 4 1
11 0.006 4 1
Bus bars
33 0.001 2 1
11 0.001 2 1
Feeder
11 0.065 5 1
Table 3 Peak load data of load points

Load points connected to Type of customer connected Load point customers Peak load
feeders (MW)
1, 2, 3, 10, 11 Residential 210 0.8668
12, 17, 18, 19 Residential 200 0.4500
4, 5, 13, 14, 20, 21 Government and industrial 1 0.566
8 Industrial 1 1
9 Industrial 1 1.15
6, 7, 15, 16, 22 Commercial 10 0.454
4 Reliability Evaluation of RBTS Bus-2 System Integrated

with Distributed Generation
A wind turbine generator is taken as DG source and integrated into feeders 1, 3 and
4 of RBTS Bus-2. Feeder-2 consists of only two load points so integrating wind
turbine will not have more impact on the system performance so remaining feeders
are taken for reliability analysis. A Wind turbine with a rating of 1 MW has been
integrated for feeder 1 of RBTS Bus 2. The failure rate and average repair time of
wind turbine are 0.25 f/yr and 20 h respectively. The wind turbine integrated with
feeder is considered to operate with the average wind velocity Vavg = 4.05 m/sec,
rated velocity Vr = 10 m/sec, cut in velocity V ci = 2.0 m/sec, cut off velocity Vco
= 22 m/sec. The Test system RBTS Bus 2 Feeders 1, 3 and 4 have been simulated in
ETAP (Electrical Transients Analysis Program) environment [11, 12]. The reliability
indices were calculated at various distances from supply terminal of the Feeders. The
wind turbine generator has been placed at four distant locations and indices were
calculated for Feeders -1, 3 and 4 (Fig. 2).
Fig. 2 RBTS Bus-2 system

feeder-1 connected with DG
Case study-1:
Feeder-1 of RBTS consists of seven load points, i.e., LP-1 to LP-7 with different
categories of customers. The reliability indices were calculated for the following
scenarios by placing wind turbine generator at various distant points from supply
terminal of feeder-1.
Scenario -1: RBTS Bus -2 Feeder 1 without Distributed Generator (DG).
Scenario -2: RBTS Bus -2 Feeder 1 with DG at 1.35 km away from supply point.
Case study-2:
Feeder-3 of RBTS consists of six load points, i.e., LP-10 to LP-15 with different
terminal of feeder- 3 (Fig. 3).
Scenario -1: RBTS Bus -2 Feeder 3 without Distributed Generator ( DG).
Scenario - 4: RBTS Bus -2 Feeder 3 with DG at 2.9 km away from supply point.
Case Study-3:
Feeder-4 of RBTS consists of seven load points, i.e., LP-16 to LP-22 with different
Fig. 3 RBTS Bus -2 system

feeder -3 connected with DG
Fig. 4 RBTS Bus -2 system

Feeder -4 connected with DG
terminal of feeder- 4 (Fig. 4).
Scenario -1: RBTS Bus -2 Feeder 4 without Distributed Generator (DG).
By using ETAP RBTS bus-2 feeder 1, 3 and 4 has been simulated and overall
indices are listed in the following Table 4, Table 5 and Table 6.
Table 4 Reliability indices of bus-2, feeder-1 connected with DG at various distant points
Reliability indices of feeder-1 Scenario 1 Scenario Scenario 3 Scenario 4 Scenario 5
2
SAIDI (hr/customer.yr) 6.5794 6.8617 5.6899 5.1305 5.0982
SAIFI (f/customer.yr) 1.2429 0.6538 0.5839 0.5549 0.5527
CAIDI (hr/customer.yr) 6.579 6.862 5.690 5.130 5.098
ASAI(pu) 0.9992 0.9992 0.9994 0.9994 0.9994
ASUI(pu) 0.00075 0.00078 0.00065 0.00059 0.00058
AENS (MW hr/customer.yr) 0.0394 0.0395 0.0364 0.0336 0.0317
EENS (MW hr/yr) 75.160 75.431 69.433 64.122 60.400
Table 5 Reliability indices of bus-2, feeder-3 connected with DG at various distant points
Reliability Indices of Feeder- 3 Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario
5
ASAI(pu) 0.9992 0.9992 0.9993 0.9994 0.9994
ASUI(pu) 0.00075 0.00078 0.00070 0.9994 0.00058
AENS(MW hr/customer.yr) 0.0394 0.0387 0.0369 0.0341 0.0317
EENS (MW hr/yr) 75.226 73.844 70.322 65.082 60.544
Table 6 Reliability indices of bus-2, feeder-4 connected with DG at various at various distant
points
Reliability indices of feeder-4 Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5
ASAI(pu) 0.9992 0.9992 0.9993 0.9994 0.9994
ASUI(pu) 0.00075 0.00077 0.00070 0.00058 0.00058
AENS (MW hr/customer.yr) 0.0394 0.0393 0.0366 0.0342 0.0327
EENS(MW hr/yr) 75.226 75.029 69.915 65.301 62.451
5 Discussions
It is analyzed from Fig. 5 that, the interruption duration is more in the base case,
i.e., without DG in three feeders. If any fault occurs in the feeder end or in lateral
section, breaker disconnection will result in unsupplied energy for the customers in
all the load points. Integration of 1 MW wind turbine generator at various distant
points reduces the interruption duration especially at the ends of feeder.
It is observed from Fig. 6 that the failure rate of three feeders is high for the base
case. Connecting DG of 1 MW at various distant points has reduced the failure rate
considerably, when placed at 3.65 km away from supply point for feeder-1, 3.5 km
away from the supply point for feeder-3 and 3.65 km away from supply point for
feeder-4 hence increased the continuity of power supply to the feeder.
From Fig. 7, it is explored that, the Expected energy not supplied for the three
feeders has reduced after the placement of DG of 1Mw and helps in providing the
quality power to the consumers.
Fig. 5 SAIDI for RBTS bus-2- feeders-1, 3 and 4

Fig. 6 SAIFI for RBTS bus-2 feeders-1, 3 and 4
Fig. 7 Expected energy not supplied for RBTS bus-2 feeders-1, 3 and 4
6 Conclusion
The significances of reliability indices of Roybillinton Test system bus-2 feeders-1,

3 and 4 with the inclusion of the distributor generator at various distant points are
analyzed. Five different scenarios are considered for the three feeders of bus-2 and
the system performance indices are obtained by using ETAP simulation Tool. Asso-
ciating the results of all the five scenarios of three feeders shows that the reliability
indices SAIDI, SAIFI and EENS have been improved for all the load points due
to the presence of DG. Injecting the Wind turbine generator as distribution gener-
ation source into feeder configuration has provided positive impact on distribution
system reliability. Reliability analysis of distribution system provides an outline for
the performance assessment of distribution system and supports in progress of future

load operations.
References
1. Chandhra Sekhar P, Deshpande RA, Sankar V (2016) Evaluation and improvement of reliability
indices of electrical power distribution system. 2016 Natl Power Syst Conf NPSC, no 4. Nov
2017
2. Divakar V, Raviprakash MS, Keshavan BK (2016) A survey on methods of evaluation of
reliability of distribution systems with distributed generation. Int J Eng Res V5(08):220–226
3. Sailaja CVSS, Prasad PVN (2016) Reliability and cost benefit analysis of DG integrated
distribution system. Bull Electr Eng Informatics 5(4)
4. Subramanya Sarma S, Madhusudhan V, Ganesh V (2018) Reliability constrained planning of
distribution system under high penetration of stochastic DG units. Int J Eng Sci Res Technol
7(3):538–548
5. Kirubarani K, Peer Fathima A (2019) Distribution system reliability assessment for improved
feeder configuration. Int J Eng Adv Technol 8(6):4416–4421
6. Al-Maghalseh MMA (2018) Evaluating the reliability worth indices of electrical medium
voltage network: case study. Procedia Comput Sci 130:744–752
7. Chiradeja P, Pothisarn C, Ngaopitakkul A (2020) The study of reliability indices, power losses,
and economic effects when distributed generators (DG) connected to a distribution system. Int
J Smart Grid Clean Energy 9(2):338–345
8. Sonwane PM, Kushare BE (2015) Distribution system reliability: an overview. 5(3):148–153
9. Billinton R (1996) A test system for teaching overall power system reliability assessment. IEEE
transactions on power systems 11(4):1670–1676
10. Shiwen Y, Hui H, Chengzhi W (2017) Review of risk assessment of power system. Sci Direct
Proceedia Comput Sci WTISG 2017 1200–1205
11. Ahmad S et al (2017) Reliability analysis of distribution system using ETAP. Int J Computer
Sci Inf Secure 15(3):197–201
12. Ahmad S, Sardar S, Noor B, Ul A (2016) Analyzing distributed generation impact on the
reliability of electric distribution network. Int J Adv Comput Sci Appl 7(10):217–221
Retrofitting of Internal Combustion
Engine Vehicles with DC Motor
Femi Robert, Muskan Puri, Ashay Kumar Thakur,

and Gajendran Marimuthu
Abstract This article gives a detailed report on the retrofitting of an IC engine

vehicle with a specially designed DC motor. The emission losses of an old ICE
vehicle increase and saturates to its highest value at around this time. The load and
rpm analysis of ICE shows that the available usable torque to produce the desired
RPM is in a very small range and thus replacing such engines with electric engines
which have high torque to speed ratio will increase the performance of the vehicle
greatly. The DC motor used in this application has been specially designed for the
vehicle. The Total Tractive Effort (TTE) is calculated considering the curb weight of
the chassis and then eliminating the components which are specific to the ICE such
as engine and fuel tank. Other factors such as the static and dynamic friction, rolling
resistances and other constants have been considered taking into account the road
safety, road transportation and NHAI standards. The retrofitted vehicle has a curb
weight less than that of the ICE counterpart. The high torque DC motor gives better
acceleration and the high RPM provides better maximum speeds. The main goal is
to have a used ICE vehicle and to completely renew it with a DC motor. This would
eliminate the use of non-renewable fuels while also extending the life of the vehicle.
Keywords Electric Vehicle · ICE vehicles · DC motor · Sensors
F. Robert (B) · M. Puri · A. K. Thakur

Department of EEE, SRM Institute of Science and Technology, Kattankullathur 603 203, India
M. Puri
A. K. Thakur
G. Marimuthu
Department of EIE, RMK Engineering College, Kavaraipettai 601 206, India
https://doi.org/10.1007/978-981-16-4943-1_50
538 F. Robert et al.
1 Introduction
Electrical vehicles are green traction drives that are currently used to replace the
conventional type of vehicle such as internal combustion engine [1]. Electrical vehi-
cles are very easy to operate, quite energy efficient while being very effective and
better than LPG engines on all grounds such as cost, impact on environment. The
development of electrical vehicles is increasing very rapidly, due to the environ-
mental issues and fossil fuel shortages [2]. Electrical vehicles consist of a fuel cell
electric vehicle (FCEV) and battery electric vehicle (BEV). EVs basically are gener-
ally considered to be consisting of different sub-systems. Each of the subsystems
needs to work in tandem with each other to make an EV work [3]. There are many
technologies used to develop coordination in between the systems. The heart of the
EV is a motor. Normally for such applications, a DC series motor drive with armature
control is used, which provides a very good high starting torque and a good speed
regulation. The maximum speed of EV is equivalent to the speed of the motor [4]. A
major sub-system for the EV is the drive circuit of the motor and its necessary sensor
circuitry. Sensors are used for continuously reading the output of the battery in terms
of the voltage and current delivered to the motor drive circuit. It also continuously
monitors the temperature of the battery [5]. In this paper both the subsystems are
used to make a setup for the EV. All the sensors used in the system are custom made.
2 Sensor Simulation Profile
The simulation is done on the MATLAB/Simulink and Proteus 8 Professional with

some pre-defined and measured initial conditions.
2.1 Voltage Senor
It continuously senses the output voltage of the battery [6] and displays the output
in digital form by the voltage division rule using the microcontroller. Equation for
calculating the Voltage in the circuitry is shown below in Fig. 1
V = 24.0 − (0.4x) (1)
x = range of the voltage increase to the Arduino.

Retrofitting of Internal Combustion Engine Vehicles with DC Motor 539
Fig. 1 Voltage Senor
2.2 Position Sensor (Hall Effect)
This sensor is used to control the speed of the DC motor by determining the position
of the rotor magnet [7]. There are three sensors which are used for analyzing the
rotor position as shown in the given Fig. 2. Each senor will give output in volt as the
magnet passes it. By using the comparator and voltage division rule having value
of two resistors (R1 = 1000 and R2 = 250 ), the sensor is then interfaced with
the microcontroller to get the readable output. The output sent by the sensor can be
varied by the PWM signal for varying the speed of the motor.
Fig. 2 Position sensor

Fig. 3 Current sensor
2.3 Current Sensor Simulations
It is used to continuously monitor the current drawn from the battery to the motor
[8]. The output of the sensor is in volts, that is derived from the voltage division law,
the relation between the voltage and current is of difference 0.15x. The maximum
current drawn from the battery is 15.4 A as shown in Fig. 3 and with that reference
the input voltage to Arduino is 4.34 V as shown in Fig. 4.
Equation for calculation of the current in the circuitry is shown below
I = 13.4 − (0.2x) (2)
x = range of the voltage increase to the Arduino.
2.4 Temperature Sensor
It continuously senses the temperature of the battery at the time of charging as well
as at discharging. The range of the temperature measurement is between −40 C and
125 C. The output of the senor is 10 V thus it is stepped down by voltage divider
rule for feeding the output to the microcontroller [9] which as a input limit of 5 V
as shown in Fig. 5. By interfacing it with the microcontroller, the digital values are
obtained as output.
Fig. 4 Current sensor interface with Arduino
Fig. 5 Temperature sensor circuitry
3 Drive Circuit Simulation Profile
Now for the retrofitting of the concerned electrical vehicle, the concerned article
covers two kinds of driver circuits for PWM generation and speed cum direction
control of the concerned DC motor for the speed control. The first circuit depicts
an analog-based circuit for speed control while the second circuit depicts a micro-
processor-based method for the overall speed control cum direction control.
3.1 Analog Circuitry
As shown in Fig. 6, it is a basic analog-based controller that controls the speed of

the motor with a fixed analog input. The output is controlled with the help of a 500
potentiometer. The potentiometer varies the motor armature current of the motor [10]
and accordingly the RPM of the motor. For ensuring constant speed of the motor, a
throttle switch is provided to ensure the supply of rated current to the motor whenever
required [11]. In the above circuit, a 24 V DC input is connected to drive the control
circuit.
A Schottky diode is connected anti-parallel to the motor to protect the circuitry
against reversal of high magnitude currents. Since the DC motor [12] has an inductive
load, when driver desires to switch off the motor and reduces the voltage across the
motor terminal to zero, the residual current in the motor winding will generate a high
back emf if there is no discharge of current. So, the Schottky diode acts as a short
circuit for discharge and has a low power drop. The battery supplies 24 V but the
main op-amp of the circuit requires a maximum of 12 V across its terminal.
This is achieved by the use of LM7812 voltage regulator. It bucks the 24 V to a 12 V
signal and contains a capacitor in parallel in order to filter out the ac components. This
signal is then passed to the comparator IC which helps in triggering the transistor.
When the comparator output is high, the gate of the transistor is turned ON and vice
versa when the comparator output is LOW. The output of the comparator changes
each time the pulse changes its polarity. The output of the transistors serves as the gate
Fig. 6 Analog motor drive circuit

Fig. 7 Microcontroller-based drive circuit
pulses for the two MOSFETs (Metal Oxide Semiconductor Field Effect Transistor)
which generates an electric field when voltage is applied at the gate) [13]. This
helps to control the current flow through the channel between drain and source. The
potentiometer mentioned earlier is used at the gate of the MOSFET to change the
duty cycle of switching of the MOSFET [14]. This helps in controlling the output
current of the MOSFET which further controls the current through the motor and
thus ensuring a uniform control of speed as far as the motor is concerned [15].
3.2 Microcontroller-Based Circuitry
Figure 7 shown above shows the circuit design and simulation of another alternative
motor control circuit which is purely microcontroller-based as shown in the simula-
tion diagram. In this diagram, an external potentiometer connected to a 5 V supply is
read by the microcontroller which then maps into the corresponding gate pulse with
the corresponding duty cycle [16]. A single MOSFET driver IR2104 is used to run
at the same time two high power MOSFETs thereby controlling the speed of the DC
Motor on a voltage basis and ensuring uniform open loop control of the DC Motor.
The circuit is enabled with the help of decoupling capacitors and gate resistors and
input filter circuit. The concerned microcontroller used for this operation is the
Arduino Uno or the AtMega 328P IC [17]. It reads the analog input of the driver
from the variable rheostat connected to one of its analog pins. The IC accordingly
generates the PWM Signal from one of its PWM pins and feeds it to the high power
IR2104 MOSFET Driver [18]. This driver with the help of an external 12 V Vcc
generates a reference signal higher than the 5 V PWM output of the AtMega 328P
IC for proper conducting and switching of the two concerned high power MOSFETs
IRFP120N. Capacitors of various values are connected in parallel to filter out the
unnecessary AC components. In this method of speed control, contrary to the previ-
ously mentioned analog method, the speed of the motor is varied by varying the
Fig. 8 Open state
net supply voltage across the motor input terminals. Thus the microcontroller-based
method of controlling speed is based on voltage control method of speed control
as against the current control-based method of speed control of the aforementioned
analog controller in Fig. 6.
4 Kill Switch Mechanism
The kill switch mechanism [19] is used to ensure the over-current protection of the
motor. In case there is any anomaly in the operation of the motor, the kill switch is
switched on which in turn shorts the relay with its 12 V actuation voltage thereby
cutting off the motor with its supply voltage. Vice versa, when normalcy is restored,
the kill switch is turned off to restore normal running of the motor [20] as shown in
Figs. 8 and 9.
5 Results and Observations
The results and the corresponding variations in the speed of the motor with respect to
various parameters like the net armature current and the rheostat resistance connected
in series with the gates of the MOSFETs are graphically represented. Figure 10 shows
a variation of speed vs the net armature current of the motor, we observe different
stages of variation. While initially, the speed rises steeply with increase in the current,
after a certain point, we observe that beyond a certain point of saturation the rise in
speed with current becomes less steep and more or less uniform giving the resultant
curve the shape of an Ogive curve. The second curve which is shown in Fig. 11
Fig. 9 Close state
Fig. 10 Speed versus

current for the DC motor
shows that the rheostat or the potentiometer is directly responsible for controlling or
limiting the speed of the motor. The curve shown in Fig. 11 is that of a uniform curve
showing decrease in speed with increase in value of net resistance on the rheostat
which is varied by varying the knob on the potentiometer.
Although the rheostat offers a maximum resistance of 500 ohms, we can observe
that the concerned motor’s rpm comes to a standstill zero at a value of around
300 ohms from the graph shown in Fig. 11.
Figure 12 shows a screenshot of the drive circuit when the rheostat resistance in
series with the gate is minimum, and thus the current flowing through the motor is
maximum as is donated by the denoted ammeter reading, and thus the concerned
rpm of the motor is also maximum. Figure 13 shows a screenshot of the drive circuit
when the rheostat resistance in series with the gate is maximum, and thus the current
Fig. 11 Speed versus

rheostat resistance for the
DC motor
Fig. 12 Maximum current to the motor
flowing through the motor is minimum and thus denotes the standstill or rest position
of the concerned DC motor. It is to be noted that all the concerned simulations in the
article have been performed using the software Proteus 8 Professional.
Fig. 13 Minimum current to the motor
6 Conclusion
This article gives a detailed view on the retrofitting of ICE-based vehicles with corre-
sponding electrical vehicle designs which helps in providing effective alternatives to
the generic fossil fuel run vehicular transmission systems sometimes using both fossil
fuels and natural gas. By using the proposed simulation profile which implements
the use of real-time and active simulations of various vehicle running and sensing
operations, the overall energy and power efficiency for the accustomed power elec-
trical vehicular transmission system can be calculated and further implementations
can be executed accordingly. Along with the promised performance efficiency, which
is more or less equivalent to its corresponding ICE-based vehicle design, the motor
transmission is comparatively way less polluting and offers less losses due to noise
and other external physical disturbances. However the shortcoming of this article,
it provides detailed calculations and simulations for very low HP vehicles, and the
net electrical output can be compared to that of those clean energy tow-tow vans.
This design can be further used as reference for high energy electrical output vehicle
designs like ICE-based sedans offering a HP rating of as high as 75 HP. The fossil
fuels levels are declining rapidly, and with this, the need for alternative energy sources
has risen. Over one-third of fossil fuels are used to power the transportation industry
and thus it is important to limit the use of these fuels in vehicles. Electric Vehicles
have begun to capture the market slowly but the industry is still in its budding stages
and thus catering to the complete needs of the people is far from a reality. Even if
EVs take over the industry completely then we would face the issue of excess road
traffic with the already existing ICE vehicles and the new EV. Thus, eliminating ICE
vehicles is just as important as introducing EVs.
References
1. Abas N, Kalair A, Khan N (2015) Review of fossil fuels and future energy technologies. Futures
69:31–49
2. Hasnira H, Didi I (2014) Energy estimation on BLDC electric vehicle using MATLAB. IPTEK
J Proc Series 1:485–489
3. Sundriyal P, Shantanu B (2019) Sensors for automotive and aerospace applications. Springer,
Singapore pp 1–6
4. Bimal K, Robert L (1978) A DC motor control system for electric vehicle drive. IEEE Trans
Ind Appl IA-14(6):56572
5. Rowlette MR (1994) Temperature sensor. U. S. Patent
6. Jose S (2001) Rechargeable battery pack and the battery pack charger with safety mechanism.
United States Patent. 1(12)
7. Murray A (1998) Position and speed sensors. United States Patent.19
8. Patel A, Ferdowsi M (2009) Current sensing for automotive electronics—a survey. IEEE Trans
Veh Technol 58:4108–4119
9. Carkhuff BG (2019) Battery internal temperature sensing battery management system.
US10461374
10. White KE (1980) Control system and method for operating a DC motor. US4240015
11. Morisset MR (2019) Throttle control system and method. US10479200
12. Thattil A, Vachhani S, Raval D, Patel P, Sharma P (2008) Comparative study of using different
electric motor for EV. Int Res J Eng Technol 9001:4601–4604
13. Zhou Q, Gao F, Jiang T (2015) A gate driver of SiC MOSFET with passive triggered auxiliary
transistor in a phase-leg configuration. IEEE Energy Convers Congr Exposition 7023–7030
14. Smith TA, Dimitrijev S, Barry Harrison H (2000) Controlling a DC-DC converter by using the
power MOSFET as a voltage controlled resistor. IEEE Trans Circ Syst I: Fundam Theory Appl
47:357–362
15. Silvestre J (2008) Half-bridge bidirectional DC-DC converter for small electric vehicle. In:
SPEEDAM 2008—international symposium on power electronics, electrical drives, automation
and motion, pp 884–888
16. Ali YSE, Noor SBM, Bashi SM, Hassan MK (2003) Microcontroller performance for DC motor
speed control system. In: Proceedings national power engineering conference, pp 104–109
17. Gavran M, Fruk M, Vujisi G (2017) PI controller for DC motor speed realized with Arduino
and Simulink, pp 1557–1561
18. Shetti PR, Mangave AG (2014) DC motor speed control with feedback monitor based on C#
application. Int J Res Eng Technol 03:398–401
19. Pedersen BB, Reese DA, Joyce J (2013) Method and apparatus for securing a programmable
device using a kill switch. U.S. Patent 8,461,863
20. Gadade JB, Bharane GR, Rupanwar TP, Kale RU (2018) DC motor protection, Control and
Monitoring. pp 3458–3460
Analysis of Cascaded Multilevel Inverters
for Smart Grid Connected PV Systems
R. L. Josephine, M. Yoogesh Kumar, and S. Harishankar
Abstract With the increasing need for renewable energy systems to reduce the
dependence on conventional resources and fossil fuels and to counter the impact of
their usage, more efficient and robust power generation, management and storage
systems are required. To best augment the capabilities of such systems, it is imper-
ative that a smart grid-based control strategy is adopted. This will improve the effi-
ciency of such a system and will lead to better integration of the system with the
grid. With increased penetration of such smart grid connected systems, the need for
securing them from cyberattacks, having the potential to cripple strategic assets of
nations, becomes quite necessary. In this paper, a Cascaded H-bridge-based Multi-
level Inverter topology is proposed, for integration of PV systems in smart grid. A
possible cloud control strategy using blackboard architecture and protection has also
been discussed.
Keywords Multilevel inverter · Smart grids · Renewable energy · Cloud ·

Blackboard architecture
1 Introduction
Global warming is a phenomenon not new to us. Emission of greenhouse gases

primarily from the usage of fossil fuels, is an extensive global problem, with many
nations pledging to reduce their emissions by fixing targets [1]. The emission of CO2
related to energy production and consumption will increase 6% in between 2015 and
2050, considering ongoing and future policies. Share from non-conventional energy
sources in the supply of primary energy would rise by 14% in the same period.
Non-conventional energy sources and efficient energy usage, along with end-use
electrification, will account for around 94% of the world’s emissions reductions [2].
Electric mobility and E-vehicles also play an important role in these goals.
R. L. Josephine (B) · M. Y. Kumar · S. Harishankar

National Institute of Technology, Tiruchirappalli, India
https://doi.org/10.1007/978-981-16-4943-1_51
550 R. L. Josephine et al.
For the purposes of this paper, we’ve implemented a PV RES system. To best
analyze the effect of adopting RES with integration into Smart Grids, the project
can be broadly divided into three distinct levels; namely the Power Generation and
Distribution (PGD) level, Commercial level, and the Domestic level. In the PGD
level, the major generation of PV power takes place, followed by the industries and
services level where the presence of heavy loads and a degree of self-sufficiency
distinguishes it from the domestic level, where a typical smart home system, with
solar power production capability is present (Fig. 1).
Energy storage systems are present at both the PGD and the domestic levels, to
divert excess power and to tap into the reserves, when sudden spikes in demand
are observed. Cascaded H-bridge Multilevel Inverter was the major converter device
adopted for this model, since it allows a scalable, modularized circuit layout and
packaging, and also requires a lesser number of components for the same output
Fig. 1 Overview of CHBMLI-based smart grid system with cloud control strategy
Analysis of Cascaded Multilevel Inverters for Smart Grid … 551
voltage levels. MPPT algorithm was used for controlling the intermediate DC-DC
boost converter, before feeding the output into the inverter.
An increasingly popular mode for computation and control for demand and
response facilities, is cloud computing [3], employing huge storage and computing
facilities at remote locations, which can be shared between a host of different users.
To effectively control the system, and to protect it from a host of different cyberat-
tacks and other sabotage, the need for a novel cloud control strategy has also been
discussed.
2 PV System Integration with Smart Grid
Smart Grids are a form of advanced digital bidirectional power flow system that is
flexible, robust, and sustainable, with the ability to forecast outcomes under a variety
of conditions, and is designed with current and future requirements of cyber-secured
components, computers, and systems [4]. In [5], a novel Multilevel Converter for
on-grid interface of RES in Smart Grids is discussed, where a front-end multilevel
power electronic converter is employed for the three levels of the electric power
system, namely the power distribution, industrial and services, and the residential
levels.
Composition and description of each of the levels mentioned above in this system
will be discussed below:
2.1 Power Generation and Distribution Level
In the power generation and distribution level, we have RES in the form of a PV array,
an intermediate DC-to-DC boost converter, connected to 3-ph, 5-level Cascaded H-
bridge Multilevel Inverter (CHBMLI). It is in turn connected to the grid. An energy
storage system along with a converter is also present, connected to the grid. The
generated voltage from the PV array is stabilized using the DC-to-DC converter.
MLI converts the DC input into a 3-ph AC output, before feeding it into the grid
via the substation. Whenever excess power is generated, the Energy Storage System
(ESS) would be charged through the AC-DC converter, which then helps in meeting
sudden spikes in demand. Here, integration of solar energy as a source, and energy
storage with batteries can be done, and its QoS and control nature ensured by both
distribution and transmission system operators [5].
2.2 Commercial Level
At the Commercial Level, we have a very similar setup as that of the PGD level,
with the notable absence of an ESS. The power flow in this particular level is uni-
directional. Commercial level represents heavy industrial loads, with a high demand
for power throughout the day. To reduce the carbon footprint and harmful emissions
from this level, and to lower the dependence of the level on the grid, RES can
be integrated into this level. It also helps in lowering the consumption of energy,
thereby reducing costs. As mentioned earlier, the I&S level also has a CHBMLI as
intermediate between the PV arrays and the industrial loads.
2.3 Domestic Level
The objective of the domestic level is to represent the growing number of smart
home communities and individuals. Smart home technology is a key component
and a live dynamic communication between the power grid and consumers that
enables intelligent and interactive electricity use while also enhancing the power
grid’s operating mode and users’ consumption habits to boost energy efficiency. It
employs cutting edge technology like IoT, smart sensors and control technologies
to link multiple facilities via the network to automate the entire system [6]. Solar
power and energy storage devices can be integrated in a smaller scale to minimize
the energy costs for the end-user, and also to improve the strategies for managing
demand response, thereby establishing a renewable energy source system which is
highly distributed.
The need for a cloud-based control interface and communication strategy arises
here, for linking the three levels described above effectively. This defines the status of
the deployed inverters and other power electronic equipment, and also reports issues
in power quality, defining reference points for operation and details regarding the
transactions of power. With such specialized technology employed, the need arises
to secure the system from external actors with malicious intent. Hence, the strategy
adopted should also be well resistant to such attacks from the cyberspace.
3 Components of the Proposed System
3.1 PV Array
The solar module used for the simulation of the PV array is 1Soltech 1STH-215-P,
with 47 strings in parallel, and 10 modules per string. The maximum power of the
PV array is 213.5 W, and the open circuit voltage (V oc ) is 36.3 V, short circuit current
(I sc ) is 7.84 A. The number of cells per module (N cell I) is 60. For an irradiance of
1000 kW/m2 , at a temperature of 25° C, the power output is 100.2 kW and the output
voltage 290 V, with an output current of 345.4 A.
3.2 DC-DC Converter and MPPT Algorithm
A DC-DC boost converter is employed for boosting the output of the PV array. The
design of the converter capacitor and inductor was done to boost the array output to
a steady 700 V.
To track the peak power output of the array at all times, MPPT algorithm was
used to control the converter output. This ensures a steady output voltage even when
changes in irradiance or temperature happen. The working of the MPPT algorithm
is shown in Fig. 2.
Fig. 2 MPPT algorithm

3.3 Inverter
Inverter topologies suitable for on-grid solar power systems is discussed in [7]. In
this system, a CHBMLI is adopted for all DC-AC conversions. The advantages of
CHBMLI are many, as discussed in [8, 9]. A higher peak voltage can be obtained,
reducing the voltage load on the switches. Multilevel inverters have also proven
to be an efficient and cost-effective way to reduce switching errors in applications
demanding high voltage and power. When number of DC voltage levels increases,
the obtained output increases in stages and generates an output that resembles a
sinusoidal wave with minimal THD (Total Harmonic Distortion), eliminating the
need for an output filter.
3.4 LCL Filter
A comparative study of L, LC and LCL filters for grid connected inverters is discussed
in [10]. The advantages of LCL filter over others, and a design methodology for an
LCL filter for a 3-phase grid connected inverter, is discussed in [11].
4 Results
Power output from the solar array, and output voltage of the boost converter can be
seen in Fig. 3. These correspond to the PV array parameters specified in Sect. 3. We
observe a steady DC voltage output from the converter, which is controlled by the
MPPT algorithm, also discussed in Sect. 3. In Fig. 4, the grid voltage and current of
the commercial level is shown. The values correspond to a grid voltage of V L-L =
415 V, and a frequency of 50 Hz. The peak solar array output for an irradiance of
0.5 kW/m2 at 25° C is 49.78 kW. We observe that the grid current and voltage is in
phase, with harmonics removed using the LCL filter described earlier.
5 Need for a Cloud-Based Control Strategy
The conceptual analysis of a grid system with and without a cloud communica-
tion system is discussed in [3]. For a smart grid system that doesn’t utilize cloud
computing, all the components can interact with the utility providers directly over
the network. But the absence of a robust cloud-based control architecture, has a host
of disadvantages as discussed below:
• The master–slave architecture leads to a high exposure to cyberattacks. These
cyberattacks can be crippling to the energy infrastructure of a nation. India recently
Fig. 3 a DC-DC converter—voltage waveform and b Output power of PV array
Fig. 4 Grid voltage and current in the I&S level

Fig. 5 Blackboard architecture
witnessed this, when reportedly state backed actors were responsible behind an
attack which led to widespread outages in the grid of Mumbai.
• When a grid provides services in islanded mode, adequate management and
control are necessary. This gained prominence recently, when a historic cold
wave swept the entire Midwest and southern regions of the USA and Texas.
The blackboard architectural pattern is a popular method to solve problems char-
acterized by their uncertain and non-deterministic nature. No solutions, which are
direct and algorithmic, exist for these problems. Often, the best effort to approximate
the solution would suffice [12]. Since a smart grid would also be required to handle
uncertain load variation over a large area, this architecture would be suited for the
job.
As seen in Fig. 5, the architecture can be divided into three major subdivisions,
namely blackboard, knowledge sources and control. Blackboard acts as a global
database, and it helps in storing all the data which are collected and produced over
the period. There can be more than one knowledge source, and each source will be
assigned to take care of a particular problem. The knowledge source would get the
data from the blackboard, process that data and store the output in the blackboard.
Control acts as a manager for the whole architecture, by taking the runtime decisions
and making sure that the right knowledge sources are active at right time.
6 Conclusion
To meet the growing need for clean energy. Renewable Energy Sources and Smart
Grids are currently leading the way for future power transmissions. This paper deals
with certain aspects through which the smart grid could become more efficient.
Using the AC module-based CHBMLI gives several advantages such as easier failure
detection of the solar modules, flexible and expandable design and so on. When
smart grid grows on a large scale, communicating and coordinating becomes a key
requirement for its management. Cloud control strategies can cover large area in
real time over the Internet, making it easy for communication, and also improves
the security of critical infrastructure. On implementing blackboard architecture over

the cloud, it becomes easier to manage the smart grid as this architecture is flexible
as well as expandable. The paper also highlights the advantages of implementing
cloud-based strategy over classical solutions.
References
1. Bose BK (2010) Global warming—energy, environmental pollution, and the impact of power
electronics. IEEE Ind Electron Mag 4(1):6–17
2. Gielen D, Boshell F, Saygin D, Bazilian MD, Wagner N, Gorini R (2019) The role of renewable
energy in the global energy transformation. Energ Strat Rev 24:38–50
3. Bera S, Misra S, Rodrigues JJ (2014) Cloud computing applications for smart grid: a survey.
IEEE Trans Parallel Distrib Syst 26(5):1477–1494
4. Dileep G (2020) A survey on smart grid technologies and applications. Renew Energy
146:2589–2625
5. Monteiro V, Sousa TJ, Sepulveda MJ, Couto C, Martins JS, Afonso JL (Sept 2019) A novel
multilevel converter for on-grid interface of renewable energy sources in smart grids. In: 2019
international conference on smart energy systems and technologies (SEST), IEEE, pp 1–6
6. Li M, Gu W, Chen W, He Y, Wu Y, Zhang Y (2018) Smart home: architecture, technologies
and systems. Procedia comput sci 131:393–400
7. Deshpande S, Bhasme NR (Apr 2017) A review of topologies of inverter for grid connected
PV systems. In: 2017 Innovations in power and advanced computing technologies (i-PACT),
IEEE, pp 1–6
8. Noman AM, A Al-Shamma’a A, Addoweesh KE, Alabduljabbar AA, Alolah AI (2018)
Cascaded multilevel inverter topology based on cascaded H-bridge multilevel inverter. Energies
11(4):895
9. Azli NA, Choong YC (Nov 2006) Analysis on the performance of a three-phase cascaded
H-bridge multilevel inverter. In: 2006 IEEE international power and energy conference, IEEE,
pp 405–410
10. Yagnik UP, Solanki MD (2017) Comparison of L, LC and LCL filter for grid connected
converter. In: 2017 international conference on trends in electronics and informatics (ICEI),
Tirunelveli, India, pp 455–458. https://doi.org/10.1109/ICOEI.2017.8300968
11. Dursun M, DÖŞOĞLU MK (2018) LCL filter design for grid connected three-phase inverter.
In: 2018 2nd international symposium on multidisciplinary studies and innovative technologies
(ISMSIT), Ankara, Turkey, pp 1–4. https://doi.org/10.1109/ISMSIT.2018.8567054
12. Dong J, Chen S, Jeng JJ (Apr 2005) Event-based blackboard architecture for multi- agent
systems. In: International conference on information technology: coding and computing
(ITCC’05)-Volume II, vol 2, IEEE, pp 379–384
Voltage Stability Improvement of PV
and Battery-Based Sliding
Mode-Controlled Microgrid System
V. N. Ganesh and S. Manivannan
Abstract This work points on improving the unique time reaction of a PV and
battery energy change framework in shut circle flying capacitor support converter-
controlled force methodologies with SMC. This work manages displaying and recre-
ation of SM-controlled current impediment of framework-associated inverter. Force
procedures are a famous gadget inside the group of force framework gadget. The
voltage security is carried out with voltage source inverter (VSI), and the exchanging
beats are created utilizing sliding mode regulator. The dynamic channel is utilized
for weakening of burden voltage and current sounds. The reenactment results are
introduced to discover the impact of dynamic and responsive channel utilizing SMC.
The reproduction results with PID and sliding mode (SM) controller-based voltage
soundness are looked at, and the steady time–space restrictions are introduced. The
outcomes show that sliding mode regulator framework has improved reaction than
PID-controlled framework.
Keywords Flying capacitor boost converter · Solar panel · PWM inverter · PWM
pulse · MOSFET battery · Inverter · PI · SMC
1 Introduction
Before, power quality signified the capacity of electric utilities to give electric force
without interference. In any case, with higher infiltration of non-straight loads,
computerized gadgets and other progressed power electronic gear, power quality
at present incorporates any deviation from an ideal sinusoidal waveform which
incorporates electromagnetic interference (EMI), radio-frequency interference (RFI)
commotion, drifters, floods, lists, earthy colored outs, power outages and some other
mutilations to the sinusoidal waveform. Symphonious twists, a genuine force quality
issue can cause overheating of transformers, breaking down of hardware and even
V. N. Ganesh (B) · S. Manivannan

Dr. MGR Educational and Research Institute, Maduravoyal, Chennai, India
S. Manivannan
https://doi.org/10.1007/978-981-16-4943-1_52
560 V. N. Ganesh and S. Manivannan
reason harm to the computerized electronic control frameworks in activity. With

expanding thickness of delicate gear on the electric matrix, there is expanding tension
on the controllers to lay severe principles in regards to control quality issues. In a
microgrid network framework, brilliant meters introduced at end client areas have the
ability to decide the total harmonic distortion (THD) of the stockpile voltage. Such
data will permit the utilities to decide the wellspring of symphonious mutilations
fast. Industrialization, urbanization and unremitting foundation improvements have
prompted a change in perspective in the manner power is produced, communicated
and devoured, and it has brought about massive weight on the deep-rooted electrical
lattice framework. Furthermore, challenges because of expanding energy interest
with better caliber of force and dependability are mounting. The quick expansion in
entrance of nonlinear loads, for example, server farms, huge variable-speed drives
and other force electronic gadgets across the framework, has brought about expanded
dependability and force quality concerns.
The idea of brilliant matrix gives a large group of answers for a considerable
lot of the issues looked by the flow electric lattice by exploiting cutting edge
advances like disseminated age, circulation mechanization, energy the executives
frameworks, progressed metering foundation (AMI), environmentally friendly power
age advances, module mixture electric vehicles, two-way remote correspondence and
web, to give some examples. Expanded familiarity with the exhaustion of energy
utilization and climate, need for protected and consistent activity of the force matrix
to give top caliber and solid force supply to buyers in the advanced age have pushed
exercises in the domain of brilliant network to the front line all through the world.
The fundamental idea of a microgrid is to add checking, examination, control, and
correspondence abilities to the public electrical conveyance framework to expand the
throughput of the framework while decreasing the energy utilization. The brilliant
matrix will permit utilities to move power around the framework as proficiently
and financially as could be expected. The keen network can be characterized as a
framework that utilizes computerized data and control innovations to encourage the
arrangement and coordination of conveyed and inexhaustible assets, brilliant buyer
gadgets, mechanized frameworks, power stockpiling and pinnacle shaving advances.
With higher infiltration levels, the electrical network would require all the quicker
beginning and quick inclining assets to compensate for the age deficiencies when
such assets are not working at their normal yield levels. Significant measure of explo-
ration is going on in this field, and proposed framework like the future renewable
electric energy delivery and management (FREEDM) guarantees a considerable lot
of the issues. The goal is to have an effective electric force network coordinating
profoundly stochastic, dispersed and versatile elective age sources and fuel stock-
piling with existing force frameworks to encourage a green and feasible energy-
based society, alleviate the developing energy emergency and the effect of fossil
fuel byproducts on the climate. In this way, the presentation of shrewd matrices will
not just decrease ozone depleting substance discharges yet additionally empower
expanded incorporation of sustainable sources and fuel stockpiling resources with
the electric force network.
Voltage Stability Improvement of PV … 561
PV array and flying capacitor boost converter with inverter in closed loop grid-
connected system with PID and SMC controller are shown in Fig. 1. Closed loop
circuit diagram of flying capacitor boost converter with grid-connected inverter PID
and SMC controller is shown in Figs. 2 and 3.
Fig. 1 Block diagram of PV, battery and flying capacitor boost converter with inverter in closed
loop grid-connected system with PID and SMC controller
Fig. 2 Circuit diagram of closed loop PID-controlled PV and battery-based microgrid-connected

system
Fig. 3 Circuit diagram of closed loop SM-controlled PV and battery-based microgrid-connected

system
The bar outline examination of voltage THD and current THD on PV with inverter and
PV and battery of lift converter based inverter is appeared in Fig. 4. Circuit chart of
PV and battery based on microgrid-associated inverter with shut circle PID regulator
is addressed in Fig. 5 Voltage across PV and battery based on microgrid-associated
inverter framework is appeared in Fig. 6, and its worth is 22 V. Voltage across PV
and battery based on microgrid-associated inverter framework is appeared in Fig. 7,
Fig. 4 Circuit diagram of PV and battery based on microgrid-connected inverter with closed loop
PID controller
Fig. 6 Voltage across flying capacitor boost converter
Fig. 7 Voltage across RL load
and its worth is 22 V. Voltage across RL heap of PV and battery based on microgrid-
associated inverter framework is appeared in Fig. 8, and its worth is 120 V. RMS
voltage across RL heap of PV and battery based on microgrid-associated inverter
framework is appeared in Fig. 9, and its worth is 80 V. The voltage THD esteem is
fundamental 50 Hz = 100, and THD esteem is 2.12% is appeared in Fig. 10. The
current THD esteem is fundamental 50 Hz = 7.07, and THD esteem is 2.04% and is
Fig. 8 RMS voltage across RL load

Fig. 9 Voltage THD
Fig. 10 Current through RL load
appeared in Fig. 11. Output force of PV and battery based on microgrid-associated

inverter framework is appeared in Fig. 12, and its worth is 320 W.
Circuit outline of PV and battery based on microgrid-associated inverter with
closed loop SM controller is addressed in Fig. 13. Voltage across PV and battery
based on microgrid inverter framework is appeared in Fig. 14, and its worth is 22 V.
Voltage across flying capacitor support converter of PV and battery based on micro-
grid network-associated inverter framework is appeared in Fig. 15, and its worth
is 60 V. Voltage across RL heap of PV and battery based on microgrid-associated
inverter framework is appeared in Fig. 16, and its worth is 120 V. RMS voltage
across RL heap of PV and battery based on microgrid-associated inverter framework
Fig. 11 Current THD

Fig. 12 Output power
Fig. 13 Circuit diagram of PV and battery based on microgrid-connected inverter with closed loop
SM controller
Fig. 15 Voltage across flying capacitor boost converter

Fig. 16 Voltage across RL load
is appeared in Fig. 17, and its worth is 80 V. The voltage THD esteem is funda-
mental 50 Hz = 108.5, and THD esteem is 1.61% and is appeared in Fig. 18. Current
through RL heap of PV and battery based on microgrid-associated inverter framework
is appeared in Fig. 19, and its worth is 8 A. The current THD esteem is fundamental
50 Hz = 8.116, and THD esteem is 1.36% and is appeared in Fig. 20. The current
THD esteem is fundamental 50 Hz = 100, and THD esteem is 1.36% and is appeared
in Fig. 21. Yield force of PV and battery based on microgrid framework-associated
inverter framework is appeared in Fig. 22, and its worth is 300 W.
Table 1 gives the correlation of time-domain boundaries with PID and SMC
controller of PV and battery based on microgrid network-associated inverter frame-
work. Utilizing SMC regulator, the ascent time is decreased from 0.38 to 0.33 s, the
settling time is diminished to 0.84 from 1.10 s, the rush hour is diminished from 0.73
to 0.42 s, and the consistent state mistake is diminished to 0.43–1.33 V.
Fig. 17 RMS voltage across RL load
Fig. 18 Voltage THD

Fig. 19 Current through RL load
Fig. 20 Current THD
Fig. 21 Output power
Fig. 22 Bar chart comparison of time-domain parameters for PID and SMC controller
Table 1 Comparison of time-domain parameters

Controller Tr Ts Tp Ess
PID 0.38 1.10 0.73 1.33
SMC 0.33 0.84 0.42 0.43
Table 2 Comparison of
Controller Voltage THD (%) Current THD (%)
voltage and current THD
PID 2.12 2.04
SMC 1.61 1.36
Fig. 23 Bar chart comparison of voltage and current THD of PID and SMC controller
Figure 22 shows the bar graph correlation of time-domain boundaries for PID and
SMC controller. Table 2 gives the examination of voltage and current THD of PID
and SMC controller of PV and battery based on microgrid framework-associated
inverter system. Fig. 23 gives the bar graph examination of voltage and current THD
of PID and SMC controller.
4 Conclusion
The flying capacitor support converter-based dynamic and receptive channel utilized
for framework-associated voltage source inverter in shut circle is reproduced, and
the outcomes with PID and SMC are introduced. By utilizing SMC, the ascent time
is lessened from 0.38 to 0.33 s; the rush hour is reduced from 0.73 to 0.42 s; the
settling time is decreased from 1.10 to 0.84 s; the consistent state blunder is reduced
from 1.33 to 0.43 V. Thusly, SM-based dynamic and responsive channel might be
a feasible option in contrast to the current framework. The decrease in settling time
and the consistent state mistake Reponses is better on account of SMC framework.
The benefits of the proposed framework are the decrease in voltage THD and current
THD, line drop, consistent state blunder and settling time. The impediment is that it
requires an enormous inductor and a capacitor.
The present work deals with the investigation PID- and SMC-controlled grid-
connected inverter system. The improvement with HC will be done in future.
References
1. Blaabjerg F, Chen Z, Kjaer SB (2004) Power electronics as efficient interface in dispersed power
generation systems. IEEE Trans Power Electron 19(5):1184–1194
2. Wang F, Duarte JL, Hendrix MAM, Ribeiro PF (2011) Modeling and analysis of grid harmonic
distortion impact of aggregated DG inverters. IEEE Trans Power Electron 26(3):786–797
3. Asiminoaei L, Blaabjerg F, Hansen S, Thogersen P (2008) Adaptive compensation of reactive
power with shunt active power filters. IEEE Trans Ind Appl 44(3):867–877
4. Asiminoaei L, Blaabjerg F, Hansen S (2007) Detection is key—Harmonic detection methods
for active power filter applications. IEEE Ind Appl Mag 13(4):22–33
5. Chakroborty S, Weiss MD, Simoes MG (2007) Distributed intelligent energy management
system for a single-phase high-frequency ac microgrid. IEEE Trans Ind Electron 54(1):97–109
6. Brahma SM, Girgis AA (2004) Development of adaptive protection scheme for distribution
systems with high penetration of distributed generation. IEEE Trans Power Del 19(1):56–63
7. Hossain MR, Maung Than Oo A, Shawkat Ali ABM (2010) Evolution of smart grid and some
pertinent issues. In: 2010 20th Australasian universities power engineering conference (AUPEC),
pp 1–6, 5-8 December 2010
Electric Propulsion System
with Dual-Motor Power Management
Strategy
Vinoth Kumar Balan, P. Avirajamanjula, and A. Dominic Savio
Abstract Dual-motor electric powertrain offers the independent control and less
energy consumption in plug in hybrid electric vehicle model. The component sizing
and control strategy are play the important role in dynamic control, power manage-
ment and energy saving. The proposed system vehicle architecture comprises the dual
motor, downsized engine and synchronizer. For optimizing the dynamic load sharing
with the dual-motor powertrain configuration, rule-based fuzzy logic control strategy
is implemented. The variable load will be managed by the way of selecting different
transmission topologies using fuzzy logic control strategy. This dual-motor archi-
tecture configuration is simulated using Simulink software. Single motor propulsion
drive control is engaged for all driving pattern except the peak load condition. In
peak load, such as acceleration mode and high-speed mode, the dual-motor power
transmission is engaged. The powertrain configuration is test at different dynamic
and load conditions. As a result, it shows the better drivability, high efficiency and
extended driving range.
Keywords Motor · Fuzzy · Plug in hybrid vehicle · Power management · Driving

cycle · Series hybrid electric vehicle
1 Introduction
The day-by-day usage fossil fuel rapidly increasing, due to this rising oil prices,
makes more impact on transport sector and economic. On other hand, the usage
of oil causes the air pollution, global warming, acid raining, etc. So, most of the
governments are encouraging the people to use electric vehicle [1–3]. The pure
V. K. Balan (B) · P. Avirajamanjula

Department of Electrical and Electronics Engineering, PRIST University, Thanjavur, India
P. Avirajamanjula
A. D. Savio
Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_53
572 V. K. Balan et al.
Fig. 1 Powertrain configurations for plug-in hybrid electric vehicle
electric vehicle has features of zero emission, economic operation, less maintenance,
etc. Even though pure electric vehicle has problem of long-range driving problem
due to limited energy storage, long charging periods, the people are reluctant to
buy the electric vehicle [5–9]. So, auto industries and many researchers are finding
feasible solution for electric vehicle. Plug-in hybrid electric vehicles have features
of extended driving range, grid charging, multi-mode of operation the powertrain
configuration of PHEV as shown in Fig. 1.
Pem
Hybridization factor = (1)
Pem + Peng
Pem
Hybridization factor = (2)
Phve
FTotal = FI + Fs + Fr + Fa (3)
FI = ma (4)
Fs = m.g. sin θ (5)
Fr = m.g.Crr. cos θ (6)
1
Fa = ρCd AV 2 (7)
2
PTotal = FTotal .V (8)

Electric Propulsion System with Dual-Motor Power … 573
Table 1 Value of parameters

Name Value
for the simulation of the
powertrain Electric motor A power 59 KW
Electric motor B power 40 KW
Engine 50KW
Battery capacity 110 Ah
Vehicle loaded mass 1400 kg
Air resistance coefficient 0.01
Number of batteries 25
Rolling resistance coefficient 0.012
Frontal area, Af 2.33 m2
Drag coefficient, C d 0.26
Air density, ρ air 1.22 kg/m3
Wheel radius, Rwh 0.32 m

E Total = PTotal .dt (9)
1
E Total = [m.g( f cos θ + sin θ )
3600
dv
+0.386(ρCd AV ) + m + m f
2
D (10)
dt
SOCmin ≤ SOC(t) ≤ SOCmax (11)
Pmin ≤ P(t) ≤ Pmax (12)
where the Pem is the power produced by the electric motor, Peng is the power produced
by the ICE engine, Phve is the total power the hybrid power propulsion system,
Fr is aerodynamic drag, FI is inertial force, Fs is the road slope force, Fa is the
aerodynamic drag force, PTotal is the total traction power, FTotal is the traction force,
V is the vehicle speed, E Total is vehicle energy required, road load coefficient, Crr
is the road load coefficient, g is the gravitational acceleration, m is the mass of the
vehicle, θ is the road slope angle, ρ is the air density, Cd is the air drag coefficient,
A is the vehicle frontal area, D is the distance of travelling (Table 1).
2 Modes of Hybrid Drive System
The plug-in hybrid vehicle, powertrain, comprises the dual motor, downsized engine
and synchronizer. The rule-based fuzzy logic controller is developed for various
Fig. 2 a Single motor drive control, b series hybrid powertrain, c dual-motor drive control, d dual-
motor series hybrid powertrain
load conditions of transmission pattern [10–12]. The different powertrain patterns

are shown in Fig. 2. The SoC of the battery and power demand of vehicle are
the two main controlling parameters. Based these values rule-based controller will
decide the configuation of powertrain elements. The operating mode of the vehicle
is mainly classified as start/stop, acceleration, cruising, de-acceleration, idle. The
power demand of the vehicle is very high, when its operating either in cruising mode
or acceleration mode. The acceleration and high-speed mode, i.e., cruising mode the
power demand of the vehicle is so high [13–15]. In this peak power demand condi-
tion, the torque requirement of the vehicle is met by the dual-motor powertrain. The
remaining mode of operation is managed by the single motor powertrain. The result
of single motor power configuration, extended driving range, fuel economy, extended
battery life can be obtained. The result of single motor driving, this system can obtain
fuel economy, battery saving, extended driving range.
3 Operation Mode Transition
The powertrain driving propulsion pattern condition is explained through flow chart
as shown in Fig. 2. Fuzzy controller evaluated the SoC of the battery and vehicle
travelling distance (D). If energy storage system parameter is in acceptable range,
i.e., (SoC > D), the driving pattern will be identified either dual-motor drive or
single-motor drive condition. In case, SoC < D, the engine turns to on condition for
charging the battery. The transient condition is periodically evaluated and improved
by the rule-based fuzzy logic controller. If SoC is 0.6 < SoC < 0.8 and the power
demand is P_dem ≤ P_em1, condition results the motor A met the torque requirement
of the vehicle. Even SoC of the battery in the range of 0.6 to 0.8 and power demand is
high (P_dem ≥ P_em1), the dual-motor powertrain configuration will met the torque
requirement of the electric vehicle (Fig. 3).
Fig. 3 Flow diagram of the control strategy
4 Simulation
In this proposed method, rule-based fuzzy controller was designed for the plug-
in electrical vehicle to enhance the fuel economy and battery energy savings. The
control strategy of fuzzy membership is created based on the power demand and SoC
of the battery as shown in Fig. 4. The fuzzy membership for load control is shown in
Fig. 4. For the different load condition, the controller will meet the power demand
of the vehicle, based the load sharing in propulsion system (Figs. 5 and 6).
Fig. 4 Fuzzy membership for input

Fig. 5 Fuzzy membership for output
Fig. 6 3D Fuzzy control rule
In order to verify the energy consumption of the powertrain, the proposed system
is simulated for various driving conditions. Dual-motor powertrain power output is
shown in Fig. 7a. Low SoC condition was initiated for operating engine and generated
in charging mode. The powertrain element shaft output is shown in Fig. (b). In start
mode, the single motor driving pattern will meet the torque requirement of the vehicle
from 0 to 2 s time period. The battery power loss is minimum in starting condition,
and SoC of the battery is improved in this mode as shown in Fig. 7c, d.
When vehicle moves to the accelerating mode during the time interval of 2–5 s,
the high-power demand occurs and the controller will optimize the load sharing of
propulsion elements. In this mode, the dual-motor powertrain is configured. So, the
powertrain requires consume more charge for the propulsions of the vehicle; due to
this, the battery power loss increases rapidly and SoC decreases linearly. The vehicle
was entering into cruising operation after the acceleration mode, so the controller will
estimate the power demand and identifies feasible control strategy for this mode. In
this cruising mode, vehicle operates by single motor powertrain to provide constant
speed. So, the battery power loss is constant and SoC is linearly decreasing. The
Fig. 7 a Dual-motor powertrain output. b Powertrain element shaft output. c Battery power loss.
d Battery SoC at various load conditions
effect of the control strategy results, better dynamic control and extend driving range
was obtained.
5 Conclusion
In this work, extending driving range of vehicle dual-motor propulsion powertrain

is configured. By using the rule-based fuzzy logic controller, the control strategy
is developed for different driving patterns. This system is design using Simulink
software and tested for various load dynamic conditions. Peak load condition in dual-
motor powertrain met the power demand of the load. All other load conditions are
managed with single motor powertrain configuration. Compared to the conventional
hybrid model, dual-motor powertrain offers the extended electric mode of operation.
The result shows the better drivability and improved overall performance.
References
1. Wang F, Fang Z, Zhu X (2011) Matching, simulation and optimization of the new power
transmission device for an electric vehicle. Autom Eng 33(9):805–808
2. Wang Y, Sun D (2014) Powertrain matching and optimization of dual-motor hybrid driving
system for electric vehicle based on quantum genetic intelligent algorithm, vol 2014. Hindawi
Publishing Corporation
3. Lv C, Zhang J, Li Y, Yuan Y (2015) Mechanism analysis and evaluation methodology of
regenerative braking contribution to energy efficiency improvement of electrified vehicles.
Energ Convers Manag 92:469–482
4. Zhang L, LiuW, Qi BN (2020) Energy optimization of multi-mode coupling drive plug-in
hybrid electric vehicles based on speed prediction. Energy 206
5. Wang Y, Zeng X, Song D (2020) Hierarchical optimal intelligent energy management strategy
for a power-split hybrid electric bus based on driving information. Energy 199
6. Xu J-H, Guo J-F, Peng B, Nie H, Kemp R (2020) Energy growth sources and future energy-
saving potentials in passenger transportation sector in China. Energy 206
7. Goetz M (2005) Integrated: powertrain control for twin clutch transmissions. Univ. Leeds,
Leeds, U.K
8. Wang W, Li Y, Shi J, Lin C (2018) Vibration control method for an electric city bus driven
by a dual motor coaxial series drive system based on model predictive control.IEEE Access
6:41188–41200
9. Wang W, Zhang Z, Shi J, Lin C, Gao Y (2018) Optimization of a dual-motor coupled power-
train energy management strategy for a battery electric bus based on dynamic programming
method.IEEE Access 6:32899–32909
10. Ruan J, Walker PD, Zhang N, Wu J (2017) An investigation of hybrid energy storage system
in multi-speed electric vehicle.Energy 140:291–306
11. Morozov A, Humphries K, Zou T, Martins S, Angeles J (2014) Design and optimization of a
drivetrain with two-speed transmission for electric delivery step van.In: Proceedings of IEEE
international electric vehicle conference (IEVC), Florence, Italy, December 2014, p 18
12. Kim H, Kim J, Lee H (2011) Mode transition control using disturbance compensation for a
parallel hybrid electric vehicle. Proc IMechE Part D J Autom Eng 225:150–166
13. Enang W, Bannister C (2017) Modelling and control of hybrid electric vehicles (A compre-
hensive review). Renew Sust Energ Rev 74:1210–1239
14. Naunheimer H, Bertsche B, Ryborz J et al (2010) Automotive transmissions: fundamentals,
selection, design and application. Springer Science & Business Media, Dordrecht
15. Enang W, Bannister C (2016) Robust proportional ECMS control of a parallel hybrid electric
vehicle. Proc IMechE Part D J Autom Eng 231:99–119
Realization and Implementation of Peak
Energy Management
P. Kanakaraj and L. Ramesh
Abstract Peak energy management (PEM) is an important tool in energy audit,

which will manage and reduce the excess power demand required during peak hours.
The role of PEM is to reduce the peak demand with the help of pricing tariffs and
incentives for the consumers. The PEM is reviewed with techniques, applications,
and ancillary services. Based on that, review new model is proposed. Here, we use a
PV panel to charge the battery which is connected to the load through an inverter and
switch. The grid is connected to the load through a switch as well. Hence, using this
model enables us to supply power to the load with the help of a battery during peak
hours, and the PV panel charges the battery during off-peak hours. Thus, it enables
us to reduce the peak hour consumption and the bill cost as well.
Keywords Peak energy management · Demand response program · Renewable

energy system · Battery energy storage system · Thermal energy storage · Diesel
generator · Time of use · Multi-agent systems
1 Introduction
In the power system, the balancing of power demand is the major task [1]. In recent
years, energy management is playing a vital role in balancing of power demand. It
is a perfect solution for the reduction of energy consumption with the help of solar,
wind, biomass, battery storage, and demand-side management (DSM) techniques. It
is the skill and knowledge of optimal use of energy to maximize profits (minimize
costs) and thereby increase the financial competitiveness. It can save up to 70% of
the energy consumption in a typical building or plant, which is achieved with the
help of required investment in transmission and distribution system, and also require
proper monitoring system. It has forecast and development of energy production and
P. Kanakaraj · L. Ramesh (B)

Dr.M.G.R. Educational and Research Institute, Maduravoyal, Chennai, India
P. Kanakaraj
https://doi.org/10.1007/978-981-16-4943-1_54
580 P. Kanakaraj and L. Ramesh
energy consumption units. It is used by industries, residences, commercial buildings,

and islanded systems. It is the procedure of monitoring, controlling, and preserving
energy in a building or an industry. Energy management program has the major
elements like setting your goal (size of energy reduction), know your numbers (size
of requiring consumption), reduction in consumption (major consumption units to
be reduced), and continuous review and management.
The energy should be used economically, efficiently, and optimally which can be
identified with the help of energy efficiency [2]. Energy efficiency is very important
for managers of commercial buildings. Commercial buildings are utilizing the energy
in the form of heating, cooling, lighting, and hot water. The major aim of energy
managements is resource preservation, climate protection, and cost savings. The
population is increasing day by day, and due to this, electricity demand also increased
with a higher number of electrical appliances which is affecting the reliability and
efficiency of the utility grid.
Peak electricity demand was creating the constraints and congestion for transmis-
sion, which also increases the electricity cost for all the end-use consumers. Distribu-
tion and transmission infrastructure should be upgraded for providing power during
the peak demand period. Normally, service providers charge a higher price at peak
time and normal price during the off-peak time [4].
DSM helps to maximize the energy efficiency of all consumers and reduces the
peak load and total power utilization [3]. The different DSM techniques are peak
shaving (peak clipping), load shifting, flexible load pattern, and reduction of energy
consumption. These techniques have the following benefits like reduction in elec-
tricity bills, managing the ancillary services, reduction requirement of new power
plants, and upgradation of requiring transmission and distribution networks. The
integration of storage devices into renewable energy systems requires novel load
management techniques to manage more complex energy systems. This complex
renewable energy system creates full of uncertainties like renewable energy sources,
errors in renewable energy forecast, electricity demand, and electricity prices.
The structure of this paper is as follows: Sect. 2 explained the different existing
peak energy management articles in the view of different techniques, different appli-
cations, and different ancillary services. Open issues and research directions are
discussed in Sect. 3. Section 4 is deliberated the proposed work and its results.
Finally, this paper is concluded in Sect. 5.
2 Recent Research on Peak Energy Management
Peak energy management articles were discussed based on the techniques to reduce
the peak demand, applications where peak energy management was proposed, and
ancillary services which were helpful to reduce the peak demand. Some of the above-
mentioned topics are discussed further in this section.
Realization and Implementation of Peak Energy Management 581
2.1 Techniques
Peak energy management techniques are implemented to reduce the peak demand
during peak time. These techniques are classified into the time of use (TOU),
peak shaving, peak shaving and valley filling, multi-agent program, peak strate-
gies based decision-making process, real-time pricing, variable peak pricing, critical
peak pricing, interruptible/curtailable service, capacity market program, and direct
load control. The few of the above techniques are implemented recently, and in this,
some of them are discussed below based on the results obtained.
(i) Time of Use
Ali Mehdizadeh et al. proposed an information gap decision theory (IGDT) for
obtaining the bidding strategy of micro-grid (MG) [6]. MG considered the two
different uncertainty environments like robustness decision (risk-averse) and oppor-
tunity decision (risk-taking). The main purpose of the demand response program
(DRP) in MG was a reduction in energy procurement cost.
Ilze Laicane et al. investigated a study of demand-side management, which reduces
the peak load of residential houses [7]. This article considered a household with four
members and focused on two dissimilar appliances like a washing machine and
dishwasher for load shifting. DRP was used to identify the pattern of household
appliances. It was noted with the help of a smart meter installed in the houses. TOU
or load shifting algorithm was implemented in the DRP. Figure 1 shows the impact
of optimization on peak-time demand.
(ii) Peak Shaving
Fig. 1 Impact of optimization on peak-time demand

Sanjeev et al. designed a real-time simulation of different practical scenarios in

the residential houses like peak, off-peak, and isolated modes of operations in
RSCAD/RTDS software platform to bring the comprehensive power flow control
strategy (PFCS) [9]. These modes were operated based on the RES, BESS, and
ultracapacitor.
Fakeha Sehar et al. proposed an ice storage technology to shift the air-conditioning
(AC) load of the medium-sized office building to off-peak periods, which reduced
the total electricity consumption to 17.8% [5]. This study was performed during the
summer season with a different combination of DRP, solar power generation, and
cold storage systems.
(iii) Multi-Agent System
Hantao Huang et al. proposed a multi-agent minority game-based energy manage-

ment system (MG-EMS) to reduce the peak electricity demand on the main grid [13].
The static and centralized energy management system (SC-EMS) was considered a
traditional energy management system, which was taken as a base case. Experimental
results of MG-EMS and uncertainty-aware minority game-based energy management
systems (UAMG-EMS) were compared with the SC-EMS method. MG-EMS was
mainly framed for reducing the peak demand of the grid and also providing addi-
tional support to the grid with solar energy. Figure 2 shows the summer and winder
energy cost savings comparison with different energy management systems.
(iv) Peak Strategies (Decision Making)
Cristina Azcarate et al. planned a new management policy named Peaking Strategies.
In this peaking strategy, electricity was converted into hydrogen at the off-peak price
Fig. 2 Summer and winter energy cost savings comparison with different energy management
systems
period and similarly, hydrogen was converted into electricity at peak price periods
[14]. Also, stored hydrogen was used to convert as electricity during the high peak
demand in the daytime when the grid reinforcement was high.
2.2 Applications
Peak energy management is a very important tool to reduce peak demand during
peak time and is implemented in many applications like rural areas/islanded power
systems, residential consumer, industrial consumers, and smart buildings (medium-
sized office building/high-rise residential building). In this section, some of the
applications are discussed as follows.
(i) Islanded Power Systems/Rural Areas
Spyridon Chapaloglou et al. proposed a new algorithm of EMS for an islanded power
system [12]. In this algorithm, we combined diesel generators, a PV farm, and a
battery energy storage system (BESS) with the help of load forecasting methodology
and pattern recognition procedure. Figure 3 shows the developed neural network for
load forecasting.
Michael T. Wishart et al. proposed a novel peak load management for the rural
areas with diesel generator (DG) [10]. In order to avoid peak demand during peak
time, some parts of the residential consumers were supplied with the help of DG.
(ii) Residential Consumer
Jagruti Thakur et al. designed the load balancing and load shifting for effective
management of the power grids with the help of DSM mechanisms [1]. The main
objective of this paper was to control the monthly bills as well as to discover the
dynamics of demand response-based tariff mechanisms on consumer load profiles.
DSM program was applied to obtain the economic savings for the residential
consumer and classified into price-based technique and incentive-based technique.
Fig. 3 Developed neural network for load forecasting

Krystian X. Perez et al. designed a centralized model predictive control (MPC)

scheme for residential consumer to minimize the peak demand [15]. The scheduling
operation was formulated to minimize peak demand with a mixed-integer linear
program (MILP) for time-shiftable individual appliances like dishwashers, washing
machines, and dryers in each house.
K. Keerthi Jain et al. discussed the saving of energy with the help of effective
energy audits [18]. In this paper, the energy audit was conducted for a double bedroom
house in Chennai. After the critical energy audit, the author recommended the few
implementations in that house with and without investments to achieve more energy
savings.
(iii) Industrial Consumer
C. A. Babu et al. presented optimized load management for electrolytic process

industries [8]. Load management was implemented with the help of mixed-integer
nonlinear programming (MINLP) technique to minimize the electricity cost and
reduce the peak demand by rescheduling the loads and satisfying the industrial
constraints.
S. Ashok presented a common load model for any type of process industry, which
was based on the characteristics of batch-type loads [16]. The main objective is
minimizing the total electricity cost with an optimal response for any industry under
time-varying tariff. The case study of a steel plant was considered and obtained the
significant reduction in peak-period demand around 50% and total electricity cost
around 5.7% with an optimal load schedule.
Dongmin Yu et al. implemented the TOU method in the DRP to the industrial
consumers [2]. The total electricity cost of the industry was compared between
deterministic method and the proposed method with and without considering DRP.
Finally, the total electricity cost was reduced by 8.2% for the optimization method
and 6.5% for the deterministic method with the help of DRP.
(iv) Smart Buildings
Fakeha Sehar et al. simulated a medium-sized office building to reduce the peak
load and energy utilization, which was located in Virginia, USA [5]. The building’s
peak electricity demand and energy consumption were reduced with the help of solar
power generation, DR, and cold storage systems. In Table 1, savings in peak load,
energy utilization with partial, and full ice storage are displayed. Figure 4 shows
the power consumption profiles of the simulated building with combinations of PV
generation, full ice storage, and DR.
Yu Wang et al. investigated the peak shaving and valley filling potential of the
energy management system (EMS) in a high-rise residential building with the help
of multi-agent system (MAS) framework with equipped PV storage system [17].
The simulation was done with the help of EMS, which flattened the HRB electricity
demand profiles, and also, the peak-to-valley ratio of demand profiles was reduced
much.
Table 1 Savings in peak load, energy utilization with partial and full ice storage
Peak load at 4:10 pm (kW) Energy consumption for the
simulation day (GJ)
Building HVAC Building HVAC
Without ice storage 223.40 107.34 10.42 4.52
(conventional DX unit)
With full ice storage 128.37 12.31 16.88 10.98
With partial ice storage 190.22 74.15 14.52 8.63
Fig. 4 Power consumption profiles of the simulated building with combinations of PV generation,
full ice storage, and DR
2.3 Ancillary Services
Ancillary services helped to compensate the peak demand during peak time. It is clas-
sified into renewable energy sources (RES), battery energy storage system (BESS),
thermal energy storage (TES)/ice storage, diesel generator (DG), and ultracapacitor.
The following ancillary services are discussed in the recent peak energy management
articles.
(i) Renewable Energy Sources (RES)
Surekha Dudhani et al. proposed the optimal allocation of various renewable energy
sources (RES) to meet the peak load demand at various places in India [11]. This
proposed algorithm was implemented by using a linear programming algorithm with
various constraints. Different RESs were small hydel power, solar PV, wind power,
co-generation, and biomass energy which was varied depending upon the places.
Table 2 Cost savings of combined TES and PV power generation

Cost savings (e) Cost savings (%)
Demand Energy Total Demand Energy Total
Denver 6167 12,636 18,803 12.7% 8.8% 9.8%
Nebraska 5697 12,526 18,223 11.7% 8.8% 9.5%
Syracuse 5580 10,834 16,414 11.5% 7.6% 8.6%
(ii) Battery Energy Storage System (BESS)
Dongmin Yu et al. proposed a novel optimization technique for modeling of uncer-

tainty in the electricity market [2]. The industrial customer was taken energy from
the electricity market, reciprocal agreements, small-scale turbines, energy storage
devices (ESD), wind generation, and solar panel generation with DRP. The newly
proposed optimization technique was implemented with the help of the risk-averse
method, which was also recommended to compare with the risk-neutral method.
(iii) TES/Ice Storage
Mohammad Saffari et al. proposed the optimized DSM techniques with TES and
PV power generation for peak load reduction in commercial and industrial power
sectors [3]. The proposed DSM technique results confirmed the effectiveness of the
peak demand reduction when compared with TES and PV individually. These tech-
niques were implemented in the dairy factory which is located in the three different
places like Denver, Nebraska, and Syracuse in Spain. Table 2 displays the cost savings
of combined TES and PV power generation system.
(iv) Diesel Generator (DG)
Michael T. Wishart et al. presented a new peak load management for the rural
areas with the help of diesel generator (DG) [10]. In this peak load management,
some customer’s power demand was shifted to the local non-embedded generators
at peak time. This paper considered a case study in Central Queensland, Australia.
(v) Ultracapacitor
P. Sanjeev et al. proposed the reduction of the peak demand shortage available in
the grid-connected system and also reliable power supply provided even in case of
blackout or grid failure [9]. A comprehensive PFCS was designed as a real-time
simulation of different practical scenarios in the residential houses like peak, off-
peak, and isolated modes of operations in RSCAD/RTDS software platform with the
help of RES, BESS, and ultracapacitor.
3 Open Issues and Research Directions
The following inferences are identified from the peak energy management articles
which are discussed above.
• One of the major issues is creating an awareness among the public about peak,
off-peak, and mid-peak hours in peak energy management [1].
• The requirement of customized tools of demand-side management is important
for flexible loads in smart buildings [5].
• The proposed method (combined TES and solar PV) in the dairy factory should
be applied to all other food processing factories, supermarkets, and any other
industries which required significant cooling demand [3].
• The renewable generation’s uncertainty modeling should be designed with the
help of combined stochastic programming, information gap decision theory, and
robust optimization approach [6].
• The home energy management systems (HEMS) would be designed for the indi-
vidual residential consumer with the help of energy storage systems and the
influence of distributed generation systems [15].
• The development of solar PV storage technique would be made with higher effi-
ciency and lower investment cost to promise larger peak shaving and valley filling
[17].
• Real problems in industrial and healthcare contexts would be addressed with the
combination of simulation as well as optimization techniques [14].
• In the future, the fuel consumption of the diesel generators and detailed power
inverter configuration would be considered [12].
• The DCMG will be implemented only where the conventional grid does not
exist such as low power domestic consumers, remote villages, data-centers, and
telecommunication systems [9].
• The latest optimization technique would be extended to industries where the
controllable loads are running continuously. Here, peak demand would be reduced
with the help of rescheduling of their process [16].
4 Proposed Work with Results
This proposed work consists of buck/boost converters, a load (bulb), a battery for
storage of energy, and a solar panel. Figure 5 shows the block diagram of the proposed
work. The switch S2 connects the grid and load. The switch S3 connects the battery
and load. Figure 6 shows the hardware model of the proposed system.
During the off-peak hours, the power is directly supplied to the load from grid as
no excess load is demanded at that time. During the daytime, the battery is directly
charged with the help of the solar panel as ample amount of sunlight is available. If
the SOC of the battery is low then the battery is charged when the off-peak period
Fig. 5 Proposed block diagram
Fig. 6 Proposed system hardware model
starts (after 12 am). And during the peak hours, the grid is disconnected from the
load, and the battery is connected to the load for supplying power.
The battery can be connected directly to the solar panel for charging as the system
deals with a power of 5 W and 12 V, and no charge controller is required. However,
a proper and standard charge controller will be required for a higher rating system.
From Fig. 7, it is shown that the model is in OFF condition and no supply is given
to the bulb of 0.5 watts, neither from the grid nor from the battery.
From Fig. 8, it can be depicted that the system is working. However, it is not
possible to distinguish from the picture whether the supply is given by grid or battery.
When Switch 2 is turned on, the supply to bulb is given by the grid. When Switch
Fig. 7 System at OFF condition
Fig. 8 System at ON condition

2 is turned off and Switch 3 is turned on then the supply to the bulb is given by the
battery.
5 Conclusion and Future Scope
In this paper, a review of PEM is discussed with some inference. Based on that infer-
ence, new hardware model is proposed with an objective of reduction in peak load
and reduction in consumer’s electricity consumption cost. This proposed system can
also be implemented on a larger scale. Some of the future scopes for improvements
in the proposed systems are listed below.
• As the above process requires collection of readings and detailed analysis of load
during night time, another renewable source such as wind can be utilized to charge
the battery during the night time instead of the grid.
• Further, our system at present plans to deal with the load of a single house. With
proper implementation and planning and strategy, this system might be utilized
for an apartment and later a community with about 40–50 houses.
• This system can also be modified into one where the battery will be used to supply
power during the peak hours to the lighter loads in a household like fans and tube
lights.
• Also, the system that we have proposed is improvised into an automated one using
intelligent controller unit.
References
1. Thakur J, Chakraborty B (2016) Demand side management in developing nations: a mitigating

tool for energy imbalance and peak load management. Energy 114:895–912
2. Yu D, Liu H, Bresser C (2018) Peak load management based on hybrid power generation and
demand response. Energy 163:969–985
3. Saffari M, de Gracia A, Fernández C, Belusko M, Boer D, Cabeza LF (2018) Optimized
demand side management (DSM) of peak electricity demand by coupling low temperature
thermal energy storage (TES) and solar PV. Appl Energ 211:604–616
4. Yao L, Yang B, Cui H, Zhuang J, Ye J, Xue J (2016) Challenges and progresses of energy storage
technology and its application in power systems. J Mod Power Syst Clean Energ 4(4):519–528
5. Sehar F, Pipattanasomporn M, Rahman S (2016) An energy management model to study energy
and peak power savings from PV and storage in demand responsive buildings. Appl Energ
173:406–417
6. Mehdizadeh A, Taghizadegan N, Salehi J (2018) Risk-based energy management of renewable-
based microgrid using information gap decision theory in the presence of peak load
management. Appl Energ 211:pp. 617–630
7. Laicane I, Blumberga D, Blumberga A, Rosa M (2015) Reducing household electricity
consumption through demand side management: the role of home appliance scheduling and
peak load reduction. Energ Procedia 72:222–229
8. Babu CA, Ashok S (2008) Peak load management in electrolytic process industries. IEEE
Trans Power Syst 23(2):399–405
9. Sanjeev P, Padhy NP, Agarwal P (2018) Peak energy management using renewable integrated
DC microgrid. IEEE Trans Smart Grid 9(5):4906–4917
10. Wishart MT, Turner J, Perera LB, Ghosh A, Ledwich G (2011) A novel load transfer scheme
for peak load management in rural areas. IEEE Trans Power Deliv 26(2):1203–1211
11. Dudhani S, Sinha AK, Inamdar SS (2006) Renewable energy sources for peak load demand
management in India. Int J Electr Power Energ Syst 28(6):396–400
12. Chapaloglou S et al (2019) Smart energy management algorithm for load smoothing and peak
shaving based on load forecasting of an island’s power system. Appl Energ 238:627–642
13. Huang H, Cai Y, Xu H, Yu H (2017) A multiagent minority-game-based demand-response
management of smart buildings toward peak load reduction. IEEE Trans Comput Des Integr
Circ Syst 36(4):573–585
14. Azcárate C, Blanco R, Mallor F, Garde R, Aguado M (2012) Peaking strategies for the
management of wind-H 2 energy systems. Renew Energ 47:103–111
15. Perez KX, Baldea M, Edgar TF (2016) Integrated HVAC management and optimal scheduling
of smart appliances for community peak load reduction. Energ Build 123:34–40
16. Ashok S (2006) Peak-load management in steel plants. Appl Energ 83(5):413–424
17. Wang Y, Liu L, Wennersten R, Sun Q (2019) Peak shaving and valley filling potential of energy
management system in high-rise residential building. Energ Procedia 158:6201–6207
18. Jain KK, Kumar NK, Ramesh L, Raju MM (2014) An analysis to save electrical energy in a
residential house. Int J Eng Sci 6(2):59–66
Novel Template Protection Scheme
for Multimodal Data
C. Hari Akhilesh, M. Gayathri, and C. Malathy
Abstract A popular form of technology used for securing data is ‘biometrics’.

Biometrics are stored as templates, which are nothing but a digital reference of
the unique characteristics extracted from the biometric, usually stored in a database
which are used for authorizations and encryption. The concern for security of storage
of templates is mainly due to the fact that the templates may not always be securely
stored, and if a hacker does manage to crack the database, the attacker may find all
the data stored in the template use it pilfer other important data or obtain access to
confidential secrets. This paper suggests a novel scheme where in the fingerprint are
stored safely using multi-modal biometric, namely iris, salting and chunking. It also
proposes an efficient way to detect presentation attacks which using salt.
Keywords Biometric templates · Fingerprint template protection · Multimodal

biometrics · Distributed template storage · Detection of presentation attack using
salting
1 Introduction
To understand about template protection, we need to understand first what the

biometrics are and why are the biometrics needed or have to be protected.
C. Hari Akhilesh (B) · M. Gayathri · C. Malathy

Department of Computer Science and Engineering, SRM Institute of Science and Technology,
SRM Nagar, Kattankulathur, Chengalpattu District, Tamil Nadu 603203, India
M. Gayathri
C. Malathy
https://doi.org/10.1007/978-981-16-4943-1_55
594 C. Hari Akhilesh et al.
1.1 Biometrics
Biometrics are physical attributes or characteristics that help us identify a person

uniquely. The world progressing towards a digital era has made the biometrics turn
digital as well. There are multiple parameters that have to be taken in for consid-
ering a characteristic as a biometric. A few of them that are generally considered
are universality which means majority of the people should have that particular trait
or characteristic, another parameter is uniqueness which means that the character-
istic should be vary from person to person, and a more common parameter used is
permanence meaning the time before which the characteristic may change. An ideal
biometric would one which is universal, unique and has less permanence.
These biometrics cannot be directly stored in the digital mode and hence are
generally stored in the form of a template. These templates are generated from char-
acteristics that are unique to that biometric, and there are many methods to generate
the templates for different biometrics.
1.2 Need for Biometrics
Biometrics have begun to replace traditional passwords and pins as an authentication

method; the main reason for this is because these types of authentication methods
have become easy to break over the period of time due to increase in computation
power. The shortcomings of traditional authentication methods are overcome by
biometrics as it is very hard to crack or imitate a biometric.
Apart from authentication, biometrics are used to encrypt data and other day-to-
day activities such as logging attendance of staff and students and solving crimes.
This is mainly because a biometric cannot be forged and the person can be traced
down easily using the biometric. All these have led to a huge prominence in the usage
of biometrics in day-to-day life in the digital world.
1.3 Importance of Protection of Biometrics
Biometrics now being a key part of day-to-day routines is now being targeted by
hackers who want the same for their personal benefits. The exposure can lead to
data leaks by decrypting data that had been encrypted by a biometric or can lead to
digital impersonation and hence access secure data. During impersonation, one can
even defame the person who’s biometric has been leaked, leading to trauma or false
accusation.
When a person is able to get access to the biometric template, through imperson-
ation, there can be a lot of other dependent services which are breached. There exist
services that are linked using a biometric ID such as banks and other government
Novel Template Protection Scheme for Multimodal Data 595
sectors, and if such an ID is accessed by a hacker with the help of the biometric
template of the person, then it will be giving the hacker almost all details of the
person right from his phone number to his physical address.
With all this being said, the need of protecting the biometric templates has
increased and has provided a huge scope of research. Researchers have been trying to
find and implement better biometric template protection schemes and hence fortify
our privacy and security.
2 Proposed System
The proposed system aims to protect the biometric against multiple attacks using
other biometrics. It also aims to detect presentation attack when the template is
stolen and fed to the biometric system in a faster way. This detection is achieved
by detection of salt. The enrolment steps of the proposed system in a brief manner
are combining of biometrics, addition of salt, chunking, encryption and then random
distributed storage. The main point to note here is that the main biometric is used for
matching (first given one); the rest are used to protect this biometric. The verification
process is of a similar manner where in the given biometrics are combining of given
biometrics, fetching and rebuilding template from database, checking of salt in given
template, then sending the given main template and defused template to matcher. The
process is demonstrated using the physiological biometrics, namely fingerprints and
iris. The processes of enrolment and verification are as shown in the Figs. 1 and 2,
after which the system is elaborated based on its modules in the following sections.
2.1 Enhancement, Combining and Decombining
In this module, the system takes care of enhancing the received iris and fingerprints
and then combines them or recombines them as required. This is done using the XOR
method as this enables removal of the biometrics later when the same biometrics
are given. For combining of the iris and fingerprint, we first normalize the iris and
enhance the fingerprint and then we fuse them [1]. Enhancement of the fingerprint
is done using a loosely followed method as proposed in the paper [2] wherein the
fingerprint is enhanced using oriented Gabor filters and made into binary images.
After enhancement, the fingerprint is skeletonized to enable combination. The iris
is enhanced using canny edge detection followed by Hough circular transform after
which we normalize it to a rectangular block with the help of 1D Gabor filters and
phase quantization as suggested in [3] following which again canny edge detection is
taken place to obtain the iris edges. Before combining, data can be resized to match
a standard format to enable ease of combining. For decombining, the same process
of enhancement and resizing is done after which the main biometric is skipped and
Fig. 1 Enrolment activity diagram
rest of the given biometrics are xored with the stored template to obtain the main
biometric.
2.2 Salt
In the salt module, three major functions are achieved, namely salting, desalting and
salt check. While the first two do what they imply, the third salt check function is
for checking of salt existence in the given template for verification. To generate salt,
we use password in conjunction with blake2s (which is a secure, lightweight and
fast hash function) [4], wherein the key for blake2 is hardcoded in the machine and
can only be accessed by direct access alone. Now, we convert this hash to a binary
format which in turn is coded into a different number to avoid collision with image.
Then, it is added or removed from template as required. For checking of salt, we first
compare the given template directly with the assembled template from database. In
the case it is the same the main, admin is notified for breach and the user is denied
access. If it turns out to be negative, then we create a temporary unsalted template
from the assembled template which is used to subtract data from the given template.
Fig. 2 Verification activity diagram
Now, we check the modified given template for remnants of the salt. If remnants
exists, then we again notify the main admin and deny access.
2.3 Chunking, Encryption and Decryption, Assembly
This module takes care of chunking, assembling, encryption and decryption of the
template. The chunking is done using array splitting as the images are nothing but
data. For encryption, we obtain the password, which in turn is used to generate a
key. Key generation is done using scrypt which is a strong key generator which may
prove to be memory intensive for attackers [5]. For key generation, we generate a
salt to be used using the SHA256 algorithm which is a strong hash function used to
generate 256 bits of hash [6]. The key given to SHA256 is nothing but the password
concatenated with a hardcoded string which is ideally different from the one used
for salting module of template. Thus, making the salt a very strong one and hence
making the key to be stronger. Now this generated key from scrypt algorithm is
used in conjunction with AES. AES is chosen because it is a strong symmetric
cipher compared to other symmetric cipher algorithms [7]. Now, the chunks and the
required meta data such as the shape of the chunk, chunk position are encrypted kept
ready for uploading. For decryption and assembly, the process of key generation
is repeated, and then, the chunks and meta data are decrypted using the key and
assembled.
2.4 Database
The database module will handle the part where the chunks are randomly uploaded
to different databases across the world, and this is done to ensure that all chunks
are not available in the same database, which might enable the attacker to actually
reassemble the salted template. Though the data can be stored in relational database,
it is recommended to put in non-relational database (NoSQL) as it is optimized for
key value storage and has better query times [8, 9]. This comes in handy, especially
when there are multiple data to be stored, which will be the case due to chunking.
For retrieval and storage, a username will be used, so therefore ensuring that the
password is not stored in any form in any database.
3 Evaluation
3.1 Template Protection Scheme Analysis
A good template protection is one said to be one with good revocability, performance,
security, unlinkability and diversity.
Revocability of the proposed system is achieved because of salting, encryption
and combination of different biometrics, and the stored template chunks can be
modified by simply modifying the password; in case that feels to be inadequate, a
different combination of biometrics can be used to secure the template, hence making
it different from the previous template. Diversity is attained as the combination of
multiple biometrics along with salting, and in turn, encryption with AES will result
in a different chunk of template even if biometrics given are same. Security is also
achieved since visually the attacker will not be able to figure out the chunks due to
encryption and their randomized upload order.
Performance of the template matching scheme actually depends on the matcher
since the proposed system brings back the original main biometric and gives it to
the matcher along with the given main biometric, so therefore no major performance
hinderance is given by the proposed system. The time consumed for processes apart
from enhancement is almost constant as shown in Fig. 3 (Table 1). Thus, we can say
that the performance is unhindered in the proposed template protection scheme.
Fig. 3 Comparison of time spans apart from enhancement
Table 1 Time spans apart from enhancement

S. No. Time for combining, salting, chunking and Decryption and assembly time (s)
encryption (s)
1 0.171608925 0.329058409
2 0.152110367 0.321273565
3 0.162323952 0.325931311
4 0.149440494 0.344387054
5 0.171329737 0.345149041
6 0.160823584 0.329431295
7 0.150390158 0.324714899
8 0.148513556 0.332785131
9 0.171221849 0.333497763
10 0.169684069 0.341736059
11 0.172198019 0.330428123
12 0.171744553 0.337514412
13 0.164268494 0.332345486
14 0.156521332 0.337944508
15 0.142908838 0.339892511
16 0.156084385 0.341259861
17 0.153265476 0.352910995
18 0.162986685 0.348599911
19 0.157442571 0.333347082
20 0.162786774 0.357174873
Also, due to the combination of multiple fingerprints and Iris, the proposed system
manages to achieve unlinkability due to its complex combination possibility. There-
fore, the proposed system gives a good balance of revocability, performance, security,
unlinkability and diversity which are desired for a template protection scheme.
3.2 Salt Detection Analysis
The proposed system to detect salt and use against presentation attacks was trailed
using custom-made template samples which resulted with the FAR and FRR as 0%
and 1.6%, respectively, which is better than the results obtained by the proposed
system in [1] even though it is a custom matching module versus an attack detection
module, the FAR being better can reduce unnecessary time used up.
4 Conclusion
The proposed system uses a combination of salting, encryption and chunking using
various modules to secure the template while ensuring revocability, performance,
security, unlinkability and diversity apart from which a presentation attack either
direct or using tampered templates is also detected with ease using the proposed
salt detection method, hence ensuring further safety. Since this acts as a middle man
between feature obtaining and matching, it can be used by a system for its betterment.
The proposed system can also be extended to other biometrics such as finger knuckle
print and finger vein. The system can be made better by using better enhancement
techniques and using other better salt generation or key generation functions that are
created in the future. Future works can be targeted for the same.
References
1. Conti V, Militello C, Sorbello F, Vitabile S (2010) A Frequency-based approach for features

fusion in fingerprint and iris multimodal biometric identification systems. IEEE Trans Syst Man
Cybern Part C (Applications and Reviews) 40(4):384–395. https://doi.org/10.1109/TSMCC.
2010.2045374
2. Hong L, Wan Y, Jain A (1998) Fingerprint image enhancement: algorithm and performance
evaluation. IEEE Trans Pattern Anal Mach Intell 20(8):777–789. https://doi.org/10.1109/34.
709565
3. Masek L (2003) Recognition of human iris patterns for biometric identification. Doctoral
dissertation, Master’s thesis, University of Western Australia
4. Aumasson JP, Neves S, Wilcox-O’Hearn Z, Winnerlein C (2013) BLAKE2: simpler, smaller, fast
as MD5. In: Jacobson M, Locasto M, Mohassel P, Safavi-Naini R (eds) Applied cryptography
and network security. ACNS 2013. Lecture notes in computer science, vol 7954. Springer, Berlin,
Heidelberg. https://doi.org/10.1007/978-3-642-38980-1_8
5. Alwen J, Chen B, Pietrzak K, Reyzin L, Tessaro S (2017) Scrypt is maximally memory-hard. In:
Coron JS, Nielsen J (eds) Advances in cryptology—EUROCRYPT 2017. EUROCRYPT 2017.
Lecture notes in computer science, vol 10212. Springer, Cham. https://doi.org/10.1007/978-3-
319-56617-7_2
6. Gilbert H, Handschuh H (2004) Security analysis of SHA-256 and sisters. In: Matsui M,
Zuccherato RJ (eds) Selected areas in cryptography. SAC 2003. Lecture notes in computer
science, vol 3006. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-24654-1_13
7. Advani N, Rathod C, Gonsai AM (2019) Comparative study of various cryptographic algorithms
used for text, image, and video. In: Rathore V, Worring M, Mishra D, Joshi A, Maheshwari S
(eds) Emerging trends in expert applications and security. Advances in intelligent systems and
computing, vol 841. Springer, Singapore. https://doi.org/10.1007/978-981-13-2285-3_46
8. Li Y, Manoharan S (2013) A performance comparison of SQL and NoSQL databases. In: 2013
IEEE pacific rim conference on communications, computers and signal processing (PACRIM),
Victoria, BC, Canada, pp 15–19. https://doi.org/10.1109/PACRIM.2013.6625441
9. Čerešňák R, Kvet M (2019) Comparison of query performance in relational a non-relation
database. Transp Res Procedia 40:170–177. ISSN 2352-1465. https://doi.org/10.1016/j.trpro.
2019.07.027
Performance Optimization of Stepper
Motor Using MATLAB
R. Gopalakrishnan, A. Dominic Savio, K. Dhayalini,

and Nithya Rani Navaneethan
Abstract The importance of special machines has become an utmost necessity in

the current scenario. The implementation of the different types of special machines
in various types of applications has become quite challenging, keeping their mechan-
ical and electrical characteristics in mind. In this paper, we have proposed the
hybrid stepper motor performance analysis. The performance of the given motor is
controlled for user-defined value, and errors are reduced to make the system more reli-
able. The performance stepper motor analyzed in MATLAB and the corresponding
verified through hardware implementation.
Keywords Stepper · MATLAB · Special machines
1 Introduction
The machines have outgrown in numbers more than humans, and the purpose of
the machines is so vast that their growth is nonstoppable nor the invention of new
techniques, machines for each purpose going to be very normal. The applications
of special machines like, switched reluctance, stepper motor, brushless DC have
become so common and more advisable these days, because of their competency.
For various stepping applications like angled movement, precise rotation, and other
angular rotation applications, Le and Jeon [1] explain about the low-speed control of
stepper motors with advanced FPGA technology, whereas the [2] paper describes the
position control of a sensorless stepper motor. The more the electronic component in
the circuitry, the complications and other disturbance in the performances increase.
R. Gopalakrishnan (B) · K. Dhayalini

K. Ramakrishnan College of Engineering, Tiruchirappalli, India
A. Dominic Savio
School of Electrical Engineering, SRM Institute of Science and Technology (formerly known as
SRM University), Kattankulathur Campus, Chennai 603203, India
N. R. Navaneethan
Sri Sairam Engineering College, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_56
604 R. Gopalakrishnan et al.
The sensorless control of machines gives the open loop direct control of any motor.
Masi et al. [3] give the sensorless stepper motor drive with extended Kalman filter;
[4] studies the overview of the permanent magnet stepper motor for microstepping
using a disturbance observer. The controlling of a motor for any nonlinear module
is going to be very difficult; [5] explain the microstepping with nonlinear torque
modulation for a permanent magnet stepper motor. Unlike other motors, the stepper
motor construction is different with the input which is phased sequence, [6] where
the phased-compensated microstepping for permanent magnet stepper motor is vali-
dated. Butcher et al. [7] it is stated the use of hybrid machines in intelligence drives,
where we need go in for control technique algorithms for intelligence. The special
machines have also become an important component for an automatic test bench
system; [8] gives the complete characterization of vibration energy harvesters in an
automated system. Wilcox and Devasia [9] describe the detailed study of stability
of velocity control of piezoelectric stepper machines. Wei et al. [10] where the new
design technique of 3-D printing technology using stepper motors is discussed, nowa-
days the printing system has become more advanced with the help of robotic guiding
technology which is only possible cause of the new machines, thanks to stepper
motors. Shin et al. [11] validate the nonlinear control interconnected system. Druti
ranjan in [12] proposes the real-time precision control tracking with stepper motor;
here it is controlled with microstepping technique. Lim and Yan [13] work on high-
precision motion control for industrial microscopes. Arias [14] proposes the tech-
nique for reducing the cogging torque in hybrid stepper machines by using resonant
controllers. Tran [15] validates the adaptive current controller for the stepper motor
which uses a neural network and double compensators. Although the controlling
for speed is quite easy, the rotating angle estimation in hybrid motors is going to
be a very difficult task; Wang et al. [16] propose the angle estimation for hybrid
stepper motor with fault diagnosis. Also, some of the hybrid motors have the issues
of vibration due to their high speed; Pillans [17] proposes a method of reducing
positional errors by going in for vibration optimization of the drive. Kim [18] gives
a nonlinear gain position control using positional feedback. The control of feedback
gain plays an important role in any linear or nonlinear control. Blauch [19] demon-
strates the high-speed parameter estimation of stepper motor. In our proposed work,
the difficulty in motion control and other stepping errors are neglected. The control
technique is easily implemented using the drivers. Both MOSFET and IGBT func-
tions are analyzed. The IGBT control is comparatively faster than precise than the
MOSFET, although the current rating makes the IGBT tougher for implementation.
2 Working of Hybrid Motor
The motor is excited by a combination of winding and by permanent magnets in it.

The pole in the motor has two excitations. The number of poles or rotor teeth gives
the length of each step, which is calculated by
Performance Optimization of Stepper Motor Using MATLAB 605
90
steep length = (1)
P
The hybrid stepper motor has small steep typically 1.8 when compared to reluc-
tance motor. This decides the advantage over the other motors, where the hybrid
motor requires limited operating space. Small signal modeling is used to reduce the
inductance in the magnetic coupling,
di a Ua + K m ω sin(Nθ ) − Ri a
= (2)
dt L
di b Ub + K m ω sin(Nθ ) − Ri b
= (3)
dt L
dw K m i b cos(Nθ ) − Ti − K m i a sin(Nθ ) − K v ω
= (4)
dt J
dθ
=ω (5)
dt
ib is the current phase B and ia is the current phase A; correspondingly, the U b V,

U a V is the voltage in phase in A and phase B.
The motor phases are fed by two H-MOSFET PWM converters connected to a 12 V
DC voltage source. The motor phase currents are independently controlled by two
MOSFET drive signals by comparing the measured currents with their references.
The initial input signal for driving the phases is given by the step frequency, which
is given as gate pulse. Also, the movement of the hybrid stepper motor drive is
controlled by the step signal from by user-defined value.
The output of the simulation results is obtained for 0.25 s operation of the stepper
motor drive during which the stepper rotates during 0.1 s in the positive direction,
stops for 0.05 s and rotates in the reverse direction for 0.05 s and stops.
Figure 1 represents the Simulink block, where the load torque is purposely reduced
to study the triggering output of the waveform and as a result, the output by the stepper
motor is null with respect to any given input.
During zero torque condition, all the result parameters are measured as zero.
When the torque is input, the torque of the system is increased to 1.8. The hybrid
motor achieves some critical output without any controlled response. The response
of the system, where the voltage waveform, is initially high; the value gets settled to
threshold value after 2.5 s is shown in Fig. 2.
In the Simulink, the drive switching has been triggered from both IGBT and
MOSFET, which gives the output to a controlled value projected. Figure 3 represents
Fig. 1 Simulink module of hybrid stepper motor with zero torque

VPH (V)
Time(ms)
(a)
IPH (A)
Time(ms)
(b)
Te(N*m)
Time(ms)
(c)
w (Rad/s)
Time (ms)
(d)
Theta(degrees)
Time (ms)
(e)
Fig. 2 Different output at the torque value of 1.8 N/M with MOSFET drive
VPH (V)
Time(ms)
(a)
IPH (A)
Time(ms)
(b)
Te(N*m)
Time(ms)
(c)
w (Rad/s)
Time (ms)
(d)
Theta(degrees)
Time(ms)
(e)
Fig. 3 Different output at the torque value of 1.8 N/M with IGBT drive
the value triggered from an IGBT, the second waveform is much more precise than
the former one, which is because of the range of the IGBT. In the above Simulink, the
values for voltage, current, torque, speed, operating drive temperature are provided
and been analyzed. The value of the waveform is shown for 2 s. Due to the initial
torque produced, we could see the starting response which is much distorted and the
response gets settled after the time period 0.5 s.
The observations are made from Fig. 4 response value for both current and voltage
which is the user-defined input value is gets settled. The value torque is at 1.8, still
needed to fine-tune. The value for given rpm is not settling proper value; also, the
temperature value is more deviating when compared to the value obtained from
VPH (V)
Time(ms)
(a)
IPH (A)
Time(ms)
(b)
Te (N*m)
Time(ms)
(c)
w (Rad/s)
Time (ms)
(d)
Theta(degrees)
Time(ms)
(e)
Fig. 4 Different output at tuned condition with 1.8 torque with IGBT drive
MOSFET. Therefore, the shows more precise value, so that the finer value can be
obtained.
The values of voltage, current, torque, speed and the system temperature are
analyzed and produced at a precision value with use of higher tuning method
employed.
In the above Simulink module, the input to the IGBT driver circuit is given after
tuning using a comparator, necessary filter networks and gain, which resulted in the
output waveform value more tuned than the previous. The current and voltage values
are more accurate 12 V and 1.1 A, respectively. The speed and the torque values show
significant steady-state response and the value more accurate to the given input. The
temperature value while using IGBT drives is significantly very low compared to the
higher value, when the tuning is not employed.
4 Hardware Implementation
The experimental validation of the stepper motor control is done using laboratory
prototype shown in Fig. 5. The input to the motor is initially converted from an AC to
DC, and with help of switching mode, the current rating to the motor is maintained.
The proposed motor is 12 V, DC, 1.1 A and 13 Watts (Fig. 5).
Hybrid stepper motor
Power
supply
Converter design
with IGBT , MOSFET
Fig. 5 Experimental setup

VIGBT(V) IMOSFET (A) VMOSFET (V)
V = 12.3V 10 V/Div
I = 1.2A 1A/Div
V = 11.9 V 10V/Div
I = 2A 2A/Div
IIGBT (A)
Time (ms)
Fig. 6 Experimental result

The experimental results validate the output obtained from the Simulation results.
Both the values are analyzed for performance improvement. Initially, the input value
given to the converter, both IGBT and MOSFET which, due to the integral design
improves the output value. From the analyzed hardware results, it is suggested that
the result from IGBT has better faster triggering than MOSFET, but the MOSFET
result shows better performance in sequential time period. So, IGBTs can be used
for high current ratings, whereas MOSFETs can be used for low speed and better
performance optimization.
5 Conclusion
This paper demonstrates the working of stepper motor with converter control tech-
nique. Initially, the working and optimization of the motor are simulated using the
MATLAB/Simulink software. The performance of the simulated machine improves
with the changing the converter technique. A hardware prototype module is devel-
oped, and the performance of the motor is improved considerably, with respect to the
simulated results. The hybrid stepper motor performance characteristics are deter-
mined for various conditions. The different motion directions are also examined with
different torque values.
References
1. Le QN, Jeon J-W (2010) Neural-network-based low-speed-damping controller for stepper

motor with an FPGA. IEEE Trans Ind Electron 57(9)
2. Bendjedia M, Ait-Amirat Y, Walther B, Berthon A (2012) Position control of a sensorless
stepper motor. IEEE Trans Power Electron 27(2)
3. Masi A, Butcher M, Martino M, Picatoste R (2012) An application of the extended Kalman
filter for a sensorless stepper motor drive working with long cables. IEEE Trans Ind Electron
59(11)
4. Kim W, Shin D, Chung CC (2012) Microstepping using a disturbance observer and a variable
structure controller for permanent-magnet stepper motors. IEEE Trans Ind Electron 60(7)
5. Kim W, Shin D, Chung CC (2013) Microstepping with nonlinear torque modulation for
permanent magnet stepper motors. IEEE Trans Control Syst Technol 21(5)
6. Shin D, Kim W, Lee Y, Chung CC (2013) Phase-compensated microstepping for permanent-
magnet stepper motors. IEEE Trans Ind Electron 60(12)
7. Butcher M, Masi A, Picatoste R, Giustiniani A (2014) Hybrid stepper motor electrical model
extensions for use in intelligent drives. IEEE Trans Ind Electron 61(2)
8. Ruan JJ, Lockhart RA, Janphuang P, Quintero AV, Briand D, de Rooij N (2013) An automatic
test bench for complete characterization of vibration-energy harvesters. IEEE Trans Instrum
Meas 62(11)
9. Wilcox S, Devasia S (2015) Stability of velocity control for a piezoelectric stepper. IEEE/ASME
Trans Mechatron 20(2)
10. Wei Y, Chen Y, Yang Y, Li Y (2018) Novel design and 3-D printing of nonassembly controllable
pneumatic robots. IEEE/ASME Trans Mechatron 21(2)
11. Shin D, Kim W, Lee Y, Chung CC (2017) Nonlinear control with state-dependant reset integrator
for a class of singularly perturbed interconnected nonlinear systems. IEEE Trans Control Syst
Technol 25(4)
12. Gaan DR, Kumar M, Sudhakar S (2018) Real-time precise position tracking with stepper motor
using frequency modulation based microstepping. IEEE Trans Ind Appl 54(1)
13. Lim LHI, Yang D (2019) High-precision XY stage motion control of industrial microscope.
IEEE Trans Ind Electron 66(3)
14. Arias A, Caum J, Ibarra E, Grino R (2019) Reducing the cogging torque effects in hybrid
stepper machines by means of resonant controllers. IEEE Trans Ind Electron 66(4)
15. Tran HN, Le KM, Jeon JW (2019) Adaptive current controller based on neural network and
double phase compensator for a stepper motor. IEEE Trans Power Electron 34(8)
16. Wang X, Lu S, Zhang S (2020) Rotating angle estimation for hybrid stepper motors with
application to bearing fault diagnosis. IEEE Trans Instrum Meas 69(8)
17. Pillans J (2021) Reducing position errors by vibration optimization of stepper motor drive
waveforms. IEEE Trans Ind Electron 68(6)
18. Kim W, Lee Y, Shin D, Chung C (2021) Nonlinear gain position control using only position
feedback for permanent magnet stepper motors. IEEE Trans Power Electron 36(7)
19. Blauch AJ, Bodson M, Chiasson J (1993) High-speed parameter estimation of stepper motors.
IEEE Trans Control Syst Technol 1(4)
Performance and Analysis of Voltage
Compensation in Transmission Line
Using SMES-Based IDVR
M. Jagadeesh Kumar, T. Muthamizhan, P. Rathnavel, G. Ezhilarasan,

and T. Eswaran
Abstract Rise in the power demand and existence of loss in transmission sector
is increasing linearly. This is due to the increase in use of nonlinear load which
causes voltage fluctuation in transmission lines. Some of the problems in it are sag,
swell, voltage spikes, harmonics, open and short circuit faults. Among the above
mentioned power quality issues, sag and swell will be considered as major fault, and
it can be resolved and overcome by including the interline dynamic voltage restorer
(IDVR). IDVR has enough capability to perform voltage compensation in an appro-
priate rate, and it maintains stable energy across DC link capacitor. The proposed
method can have several DVRs connected to various distribution feeders in power
system sharing common energy storage. A superconducting magnetic energy storage
system (SMES) is selected to feed DVR, whenever DVR undergoes compensation.
Further, fuzzy logic control also took part in it. The Simulink results convey the
effectiveness and performance of proposed method.
Keywords Interline dynamic voltage restorer (IDVR) · Sag/swell · DC link

capacitor · Fuzzy logic control
M. Jagadeesh Kumar (B) · T. Muthamizhan · P. Rathnavel · G. Ezhilarasan · T. Eswaran

Department of Electrical and Electronics Engineering, Sri Sairam Institute of Technology,
Chennai 44, India
T. Muthamizhan
P. Rathnavel
G. Ezhilarasan
T. Eswaran
https://doi.org/10.1007/978-981-16-4943-1_57
614 M. Jagadeesh Kumar et al.
1 Introduction
Variation in load level will cause failures in electrical devices due to power quality
disturbances. Power quality took major part in functioning entire system at desired
rate. Most of the distribution systems encounter sag, swell, switching transient,
harmonics, etc. The sudden fall in voltage level is said to be as voltage sag, and
rise in voltage at a particular portion is described as voltage swell. The existing
voltage sag compensation techniques are in-phase, pre-sag and optimized energy.
By comparing amplitude of positive sequence component with its reference value,
pre-sag method of voltage compensation is studied. Another method [3], implies
reduction in injecting voltage of DVR and mitigate nearer to load. To resolve these, lot
of power conditioning devices is available. Among that, DVR is preferred to provide
effective solution against sag and swell. DVR is said to be as power electronic-based
device which is connected in series with feeder of sensitive load. The structure of DVR
is comprised of voltage source converter (VSC), an energy storage element, passive
filter, injecting transformer and a control topology to deliver controllable voltage
in distribution sector such as amplitude and phase angle. DVR is used mostly for
low-level and medium-level voltage applications. Simpler structure and fast dynamic
response will be the key factors for choosing it. Other than that a short-term energy
storage system is mandatory to perform restoration; for that, SMES is chosen. It is
responsible to protect consumer from voltage fluctuation. This combination plays a
vital role in compensating sag/swell and also concentrates upon power flow control
too [1–3]. The concentration upon selection of control topology is unavoidable. In
addition, ANFIS controller (Adaptive Neuro Fuzzy Inference System) is proposed
to stabilize the functioning of DVR by maximizing the system efficiency. Fuzzy
and PI controller analysis predicts the rate of dynamic response in accordance with
voltage compensation. Replacement of PI control is carried out through fuzzy logic
control in which predecessor method has improper tuning of KP and KI. It leads to
rise in settling time and affects speed of response. Due to this, fuzzy logic control is
progressed in [4, 5]. To bring this into next stage, IDVR is proposed. It is a combina-
tion of several DVR connected in it to split the part of work for each one (i.e., one for
performing sag/swell and another one for power conditioning). The SPWM technique
in turn makes harmonic reduction by regulating switching action in an appropriate
rate [6–10]. The main drawback in DVR is that it receives abundant quantity of
power from DC link capacitor. Even it has certain limitations, this FACTS device is
unavoidable due to its plenty of advantages, by combining more devices to overcome
the limitations of the IDVR. It is evident that IDVR is considerably a perfect solution
to raise the dynamic response and performance. With proper selection of control
technique, the quality of real and reactive power injection is considerably at a steady
state. In Sect. 2, a detailed study upon modelling of DVR is presented. In contin-
uation with compensation method, energy storage system is represented in Sect. 3.
Section 4 dealt with structure and operating principles of IDVR. The modelling of
IDVR and occurrence of voltage sag/swell is represented in Sect. 5. Further, the level
Performance and Analysis of Voltage … 615
of compensation achieved through proposed topology is defined in both Sects. 5 and

6.
2 Structure of DVR
The structure of DVR is depicted in Fig. 1. It is more efficient and effective custom
power device. DVR is solid state device that supplies voltage which in order to regu-
late line voltage. Other than sag/well compensation, it focuses on harmonics compen-
sation, fault current reduction and reduction in voltage transient, respectively. The
role of injecting transformer is to limit noise and transient from primary winding to
secondary. And it connects distribution network with DVR; also, it couples injected
compensating voltage with supply voltage. Whenever the isolation of feeder and
VSC is needed, it performs that too. If any kind of harmonics is developed by VSC,
harmonic filter took care upon it and eliminates as much as possible. VSC is a collec-
tion of power semiconducting devices which produce sinusoidal voltage at desired
rate. Further, energy required for whole process is delivered through electromagnetic
energy storage element. The preliminary tasks need to perform by it are to energize
the entire section to perform sag compensation and after that maintain a constant
voltage at DC link.
Fig. 1 Structure of IDVR

3 Principle of IDVR
The structure of DVR is shown in Fig. 1 comprising voltage source converters config-
ured at front and back with link capacitor. From this structure, it is evident that IDVR
is capable to transfer active power in-between feeder and another one. A common DC
link capacitor connected between both would share power whenever DVRs needed.
If one among DVR presented in IDVR continues to absorb active power from
DC link capacitor, then another tends to be operating under rectification mode, and
it would supply energy towards DC link to maintain a constant voltage across it.
This compact structure concentrates upon deducing the components rating. Due
to this strategy, reduction in size and cost has been progressed without affecting
the performance. Without bypassing normal condition of DVR, the DVR present in
proposed method has potential to raise the displacement factor of an individual feeder.
The functioning of DVR is said to be exchanging active and reactive power between
individual feeders. It utilizes a shunt reactance parallel with load to minimize load
power factor during sag condition. When load power factor is small, much deeper
voltage sags can also compensated. At higher power factor, the performance of DVR
should lag. To resolve these factors, a proposed method adopts some changes in it.
It would make DVR to operate at medium power.
4 Superconducting Magnetic Energy Storage System
The superconducting magnetic material stores energy in magnetic field, and it

possesses quick charging and discharging time which makes it to supply load when-
ever the system needs to perform compensation. SMES parts are categorized into
three, and they are superconducting coil, power conditioning system and cryogenic
cooled refrigerator. Once a coil is charged, the current should not decay, and it
retains its magnetic field identically. Power conditioning system does rectification or
inverting voltage as it totally depends upon the voltage it may be taken. In compar-
ison with alternative energy storage system, this method lost electrical energy level
which is least in nature (Figs. 2 and 3).
The equation represents energy storage in a superconducting coil, E = 1/2 LI *
I. Most of the industrial sectors face interrupts due to voltage sag. At that condition,
improving performance of FACTS controller requires higher performance storage
device. From above concern, SMES is said to be a better choice even the system may
fall towards sudden interruption consistently.
Fig. 2 Layout diagram of a Superconducting Magnetic Energy Storage (SMES) System
Fig. 3 Schematic
diagram of SMES-based
DVR
The proposed system is designed for 3-phase, 415 V, 50 Hz supply simulated using
MATLAB/SIMULINK R2014a software. To examine the functioning of IDVR,
a fault is applied at resistance of 0.66, and after that, DVVR is connected, and
compensation is visualized in below waveform (Figs. 4, 5 and 7, Table 1).
By implementing fuzzy logic control, the generated gate pulse for switches present
in VSC is represented in Fig. 6. Normally, a system faces critical voltage fluctuation
Fig. 4 MATLAB/SIMULINK Model of the proposed method
Fig. 5 Combination of conventional battery and SMES-based DVR

Table. 1 Fuzzy rule base (7 x 7 Matrix) for the IDVR control

CE
E NB NM NS ZE PS PM PD
NB NB NB NB NB NM NS ZE
NM NB NB NB NM NS ZE PS
NS NB NB NM NS ZE PS PM
ZE NB NM NS ZE PS PM PB
PS NM NS ZE PS PM PB PB
PM NS ZE PS PM PB PB PB
PB ZE PS PM PB PB PB PB
Fig. 6 Gate signal generation for the switches of IDVR
under severe circumstances. It may be voltage sag, swell, oscillation, etc. In trans-
mission, the line struggles with sag and swell. There are various topologies available
to it. In upcoming waveforms, how much amount of sag and swell occurred and what
are processes undertaken to eliminate it are visualized graphically.
A sudden reduction in voltage flow is said to be voltage sag, and its occurrence is
considerably 10% to 90% in manner. In Fig. 8, the level of voltage of sag occurred is
represented. It should be 12.4% in below figure. The disturbance in real and reactive
power occurred during time range of 0.5–0.25 s. (Figs. 10 and 11).
FFT analysis revealing THD value is high, and it becomes 4.81%. It makes the
entire system inefficient. For swell, the value of THD is almost high. From this, it is
evident that this range of voltage feed to distribution system may affect the consumers
load requirement. The waveform is shown in Fig. 12 witnessing compensated voltage
by IDVR with fuzzy controller. From Figs. 9 and 13, the THD of the three phase
voltage is found as 4.81% for the voltage sag, 0.12% for the voltage swell and the
THD for the total three phase system is found as 0.08%. From the above results the
Fig. 7 Rules and membership function presented in fuzzy logic control
Fig. 8 Occurrence of voltage sag in 3-phase transmission system
compensation of the controller aids the systems to get good results with minimum
percentage of THD.
6 Conclusion
The modelling and functioning of IDVR to compensate voltage sag/swell are exper-
imentally verified, and its efficiency is visualized through waveforms. The model of
IDVR for 13.08% of voltage swell and 12.4% of voltage sag is compensated in addi-
tion to fuzzy logic control method. The results examine that the proposed method
Fig. 9 FFT analysis of voltage sag
Fig. 10 Occurrence of voltage swell in a 3-phase transmission system
compensated effectively and at high accuracy. Input sources are continuously moni-
tored, and if any sag/swell is noted, the shunt reactance decreases load power factor
to maximize the performance of IDVR.
Fig. 11 FFT analysis of swell
Fig. 12 Compensated voltage of the 3-phase transmission system using SMES-Based IDVR
Fig. 13 FFT analysis for three-phase system
References
1. Prema R (2016) SMES based DVR against voltage sag using ANFIS system. Int J Innov Sci
Eng Res 3(3):23–32
2. Ravirajan S, Muthamizhan T (2016) Modelling and digital simulation of a standard IEEE—30
bus system to improve power quality of grid with UPQC using MATLAB. Asian J Res Soc Sci
Humanities 6(10):830–845
3. Hossain MI, Rahaman I, Rahman MN, Hasan MF, Hasan MM, Sarker RC (2020) Voltage
sag compensation in distribution system using dynamic voltage restorer. In: 2nd international
conference on advanced information and communication technology 2020, Dhaka, Bangladesh,
pp 492–497
4. Thaha HS, Prakash TRD (2020) Use of fuzzy controller based DVR for the reduction of power
quality issues in composite micro-grid. In: International conference on renewable energy inte-
gration into smart grids: a multidisciplinary approach to technology modelling and simulation
(ICREISG), Bhubaneswar, India, pp 131–136
5. Naveen V, Bharat Kumar P (2015) Fuzzy and adaptive control strategies of voltage source
converter for correction of power factor. Int Res J Eng Technol 2(6):160–174
6. Karshenas HR, Moradlou M (2011) Design strategy for optimum rating selection in interline
DVR. IEEE Trans Power Deliv 26(1):242–249
7. Pal R, Gupta S (2017) Interline dynamic voltage restoration for power quality compensation.
Electr Comput Eng Int J (ECIJ) 4(2):79–98
8. Xie Q, Chen XY, Feng J, Lei Y (2020) Feasibility analysis of SMES-based interline DC
dynamic voltage restorer. In: IEEE international conference on applied superconductivity and
electromagnetic devices (ASEMD), Tianjin, China, pp 1–2
9. Usha RP, Sudha R, Reddy SR (2011) Voltage sag/swell compensation in an interline Dynamic
Voltage Restorer. In: International conference on emerging trends in electrical and computer
technology, Nagercoil, India, pp 309–314
10. Meenakshi N, Manivel K (2015) Interline DVR with reduced rating control. Int J Eng Res
Technol 3(4):1–6
Multiple Regression-Based Adaptive
Protection Scheme for Microgrids
Aindrila Dey, Moumita Dhang, Rangit Ray, and C. Vaithilingam
Abstract The dynamic behavior of microgrid on the protection point of views such
as modes of operation (grid-connected/islanded) and generation levels of renew-
able energy sources makes the fault current vary randomly. Subsequently, the relay
operating conditions are to be changed based on the above factors so that relay can
differentiate between fault condition and normal operating condition. The conven-
tional protection schemes will be unable to differentiate between modes of operations.
Further, the fault current is dynamic in nature. Hence, electromechanical or static
relays may be replaced with processor-based relays. This paper proposes an adap-
tive protection scheme for the microgrid using machine learning algorithm. Multiple
regression machine learning algorithm is used to develop adaptive protection scheme.
The proposed model will be able to differentiate between normal operating condition
and fault condition and trip the relay if needed. The hardware prototype developed
for the adaptive relaying scheme has been tested successfully.
Keywords Adaptive protection · Microgrids · Machine learning · Multiple

regression
1 Introduction
Distributed generations are gaining much importance in today’s power system as it

solve the perennial prime concerns of the power systems to ensure reliable, stable,
and quality power supply [1]. From [2], it is inferred that, with the advent of DGs,
the electricity producer–consumer gap has been optimally met and helped in mini-
mizing the transmission losses. Also, with the introduction of PV and wind energy
generations, the need for power electronic systems for the reliable supply necessi-
tates the faster operation of protection system [3]. The idea of this paper into this
field has been on the fact that this adaptive protection scheme is a developing topic
and has a huge potential to be in rigorous usage in the future. As the development of
A. Dey · M. Dhang · R. Ray · C. Vaithilingam (B)

Vellore Institute of Technology, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_58
626 A. Dey et al.
such a scheme requires highly efficient technologies and devices, thorough research
has been done on this field of interest [4]. The power system requires major revamp
in protection schemes due to the penetration of renewable energy resources in the
network [5]. From [6], it has been inferred that traditional protection schemes become
inappropriate due to the following reasons:
(1) Dynamic fault current.
(2) Bidirectional power flow.
(3) Randomness associated with renewable energy generation.
(4) Quantum of power and power injection node.
Also, from the literatures, it is observed that new methods for adaptive protection
schemes for Microgrids can be divided into two groups as follows,
(i) Fuzzy logic and ANN-based approaches.
(ii) Machine learning-based methods.
The philosophy of adaptive protection is not new. More than three decades ago, the
results of an investigation into the possibilities of using digital techniques to adapt the
transmission protection to PS changes had been presented in [6]. The growing number
of publications on adaptive protection has been observed. Particularly, large no. of a
rise in the citation on this topic has been observed in 2007 and the highest number
in 2018. One important observation which has been made is that adaptive protection
scheme has been divided into three sub-areas: transmission lines, transformer, and
DSs/MGs. In [7], a rule-based adaptive protection scheme using a machine learning
methodology for microgrids in extensive distribution automation (DA) has been
modeled. In this paper, ANN–SVM-based method is used. The paper [8] presents
a study of microgrid protection scheme based on Hilbert-Huang transform (HHT)
and ML techniques. Three different machine learning models are being tested and
compared in this paper: naïve Bayes classifier, SVM, and extreme learning machine.
The observations and inferences in the aforementioned survey motivated to develop
a model based on the ML technique. An adaptive protection scheme using a multiple
regression model is developed and tested. This paper explains the unique essence
of adaptive protection scheme and also the demerits of the traditional protection
scheme.
2 Design Approach
The microgrids are becoming essential in the energy industry and may become an
inevitable part of the power system network for sustainable energy generation. There-
fore, to use the full potential of the microgrid, various technical challenges have to
be addressed. One of the major challenges is the protection scheme. The two major
factors, the randomness of the renewable energy sources and the different modes of
operation, results in dynamic fault currents. Thus, designing a suitable protection
scheme and coordinating relays would require the understanding of fault current
Multiple Regression-Based Adaptive Protection Scheme … 627
values. Hence, the conventional protection schemes, for example, a conventional

overcurrent protection scheme could not differentiate the fault condition, and normal
operating condition, hence, will face selectivity and sensitivity issues. The main
objectives are to develop an adaptive protection scheme for the microgrid using ML
algorithm and develop a hardware prototype using DC Sources, intelligent electronic
devices, and suitable relaying and switching devices.
A MATLAB Simulink model of microgrid with two solar energy sources as DGs
and provision for grid connection is developed. The simulation consists of two photo-
voltaic panels (PV panels) connected to an AC load. The simulation model consists
of PV panels, three-phase inverters with RLC filters, and then the load. The effect of
fault in microgrid side is chosen in this paper so that the proposed model is validated
by ensuring the fault is detected and the microgrid is protected in both grid-connected
and islanded modes. The notion behind this model has been that the microgrid should
not induce any fault currents into the system, and following a fault in the microgrid,
it should be isolated from the main grid. Also, when the microgrid is in islanded
mode and fault occurs at the load side, the DGs will be isolated from the fault.
2.1 Dataset Preparation
The normal operating current of the system at different operating conditions was
obtained. The generations of DGs (PV sources) are varied by varying the irradiance
values. The fault currents of the system under different operating conditions and fault
types are obtained. The fault currents when the system is operating in grid-connected
mode are also obtained for all operating conditions. The sample data obtained from
the fault analysis are given in Table. 1
The above dataset has been scaled down in the range of (0–5) volt. Also, here, the
current values are given in terms of voltage in the mentioned range as in real-time
setup, all voltage values have been taken from RPSs representing PV1, PV2, current
value, and mode of operation. Also, while importing the dataset into the real-time
system (machine algorithm), only the LG fault value is chosen as the threshold value
among all types of faults because LG fault value is minimum among all the fault
values. If the input current is more than the predicted fault current value under a
given condition, then the system is under fault and vice versa.
2.2 Machine Learning Algorithm
Linear regression is a prime statistical tool that can be used to estimate the values
using the training dataset. Multiple regression is used majorly when there are a one
dependent and more than one independent variables. The general form of the multiple
regression is given by the equation:
628 A. Dey et al.
Table 1 Training dataset samples (pre-scale down)

Irradiance Irradiance Type of Mode operation Fault current Normal
(PV1) (PV2) fault (A) current (A)
100 200 LG Islanded 628 398
LL 1860
LLG 1820
LLL 1375
400 800 LG Islanded 1893 492
LL 6398
LLG 6466
LLL 5479
600 1200 LG Islanded 1433 502
LL 7350
LLG 7500
LLL 7307
100 200 LG Grid-connected 2216 1478
LL 3685
LLG 3589
LLL 3061
Y = B1 × X 1 + B2 × X 2 + . . . Bn × X n + A
Y = dependent variable
X1, X2…Xn = independent variables.
Scatter plot is used in multiple regression to fit in single line. Multi-dimensional

voids of data patterns are used to fit multiple regression. The simplest form of this
algorithm has one dependent and two independent variables. The output or the
value to be estimated is dependent variable. The variables decide that the values
of dependent variable are independent variables.
In this paper, the number of independent variables is two, which are the irradiance
values of PV1 and PV2. With the values of the independent variables, the LG fault
current value is predicted for the unknown irradiance values.
The algorithm has been implemented using the scikit library in Python. The
following steps have been used for the process.
1. After calling all the required libraries, the dataset in CSV extension has been
imported to Python with the help of the pandas library.
2. The following step is to divide the data into “attributes” and “labels." Indepen-
dent variables are attributes, and dependent variables are called as labels. In this
paper, the PV values are the attributes, and the fault current value is the label.
3. Next, the 70% of data were used for training the model, and 30% were used for
testing the model.
4. After splitting the dataset, the algorithm is trained by importing linear regression
class. Linear regression model fundamentally finds the values of slope and
intercept; using the values of slope and intercept, the value of dependent variable
will be predicted.
5. After training the algorithm, some predictions are used to check how accu-
rately the algorithm can predict the data. To make a prediction on the test case,
“regressor.predict” command is used. By this, the actual output values of the
test cases can be compared with the predicted values.
6. Next, we can get the desired fault current value for the unknown irradiance
value.
2.3 Hardware Implementation
A microgrid model with two PV panels has been designed which will function in
grid-connected and islanded modes. In both, the modes LG, LL, LLG, and LLL faults
have been analyzed for different values of irradiances. As LG fault current values
are minimum, they have been chosen as the threshold value to decide whether the
system is in fault condition or normal operating condition. The simulation values
obtained are scaled down to suitable values for the microcontroller to handle. For the
hardware model, the irradiance values are scaled down and normalized to suitable
voltage values. Four voltage inputs are given from RPS representing two PV Sources,
modes of operation, and current value. These voltages are given to the microprocessor
called Raspberry Pi 3B + with the help of an ADC called Arduino UNO. The
real-time inputs received by the Raspberry Pi will be processed by the machine
learning algorithm embedded in Raspberry Pi. The algorithm selects the dataset
to be compared, based on the four inputs. This processor will then compare the
measured current value with the predicted fault current value obtained from the
machine learning algorithm. If the predicted fault current value is more than the
input current, then the system is not under any fault. If it is lesser than input current
value, it reflects a fault condition and the relay will be activated. With the help of
the Raspberry Pi processor, a relay has been tripped to glow an LED and turn on a
buzzer to indicate the relay trip following a fault condition. The clearance time of
the relay is lesser than 15 ms (Fig. 1).
Fig. 1 Block diagram of the

hardware model
630 A. Dey et al.
The developed prototype is tested in the laboratory to validate the effectiveness

of machine learning algorithm and the capability of processors.
• In this prototype, the irradiation levels of the PV panels have been varied to
determine fault current values, to prepare a dataset that is to be used to train the
multiple regression algorithm.
• All irradiation levels and current values are expressed as voltage levels within
(0–5)V range.
• Depending upon the mode of operation and the power generation levels of the
renewable energy sources, which are random/intermittent, the fault current will
vary accordingly.
• The multiple regression algorithm is implemented in the Python platform in Rasp-
berry Pi to predict the fault current value for a specific irradiation level and a given
mode of operation.
• The predicted fault current value is then compared with the reference current
value (depicted as a voltage value) given as input. If this reference current value
is greater, it is a fault condition, else not.
• It has been observed that the relay trips within 15 ms after the occurrence of a
fault.
3.1 Simulation Results
A simulation model of a microgrid has been designed and operated in the modes
(grid-connected or islanded mode) using two PV panels, whose irradiation levels
were varied to notice the change in fault current value.
3.2 Overview of the Prototype
1. Figure 2 shows the four voltage levels which have been used to represent the
four significant inputs in the hardware setup. The four voltage levels represent
the following:
i. PV1—Representing the input from one PV panel. Changing of voltage
here represents the changing generation level of the PV panel.
ii. PV2—Representing the input from the second PV panel. Changing the
voltage value here also represents the changing generation level of the PV
panel.
iii. Mode of Operation: Representing the mode of the microgrid, i.e.,
islanded or grid-connected mode. If this voltage level is higher than “1”,
Fig. 2 Four voltage levels are used as demonstrated above
then the microgrid represents grid-connected mode, else the microgrid is

in islanded mode.
iv. Reference Current Value: This value has been represented by another
voltage value from RPS. All values are in the range of (0–5) V as the
Arduino has a maximum capacity of 5 V.
The analog inputs are given to Arduino, as shown in Fig. 2, and are
converted to digital outputs by Arduino (ADC) and then given to Raspberry
Pi.
2. Figure 3 shows the hardware setup of Raspberry Pi 3B + , relay, LED, and

buzzer. The Raspberry Pi acts as the brain of this setup. Based on the input from
Fig. 3 Hardware setup of Raspberry Pi, LED, and buzzer

632 A. Dey et al.
Fig. 4 Console window of the Geany IDE
Arduino and the machine learning algorithm, it predicts the fault current value.
If the fault current value is greater than reference current value, then it shows
“Fault,” else “NO FAULT.” If the platform predicts fault, then the relay trips
and LED glows and buzzer buzzes simultaneously.
3. Figure 4 shows the console window of the machine learning platform.
The console window in Fig. 4 is the output window of the Geany platform in
Raspberry Pi 3B + . Python language has been used to implement the multiple
regression algorithm by importing libraries like scikit in Python. The status of the
code in this console window shows that the mode of operation is in islanded mode
as the input from the RPS is less than 1 (as per the given condition in code). The
reference current input which is given as voltage input from the RPS is compared,
and in this condition, it is less than the predicted value of the fault current from the
code implemented in Raspberry Pi platform, and hence, therefore, the result in the
console window shows “Not a Fault.”
Raspberry Pi 3B + has been chosen to make hardware prototype as it has the
latest features and advanced technology. The machine learning code has been done
in the Geany platform using Python language in Raspberry Pi.
The Arduino Uno is based on ATmega328P. It consists of 14 and 6 digital and
analog pins, respectively, and a 16 MHz resonator. This module acts as an analog to
digital converter. The analog inputs from the regulated power supplies (RPSs) are
taken by Arduino and then the digital output is given to the Raspberry Pi module as
this module only accepts digital inputs. The values from RPSs and also the training
dataset are in the range of (0–5) Volts as the Arduino Uno has a maximum capacity
of 5 V.
The 5-V one-channel relay module has been used to trip the circuit during the
fault conditions. The relay remains in normally open (NO) position, and when a fault
occurs, it moves to normally open (NO) position, and the fault is cleared within the
maximum fault clearance time. The change of contact of the relay is made possible
with the help of trigger voltage (Vin) which is in the range of (0–1.2) Volts, and it is
powered through Vcc which is 5 V. The relay is powered through Raspberry Pi 3B
+.
The LED glows and buzzer buzzes when the fault condition occurs if the relay
is normally open (NO) mode. The relay circuit closes whenever the tripping takes
place and LED and buzzer get ON simultaneously.
3.3 Summary.
The adaptive relaying scheme is developed using ML algorithm “multiple regres-

sion” which has been implemented in the Geany platform in Raspberry Pi using
scikit library in Python. The dataset finally imported as “CSV” file to the ML plat-
form which consisted of two independent variables PV1 and PV2 and one dependent
variable, i.e., LG fault current value (since LG fault value is the least fault value)
under two different modes of operation, namely: grid-connected and islanded. The
multiple regression algorithm implemented in the Geany platform in Raspberry Pi
would be able to predict the LG fault current value based on the real-time inputs from
RPSs and given dataset. The real-time inputs are given through Arduino, which acts
as analog to digital converter to the Raspberry Pi, which then commands the relay
to trip during fault conditions. The trip condition is indicated when the LED glows
and buzzer buzzes.
4 Conclusion
The protection scheme for microgrid developed in this paper is adaptive in nature.
A simulation model with two PV sources as DGs has been developed to get the
training dataset for two independent variables PV1 and PV2 and one dependent
variable LG fault value under two modes of operation. The dataset has been taken
in both islanded and grid-connected modes and imported as a “CSV” file to the
ML platform in Raspberry Pi. The ML algorithm used here is “multiple regression”
for prediction of fault current value in both the modes of operation. This system
includes four inputs: PV1, PV2, reference current value (given in terms of voltage),
and mode of operation from the RPS. The algorithm has been designed in such a
manner that if the value from the mode of operation input is greater than “1”, then
it is grid-connected mode else islanded mode. The input values from the RPSs are
given through Arduino which acts here as an analog to digital converter to Raspberry
Pi. The algorithm predicts the LG fault current value and compares it with the input
current value. If the LG fault current value is less than the input value, then there is a
fault condition, the relay trips, and the LED glows and buzzer buzzes. This system,
thus, leads to the development of more stable and reliable power system network
634 A. Dey et al.
as the incorporation of DGs changes the grid contribution to the fault current and,
thus, changes the protection needs of the system. The constant need for change in the
current settings and time settings in the relay according to the fault current level and
modes of operation makes the adaptive protection scheme an inevitable part of the
modern power system network. This research work is, thus, a venture made to make
a small but significant contribution in making the modern power system network
more durable, reliable, and stable. The adaptive protection scheme developed can
have wide range of applications and scope. The scheme developed can be used
in the microgrid connected to the conventional power system network as well as
for microgrids operates in an islanded mode. The algorithm used can be adopted
for the protection of real-time microgrids as the proposed model collected system
data as voltage signals and the processor processes those data. The software model
developed using machine learning algorithm will process these data and identify
the fault condition. This proposed method can be implemented for the protection of
real-time microgrids.
References
1. Barra PHA, Coury DV, Fernandes RAS (2019) A survey on adaptive protection of micro grids
and distribution system with distributed generators. Published in: Elsevier
2. Lin H, Sun K, Tan Z-H et al (2019) Adaptive protection combined with machine learning for
micro grids. IET Gener, Transm Distrib 13:770–779
3. Mishra M, Rout PK (2018) Detection and classification of micro grid faults based on HHT and
machine learning techniques. IET Gener Transm Distrib 12(2):388–397
4. Che L, Shahidehpour M, Alabdulwahab A, Al-Turki Y (2015) Hierarchical coordination of a
community micro grid with AC and DC Micro grids. In: Published in: IEEE transactions on
Smart Grid, vol 6, no 6
5. Anil Kumar P, Shankar J, Nagaraju Y (2013) Protection issues in Micro grid. Int J Appl Control,
Electr Electron Eng (IJACEEE) 1(1)
6. Kandukuri S, Vyjayanthi C (2018) Fault level analysis in modern electrical distribution system
considering various distributed generations. In: Proceedings of 2018 international conference
on power, energy, control and transmission systems, Chennai. 978-1-5386-3817-0/18
7. Kailun Z, Kumar DS, Srinivasan D (2018)An adaptive overcurrent protection Scheme for micro
grids based on real-time digital simulation. In: Published in:2018 71st annual conference for
protective relay engineers (CPRE). TX, USA. https://doi.org/10.1109/CPRE.2018.8349769
8. Nikkhajoei H, Lasseter RH (2007) Micro grid Protection. In: 2007 IEEE power engineering
society general meeting. Tampa, FL, USA. https://doi.org/10.1109/PES.2007.385805
Power Quality Improvement
in Distribution Network Using
Levenberg–Marquardt (LM)
Algorithm-Based Controller for Active
Current Harmonic Filter
P. Balamurugan, N. Senthil Kumar, and C. Vaithilingam
Abstract Levenberg—Marquardt (LM) algorithm-based current controller for

shunt active harmonic filter (SAHF) is implemented for minimizing harmonic
currents in power distribution network. A simple and versatile soft computing algo-
rithm for implementing controller to minimize the current error of SAHF is developed
for improving power quality (PQ) in conventional distribution grid. The effective-
ness of the control algorithm proposed is analyzed considering group of balanced
and unbalanced nonlinear supplied from a balanced sinusoidal AC power supply.
The objective is to minimize the tracking error in the output of SAHF compen-
sating current harmonics resulting in distortion-free source current. Simultaneous
compensation of reactive power by SAHF results in unity power factor operation at
the source. The algorithm is developed and implemented using MATLAB/Simulink
environment using discrete solver.
Keywords LM algorithm · Artificial neural network (ANN) · SAHF ·

Harmonics · THD
1 Introduction
In recent decade, the growth of renewable energy sources emerged as an alternate

for the conventional fossil fuels due to the development of high-power converters.
Sophisticated and fast-acting controllers enable the interfacing of renewable sources
with power grid and interconnection of several types of renewables to operate in inte-
grated manner. Electric mobility has gained attraction as an alternate for the existing
P. Balamurugan (B)
eVITRC, Vellore Institute of Technology, Chennai, India
N. Senthil Kumar · C. Vaithilingam
School of Electrical Engineering, Vellore Institute of Technology, Chennai, India
C. Vaithilingam
https://doi.org/10.1007/978-981-16-4943-1_59
636 P. Balamurugan et al.
Fig. 1 Illustration of SAHF XS

iS iL
if
K e V DC Nonlinear
~ vs Load
transportation system to provide cleaner environment. In all the mentioned applica-

tions, power converters bridge the energy conversion from one form to another form.
Power converters integrating renewables and power network introduce harmonics
into it resulting in harmonic distortion. Conventional grids are designed to operate
only at fundamental frequency which cannot withstand the harmonic pollution. This
results in proper functioning of sensitive equipment like adjustable speed drives,
lifesaving equipment in hospitals, microprocessor/controller-based process control
systems in various industries, reduced losses throughout the system [1].
Harmonics are limited by means of installing passive or active power filters at suit-
able locations in the power system. Active power filters facilitate flexible compen-
sation and advantageous compared to passive filters. SAHF is a controlled current
source connected in parallel to the sensitive load to supply the harmonics required
by the load as depicted in Fig. 1. Kirchhoff’s law at node ‘a’ yields is = iL + if . The
nonsinusoidal current drawn by nonlinear load is resolved in to sinusoids of integer
multiples of supply frequency using Fourier series. The nonlinear load current (iL )
is the sum of current component at line/fundamental frequency (iL1 ) and currents at
integer multiple of fundamental frequency components (iLh ). If SAHF delivers iLh ,
then the source current will be at fundamental frequency of iL [1].
The current injected by SAHF is essential in regulating harmonics in the grid,
and hence, the problem is formulated to regulate the current output of SAHF.
The harmonic extraction from the distorted current essentially determines the
performance. In this work, pq-theory is adopted for its flexibility in compensation.
ANN technique develops self-decisive system enabling the designer the flexi-
bility and adaptive nature for control design. ANNs are significant in the following
aspects: The input–output relationships need not to be specified, and the same is estab-
lished through learning process or adaptive through any learning algorithm. Secon-
darily, the parallel computation architecture of ANN increases speed and reliability
of the system. Adaptive networks estimate the reference compensation currents, and
a multilayer perceptron feedforward network-based hysteresis band comparator is
developed in [2] which employs backpropagation algorithm to train the feedforward
network. Intelligent adaptive power line conditioner based on supervised learning
rule to realize optimal adaptive weight vector for filter is realized in [3] with adaptive
NN for harmonic computation, and a feedforward network implements PWM control
using digital signal processor in [2]. Control efforts to balance the source currents,
canceling the neutral currents. Online extraction of fundamental voltage component
Power Quality Improvement in Distribution Network … 637
using Adaline network is proposed in [4]. It recovers the balanced and equilibrated
voltage and different methods for harmonic filtering. Several soft computing tech-
niques like fuzzy logic, neural control and genetic algorithm-based filter control
were proposed and compared in [5]. Adaline-based notch filter was implemented to
extract fundamental grid frequency estimated based on neural-based PLL employing
ADALINE NN, trained by a least square algorithm (LS). This enabled online compu-
tation of compensation currents from the values of load current and PCC voltage to
work. Dynamic performance improvement of SAHF by utilizing the dynamics of
DC link voltage is proposed in [6] using Adaline-based current. NN control strategy
for selective harmonic elimination was proposed in [7] to extract the harmonic refer-
ence for a medium voltage 6-pulse current source converter by introducing time
delay. Feedforward multilayer neural network-based current control is proposed in [8]
with step-by-step approach. Echo-state network-based fundamental current extrac-
tion technique was implemented. It evaluates weight values to extract the reference
currents. Variable leaky least mean square (VLLMS)-based controller for making
SAHF self-adaptive to the operating conditions was proposed in [9]. VLLMS-based
algorithm replaced the conventional weight updating algorithm, enhancing the execu-
tion speed and weight extraction to a greater extent, claiming variable band adap-
tive current control technique to circumvent auditory noise, and irregular switching
frequency in conventional hysteresis regulator.
In this paper, Levenberg–Marquardt algorithm is adopted to control the SAHF.
LM algorithm locates the minimum of a multivariate function in a reiterative manner,
which is expressed as the sum of squares of nonlinear real-valued functions. Evolving
as a standard solution technique, LM algorithm is applied for nonlinear least square
problems. This work utilizes LM algorithm to generate duty cycle of PWM signals
for SAHF. Discussing the LM algorithm formulation in Sect. 2, this paper covers
the harmonic compensation principle in Sect. 3. Section 4 describes the algorithmic
implementation of the proposed control in MATLAB/Simulink environment.
2 Levenberg–Marquardt Algorithm
The Levenberg–Marquardt (LM) algorithm is a combination of the Gauss–Newton

and steepest descent methods. When the deviation between currently arrived solution
and the desired one is more, the algorithm switches to steepest descent method:
providing sluggish, but assured convergence. When the deviation between currently
arrived solution and the desired solution is minimal, it follows Gauss–Newton method
[10].
The update rule of Gauss–Newton method is given by
−1
wk+1 = wk − JkT Jk Jk ek (1)
For convergence, the approximated Hessian matrix J T J approximation must be

invertible. To ensure this condition, another approximation is introduced to Hessian
matrix by the Levenberg–Marquardt algorithm expressed as in Eq. (2):
H ≈ J T J + μI (2)
where μ = combination coefficient (positive always) and I = identity matrix.

It may be observed from (2), the principal diagonal elements of the approximated
Hessian matrix will be a positive number. The guesstimate in (2) when applied to
Eq. (1) ensures matrix H is invertible always. With the approximations applied, the
new update law of Levenberg–Marquardt algorithm can be expressed as
wk+1 = wk −(JkT Jk + μI )−1 Jk ek (3)
During training process, the Levenberg–Marquardt algorithm switches between

the Gauss–Newton algorithm and the steepest descent algorithm in computing the
grouping coefficient ‘μ’. Gauss–Newton algorithm is employed when negligible
values of μ (nearly zero) approximating Eq. (3) to Eq. (1). The steepest descent
method is applied for larger values of μ, by approximating Eq. (3) to wk+1 = wk −
α gk . If the grouping coefficient μ in Eq. (3) is larger, it can be taken as the learning
coefficient: α = 1/μ, in the steepest descent method.
Node y is significant and flexible, to define a neuron output. To interpret, yj, i , is
the ith input of jth neuron. To define the output of neuron j, it is represented as yj .
The single index of node ‘y’ implies an output neuron, whereas two indices (neuron
and input) indicate the node as input of a neuron.
The output node of jth neuron is computed as in Eq. (4).

y j = f j net j (4)
where f j = the activation function of jth neuron and netj = the sum of weighted input
nodes of neuron j

ni
net j = w j,i y j,i + w j,0 (5)
i=1
where
Y j, i = ith input node of neuron j, weighted by wj, i
wj, 0 = bias weight of neuron j.
The training process for LM algorithm must be framed as follows:

1. Generate random initial weights and compute the total error.
2. Adjust and update weights using Eq. (3).
3. Evaluate total error using updated new weights.
4. After an update, if the present overall error is enlarged, then pull back weight
vector to the preceding value and upsurge grouping coefficient μ by a factor 10
or higher. Repeat step-2 and try updating weights.
5. After an update, if the present overall error is diminished, then retain the new
weight vector to its present value and lessening the grouping coefficient μ by a
factor 10 or by the same factor as in step-4.
6. Return to step-2 with the new weights until the recent total error is smaller than
the vital value.
3 Principle of Harmonic Compensation
Utilizing Clarke’s transformation, Akagi H [11] formulated a theory to compute

instantaneous real and reactive powers with instantaneous voltages (v) and currents
(i) as key parameters. Transforming a set of time varying space varying phasors in to
two orthogonal components and a zero-sequence component, powers are computed.
⎡ ⎤ ⎡ ⎤⎡ ⎤
vα 1 −0.5 −0.5 vr
⎣ vβ ⎦ = 2⎣
0 0.866 −0.866 ⎦⎣ v y ⎦ (6)
3
v0 1 1 1 vb
⎡ ⎤ ⎡ ⎤⎡ ⎤
iα 1 −0.5 −0.5 ir
⎣ iβ ⎦ = 2⎣ ⎦ ⎣
0 0.866 −0.866 iy ⎦
3
i0 1 1 1 ib
The instantaneous real power (p) and reactive power (q) are calculated in the
converted domain as in (7).
p = vα i α + vβ i β + v0 i 0
q = vβ i α − vβ i α (7)
The components of power in (7) are composed of fundamental and nonfunda-

mental components. A higher-order low-pass Butterworth filter is used to segregate
the average component of power from the computed total power. The average value
corresponds to instantaneous real and reactive powers at the fundamental frequency
of the system. The oscillating components of power correspond to the harmonic
frequencies.
The compensating reference currents are computed from (7) after extricating the
average and oscillating components using (8) for SAHF.
∗
i cα 1 vα vβ pc∗
∗ = (8)
i cβ vα2 + vβ2 vβ −vα qc∗
The inverse Clarke’s transformation of Eq. (8) results in real-time current

references for the filter controller and is expressed in Eq. (9).
⎡ ∗
⎤ ⎡ ⎤
i cr 1 0
∗ ⎦=
⎣ i cy 2⎣ i∗
−0.5 0.866 ⎦ cα ∗ (9)
∗ 3 i cβ
i cb −0.5 −0.866
The SAHF is forced to track the reference currents employing suitable current
control algorithms used in practice. The compensation for oscillating components of
power results in harmonic free source current and voltage at constant power factor
decided by the reactive power drawn from the source. For unity power factor, along
with oscillating real power, the compensator must supply total reactive power.
Typical VSC-based SAHF power circuit in [1] is essentially a voltage source
converter with capacitor-powered DC link, connected at PCC in distribution network
with an interfacing filter. The filter controller is equipped with data acquisition system
to compute the harmonic currents by processing the system voltage and load current
signals. In addition, the filter controller ensures DC link voltage regulation during its
operation. The calculated reference and filter current outputs are related to produce
the PWM switching signals for the power switches in VSC and resulting in minimal
regulating error. Harmonic currents of any order can be synthesized and injected
by appropriate selection of power components and computing the reference current
[12].
4 Design of ANN-Based Current Controller
A customized neural network controller was generated using training data obtained
from the measured samples of the active filter controlled. The input to the neural
network is the current deviation/error from the comparator. The output is the duty
ratio for the switches in voltage source converter. The converter duty ratio must be
regulated to maintain DC voltage regulation and output current regulation of the
voltage source converter. The duty ratio is then converted into PWM signals by
comparing the duty ratio output from ANN with a unipolar triangular ramp carrier.
The switching frequency of the voltage source converter limits the size of the filter
inductor.
In the design of ANN controller, error samples acquired at a sample interval of
1 µs are fed to the neural network. Using ‘nntool’ command, the NN is generated
and trained. The input samples are parameterized as vector in the workspace and
imported to the toolbox. The selection parameters like training function, number of
input and output layers and neuron contained in the layers are configured initially. The
network type is chosen as feedforward-backpropagation type for better performance
and mean square error as the performance parameter for convergence. One such
Fig. 2 Customized neural network generated and trained using MATLAB
network as shown in Fig. 2 is created and trained using LM algorithm described in

Sect. 2.
5 MATLAB implementation of ANN-based current

controller for SAHF
The MATLAB implementation of ANN-based current controller for SAHF was

developed in MATLAB as discussed in Sect. 4 and implemented using Simulink
tool. The filter configuration for distribution network is shown in Fig. 3. The condi-
tion considered for simulation study and implementation and the system parameters
taken for study are detailed in Table 1.
Ls
~ vs
6
S1-S6
ifabc iL
3Φ – Balanced
Nonlinear Load
Lf
+
VDC CDC
-
VDC VSI
-
VDC* PDC
PI
Σ
+ Controller
PDC
+
vsabc vαβ0 +
Clarke’s Instantaneous Selection of Σ
iLa bc Transformation iαβ0 Power Calculations Compensating
(abc to αβ0) (p, q) Powers
Carrier Signal
vαβ0 qc pc
ifa bc
iα Reference
6
ANN based ei i*fabc Inverse Clarke’s
Current Σ Transformation iβ Current
S1-S6 Controller (αβ0 to abc) Calculation
PWM Signals
Fig. 3 Simulink implementation block diagram of ANN current controller-based SAHF

Table 1 System configuration and parameters used for the design of ANN current controller
System parameter Value
Source parameters
Voltage (V s ) 440 V, 50 Hz, 3-phase AC supply with no
transformer
Line resistance (Rs ) and inductance (L s ) 0.1 ohms, 5 mH
Load parameters
Nature of load Three-phase diode bridge rectifier
Load resistance (Rl ) and inductance (L l ) 10 ohms, 15 mH
Filter parameters
Voltage source converter IGBT based three-phase three-leg converter with DC
capacitor in DC link with floating neutral
Filter inductance (L f ) and resistance (Rf ) 5 mH, 0.2 ohms
DC link capacitance (C DC ) 1000 microfarad
DC link voltage reference (V DC *) 800 V
DC link controller Output limited PI controller
Filter controller ANN controller
Filter interface Direct coupling with L filter
The results of simulation are shown in Fig. 4. The waveforms of source voltage,
source current after compensation, load current, harmonic current in phase ‘c’ and
DC link voltage are shown, respectively. Since only one nonlinear load is considered,
the source current is same as the load current in the absence of compensation using
SAHF. Hence, the pre-compensated source current is similar to the load current (iL ).
To analyze the effect of current controller under dynamic conditions, an additional
load is added in the DC side of the diode bridge rectifier whose rating is similar to the
first load. The transient behavior of SAHF with the proposed current controller for
load changes on DC side is studied in this study. The second load is switched ‘ON’
at the mid-range of simulation, and the transients were observed in the behavior of
parameters like filter output current, DC link capacitor voltage and source current
compensation.
The effectiveness of compensation is measured in terms of harmonic indices
performance. Total harmonic distortion in source current is the important parameter
that indicates the quality of the source current waveshape and hence the effectiveness
of harmonic compensation. Also, lesser the value of THD, closer the wave shape is
sinusoidal and power factor is near unity. Hence, the source current waveform is
analyzed using fast Fourier transform (FFT) tool with 50 Hz as the fundamental
component. The FFT is expressed as wave spectrum and is shown in Fig. 5a for
the uncompensated source current. The distortion level is as high as 28.96% for the
measured uncompensated source current. The standard allowable harmonic penetra-
tion pertaining to the operating voltage is 5% as governed by IEEE519:2014 guide-
lines. The dominant harmonics are 250, 350, 550, 650 and 850 Hz for the operating
(a) Source voltage (vs) (b) Source current (is)
(c) Load current (iL) (d) Filter current in Phase ‘c’
(e) DC-link Capacitor Voltage (VDC)
Fig. 4 Simulink results of ANN current controller-based SAHF
(a) Pre-compensation. (b) Post-compensation.
Fig. 5 Harmonic spectrum of source current in phase—‘a’
condition considered. Among these, up to 550 Hz, the harmonic elimination using
conventional passive filter fails to be a feasible solution.
In the presence of active power compensation filter, with the proposed current
controller, the THD of compensated source current is 3.36% and is better than the
results obtained with conventional current controllers in [1, 12]. Moreover, the THD
is much within the limits specified by IEEE519:2014 guidelines for the utility grid at
Table 2 Performance analysis of ANN current controller under different operating conditions
System Balanced load Unbalanced load Single phasing
parameter I RMS (A) % THD I RMS (A) % THD I RMS (A) % THD
Phase—a 3.5 3.36 3.5 2.66 3.8 4.08
Phase—b 3.5 3.36 3.5 2.68 3.8 4.1
Phase—c 3.5 3.36 5 2.09 3.8 4.09
distribution voltage levels. The harmonic spectrum of compensated source current is

shown in Fig. 5b. Similar analysis was conducted for different operating conditions
like addition of unbalanced nonlinear load and single-phasing of balanced linear
load by open circuiting phase ‘c’ at the source. The results of source currents and
the corresponding THD levels are shown in Table 2.
The performance of the proposed LM algorithm-based ANN for shunt active
harmonic filter is evident from Table 2 after compensation. The source currents are
balanced sinusoids under various operating conditions at the load side. In this work,
the proposed current control algorithm is evaluated under balanced sinusoidal supply
voltage considerations, and the performance is observed to be satisfactory. Similar
studies can be extended under supply voltage nonidealities to deal with optimum and
real-time implementation of the proposed current control algorithm.
6 Conclusion
In this paper, a feedforward-backpropagation-based artificial neural network based

on LM algorithm was developed, implemented and analyzed for current control of
shunt active harmonic filter at distribution level. The results are evidence for the proof
of concept in the application of ANN for power control and harmonic regulation at
distribution voltages. The objective of reference current tracking at nonfundamental
frequency to compensate for harmonic currents is in effect with negligible tracking
error. The results of simulation show that the source currents are sinusoids and
in-phase with their respective voltage sinusoids achieving near unity power factor
operation. Also, the problem of eliminating low-frequency harmonics with traditional
passive filters is greatly eliminated by the use of SAHF resulting in cheaper power
quality solution.
References
1. Balamurugan P, Senthil Kumar N (2019) Regulation of current harmonics in grid with dead-
beat controlled shunt active power filter. Prz. Elektrotechniczny. 95:17–23. https://doi.org/10.
15199/48.2019.12.04
2. Salmerón P, Vázquez JR (2005) Practical design of a three-phase active power-line conditioner

controlled by artificial neural networks. IEEE Trans Power Deliv 20:1037–1044. https://doi.
org/10.1109/TPWRD.2004.838513
3. Lin HC (2004) Intelligent neural-network-based adaptive power-line conditioner for real-time
harmonics filtering. In: IEE Proceedings: generation, transmission and distribution. https://doi.
org/10.1049/ip-gtd:20040757
4. Abdeslam DO, Wira P, Mercklé J, Flieller D, Chapuis YA (2007) A unified artificial neural
network architecture for active power filters. IEEE Trans Ind Electron 54. https://doi.org/10.
1109/TIE.2006.888758
5. Kumar P, Mahajan A (2009) Soft computing techniques for the control of an active power filter.
IEEE Trans Power Deliv 24:452–461. https://doi.org/10.1109/TPWRD.2008.2005881
6. Bhattacharya A, Chakraborty C (2011) A shunt active power filter with enhanced performance
using ANN-based predictive and adaptive controllers. IEEE Trans Ind Electron 58:421–428.
https://doi.org/10.1109/TIE.2010.2070770
7. Hamad MS, Gadoue SM, Williams BW (2012) Harmonic compensation of a six-pulse current
source controlled converter using neural network-based shunt active power filter. IET Power
Electron 5:747–754. https://doi.org/10.1049/iet-pel.2011.0336
8. Qasim M, Khadkikar V (2014) Application of artificial neural networks for shunt active power
filter control. IEEE Trans Ind Infor 10:1765–1774. https://doi.org/10.1109/TII.2014.2322580
9. Ray PK (2018) Power quality improvement using VLLMS based adaptive shunt active filter.
CPSS Trans Power Electron Appl 3:154–162. https://doi.org/10.24295/cpsstpea.2018.00015
10. Yu H, Wilamowski BM (2011) Levenberg–Marquardt Training. In: Industrial electronics
handbook. pp. 12–1 to 12–15. CRC Press. https://doi.org/10.1134/1.1788770
11. Akagi H (1996) New trends in active filters for power conditioning https://ieeexplore.ieee.org/
abstract/document/556633. https://doi.org/10.1109/28.556633.
12. Parandhaman B, Nataraj SK, Baladhandautham CB (2020) Optimization of DC-link voltage
regulator using Bat algorithm for proportional resonant controller-based current control of
shunt active power filter in distribution network. Int Trans Electr Energy Syst 30:1–15. https://
doi.org/10.1002/2050-7038.12369
Modified Multiverse Optimization,
Perturb and Observer Algorithm-Based
MPPT for Grid-Connected Photovoltaic
System
N. K. Rayaguru and S. Sekar
Abstract The main objective this work is to analyze the total harmonic distortion
for grid-connected photovoltaic system. A novel modified multiverse optimization
algorithm (MMVO) is proposed to unravel the issues of the conventional maximum
power point tracking (MPPT). The goal of this new methodology is to spice up the
PV array efficiency. The proposed modified algorithm has a low convergence time
and excessive efficiency compare to perturb and observe (P&O) algorithm.
Keywords Modified MVO · P&O · PV system · Utility grid · Grasshopper

optimization
1 Introduction
Many strategies of monitoring the maximum power point of a PV have been

established to decide the disadvantage of efficiency without algorithm or suitable
controllers [1]. The MPPT governs the DC-to-DC converter that serves as an inter-
face between the PVM and the load. The number of MPPT-based algorithms has been
reviewed within the literatures such as standard perturb and observe (SPO), fuzzy
logic management (FLM), incremental conductance (INC), golden part search (GPS),
gray wolf optimization (GWO) and multiverse optimizer (MVO) [2]. Alternatively,
the multiverse optimization (MVO) is analyzed to be an algorithm with benefits like
good tuned outcomes of result and quick convergence process as in comparison with
many different MPPT algorithms [3]. The new modification MVO approaches the
benefits of quick convergence time and the benefit of the high-quality tuned outcomes
with a small step measurement to trace the MPP with greater accuracy and really
low oscillations [4].
N. K. Rayaguru (B) · S. Sekar

Department of Electrical and Electronics Engineering, Hindustan Institute of Technology &
Science, Chennai, India
S. Sekar
https://doi.org/10.1007/978-981-16-4943-1_61
648 N. K. Rayaguru and S. Sekar
1.1 Modified Multiverse Optimization and It’s Flowchart
Multiverse optimization [5] is a population-based algorithm. The MVO algorithm

is simulated on the basis of the concepts that is present theoretically in astronomy
together with white holes which are the key rudiments of the formation of universes
and have never been observed in the whole universe; black holes which draw objects
toward them by gravity applied to those around them; and worm holes which are
tunnels that unite parts in the external space. Here, the three components are mathe-
matically modeled to build up an optimizer, which mimics teleportation and object
exchange among universes starting from white holes through wormholes to black
holes. The aforesaid algorithm believes on a population of growing individuals
(Fig. 1).
Each entity is a candidate solution that encodes the three concepts of black, white
and wormholes in its objects. The multicandidate solutions aid and share information
with each other to move toward promising area [6].

j ux ij + TD + (hvi − lvi ) ∗ r d + qb j
uxi = j
ux j
r c < EPW, r c ≥ EPW, r b < EPW, r b ≥ EPW (1)
Fig. 1 Flowchart of modified multiverse optimization

Modified Multiverse Optimization, Perturb and Observer … 649
j
where ux ij —jth variable in ith universe; ux j —the jth formed universe; TD stands
for distance traveling rate; EPW—represents existence probability of wormhole; lv
represents lower variable; hv represents highest variable; qb j —represents lowest jth
variable; r b , r c , r d , re represents the variable with random numbers between [0, 1].
For the two coefficients, the formulas are as follows:

low value − high value
EPW = low value − q (2)
J
where q indicates the current iteration and J indicates the iterations with maximum
number.
1
qg
TD = 1 − 1
(3)
Jg
g = 0, range from[0, 10]
1.2 The P&O Algorithm and It’s Flowchart
Here voltage is perturbed by a minute increment V. The resulting alteration in power
is P. It is analyzed that if P is positive, the next perturbance is in this direction. If
P is in negative, the system operation has moved away from the maximum power
point (MPP). The selection of duty cycle will affect the system, and the steady-state
error will be reached in the system [7]. The complete iteration process has been done
as shown in Fig. 2. P&O algorithm procedure is considered as follows.
Step1: Check and read V (k) and I (k), initialize duty cycle D.
Step 2: Calculate P (k) = V (k) *I (k).
Step 3: Calculate P = P (k) − P (k-1), V = V (k) −V (k − 1).
Step 4: If P > 0 is yes, then check V > 0. If V > 0 is yes, then D = D + D
otherwise D = D-D.
Step 5: If P > 0 is no, then check V > 0. If V > 0 is yes, then D = D-D
otherwise D = D + D.
1.3 Proposed System Description Using Both Algorithms
A PV array of 100 kW is coupled to a 25 kV grid through a DC-DC boost converter

followed by a 3-F, three-level voltage source converter (VSC). The model consists
of (a) PV array that delivers a maximum of 100 kW power at an irradiance of
Fig. 2 Perturb and observe optimization flowchart
500 W/m2 (b) DC-DC step-up converter, (c) three-level three-phase VSC, (d) a three-
phase coupling transformer rating of 100 kVA, 260/25 kV and (e) utility grid. This
configuration has proposed a boost converter in a grid-tied PV system through voltage
source converter, filter and step-down coupling transformer as shown in Fig. 3. The
MPPT algorithms are executed through boost converter (DC-to-DC converter) with
decoupled control tactic not only supplying generated PV power to the grid but also
it satisfied the load requirements of the consumer side [7, 8]. The P&O-based MPPT
with the system is depicted in Fig. 3.
Figure 4a, b shows the characteristics between PV voltage and current; PV power
and voltage shows the indication of maximum power point using P&O MPPT
controller. The simulation waveforms for PV voltage, current and duty cycle vari-
ations corresponding to changes in solar irradiance and temperature are shown in
Fig. 5. The DC-DC converter of PV system is controlled with P&O MPPT algo-
rithm. With this implementation, they obtained maximum PV current 382.08 A at a
Fig. 3 PV panel with 100 W utility grid system through P&O
Fig. 4 a PV array
characteristics of current
verses voltage. b Power
verses voltage in perturb and
observe method
duty cycle of 0.7. The maximum power given to grid is approximately 90.2 KW as
shown in Fig. 6. The grid voltage and current are shown in Fig. 7.
In Fig. 8, it is observed that the harmonic produced by the photovoltaic inverter
system is 1.26%. The multilevel inverter is very much useful for providing preferred
AC output voltage. In the MATLAB environment, a bridge inverter has used and is
integrated of PV array into the utility grid. Table 2 depicts the harmonically generated
current at the output of the Vsc (Table 1).
Fig. 5 Waveform for PV system parameters. a Irradiance. b Temperature. c PV power. d PV

voltage. e Duty cycle
Fig. 6 Waveform for PV—grid power
2 Modified Multiverse Optimization-Based MPPT
The MMVO-based MPPT with the system is shown in Fig. 9. In this case, the
proposed grid-connected PV system is implemented with MMVO optimization for
generating duty cycle required for DC-DC converter. Table 4 depicts the harmonically
generated current at the output of the Vsc [9] (Table 3).
Figure 10a, b shows the characteristics between PV voltage and current. The simu-
lation waveforms for PV voltage, current and duty cycle variations corresponding
to changes in solar irradiance and temperature are depicted in Fig. 11. With this
implementation, they obtained maximum PV current 394.08 A at a duty cycle of
0.82. The maximum power given to grid is approximately 106.2 KW as revealed in
Fig. 12. The grid voltage and current are publicized in Fig. 13.
Fig. 7 Waveform for grid voltage and current
Fig. 8 THD for inverter current

Parameters Values
with P&O method
Irradiance 1000 W/m2
Temperature 25°c
PV array’s output voltage 273.54 V
PV array’s output current 366.98 A
Maximum power extracted from PV panel 90.22 KW
Table 2 Current harmonics

Order No RMS values (kA) % of fundamental
from VSC using perturb and
observe method 1 0.01035 100
3 0.00012 0.145
5 0.00104 1.002
7 0.00090 0.850
9 0.00045 0.390
11 0.00020 0.195
13 0.00070 0.250
15 0.00095 0.455
THD 0.01100 1.26
Fig.9 PV array with 100 KW utility grid system through MMVO

No. parameters Values
with MMVO method
Irradiance 1000 W/m2
Temperature 25°c
PV array’s output voltage 284.45 V
PV array’s output current 394.08 A
Maximum power extracted from PV panel 106.20 KW
In Fig. 14, it is observed that the harmonic produced by the photovoltaic inverter
system is 0.77%
The peak power extracted from PV panel is maximum in MMVO method when
compared to P&O method. It is depicted in Table 5. The constant temperature and
power output v/s solar irradiation are shown in Table 6. It is observed that solar
Table 4 Current harmonics

Harmonic number RMS current (KA) % of fundamental
from VSC MMVO method
1 0.01012 100
3 0.00010 0.125
5 0.00009 1.000
7 0.00070 0.740
9 0.00035 0.320
11 0.00015 0.155
13 0.00055 0.230
15 0.00078 0.405
THD 0.01500 0.77
Fig. 10 PV array
characteristics of current
verses voltage and power
verses voltage graph in
MMVO method
irradiation decreases and power output decreases. Solar irradiation remains constant;
but power increases, and it is shown in Table 7.
3 Conclusion
The modified multiverse optimization method has helped the power extraction and
reduced the harmonics level than the P&O method. The peak power of the PV
array has been tracked with adjustable duty cycle using both algorithms. The modi-
fied MVO method has given good efficiency when compared to other method. The
harmonic distortion has reduced to 1% from 1.58%.The future scope of this method
has to be done the algorithmic performance in the real panels.
Fig. 11 Waveform for PV system parameters. a Irradiance. b Temperature. c PV power. d PV

voltage. e Duty cycle
Fig. 12 Waveform for PV—grid power

Fig. 13 Waveform for grid voltage and current
Fig. 14 THD for inverter current with MMVO
Table 5 Maximum power

Algorithms Maximum power extracted Harmonics level
extracted and harmonics
(KW)
levels of the systems
MMVO 106.2 0.77%
P&O 90.80 1.26%
Table 6 Constant
Solar irradiation Temperature in Power output to the
temperature, power output
(W/m2 ) degrees grid (kW)
and solar irradiation
1000 25 106.22
980 25 92.10
960 25 89.37
940 25 86.90
920 25 85.96
900 25 84.43
Table 7 Constant and

Solar irradiation Temperature in Power output to the
irradiation, output and
(W/m2 ) degrees grid (kW)
temperature
1000 25 106.22
1000 30 91.20
1000 35 87.07
1000 40 85.20
1000 45 83.69
1000 50 82.43
References
1. Physical Progress (Achievements) ministry of new and renewable energy, Government of India.
31 January 2014. Retrieved 21 February 2014
2. Alik R, Jusoh A (2018) An enhanced P&O checking algorithm MPPT for high tracking efficiency
of partially shaded PV module. Sol Energy 163:570–580
3. Lakshmi M, Hemamalini S (2019) Coordinated control of MPPT and voltage regulation using
single—stage high gain DC–DC converter in a grid-connected PV system. Electr Power Syst
Res 169:65–73. https://doi.org/10.1016/j.epsr.2018.12.011
4. SeyedaliMirjalili SM, Mirjalili AL (2014) Grey wolf optimizer. Adv Eng Softw 69:46–61
5. Benmessahel I, Xie K, Chellal M (2018) A new evolutionary neural networks based on intrusion
detection systems using multiverse optimization. Appl Intell 48(8):2315–2327
6. Mirjalili S, Mirjalili SM, Hatamlou A (2016) Multi-verse optimizer: a nature-inspired algorithm
for global optimization. Neural Comput Appl 27(2):495–513
7. Tobnaghi M (2016) A Review on impacts of grid-connected PV system on distribution network.
Int J Electr, Comput, Energ, Electron Commun Eng 10(1):137–152
8. Sreedevi J et al (2016) A study on grid connected PV system. In: IEEE conference
9. Jui JJ, Ashraf Ahmad M, Mohd Rashid MI (2020) Modified multi-verse optimizer for solving
numerical optimization problems. In: 2020 IEEE international conference on automatic control
and intelligent systems (I2CACIS). Shah Alam, Malaysia, pp 81–86. https://doi.org/10.1109/
I2CACIS49202.2020.9140097
IOT-Monitored EV Charging Stations
Using DC–DC Converter with Integrated
Split Battery Energy System
M. Arun Noyal Doss, R. Brindha, A. Ananthi Christy,

and Viswanathan Ganesh
Abstract This paper proposes a multiport converter-based idea of preparing a fully

monitored UFEVCS. Different converters are utilized in order to test the achievement
of ultrafast charging station. This charging station also seeks a factor of automation
in battery management system. This battery management is accomplished by the use
of IOT. A worldwide control project being designed to control different varieties of
power flowing directions if anything unbalances in the system. The methods are thus
simulated and therefore are being implemented on hardware. By the usage of CHB
converter, the SOC of a system can balance itself. IOT-based battery management
is carried in order to automate the charging station with much less efforts and more
efficiency. The parameters calculated from the sensors are current (I), voltage (V) and
power (W). These quantities from the sensors are transferred to things peak cloud.
Thing speak cloud can be accessed from anywhere at any moment of time. Complete
parameters of the battery of an EV can be uploaded to the user’s device
Keywords EV · UFEVCS · BMS · Cascaded H-bridge converter (CHB) · State of

charge (SOC) · Dual half-bridge converter · Internet of Things (IOT) · NodeMCU
1 Introduction
In today’s world, there are many advancements in usual cars and other vehicles, but
the most reliable and fit for the environment conditions are electric vehicle. These
vehicles have a plethora of merits. These are also suitable for certain load demand.
The best part of this technology is that it requires electric charge as fuel to operate.
These vehicles are easy to operate for the drivers and also have sufficient power
M. Arun Noyal Doss (B) · R. Brindha · V. Ganesh

EEE Department, SRM Institute of Science and Technology, Chennai, India
R. Brindha
A. Ananthi Christy
Saveetha School of Engineering, SIMATS, Saveetha University, Chennai, India
https://doi.org/10.1007/978-981-16-4943-1_62
660 M. Arun Noyal Doss et al.
[1–3]. This ideology of EV charging station also consists of IOT-based BMS. The
data measured from the battery of E are is uploaded to cloud and can be accessed
by the owner of the car or by the workers and owner of the station in their respected
devices. EVs usually attain power to operate from the grid supply. This involves
AEVs and PHEVs. AEVs are driven by usually one or more amount of motors. It
receives power from the grid and supplies electric charge to the different types of
batteries. These vehicles are pollution-free as there is no consumption of any fuel
like diesel, LPG. PHEVs motor usually accepts charging by the batteries which itself
takes power from the grid supply. This power of motor is usually utilized to power
up the engine of the vehicle. BEVs, battery electric vehicles, also known as EVs,
are fully functional electric vehicles which consist batteries as their power source.
These vehicles store charge in high-capacity storage packs. This power obtained
from the batteries is completely utilized to run the car and its other assets (gadgets).
These BEVs absorb power from the external sources. That is why they never emit
any unhealthy fumes in the environment [4–6]. The groupings are L1, L2, and L3 or
DC quick (fast) charging. L1 EV charging uses a common family (120v) outlet to
be plugged into the EVs and consumes approximately 8 h to charge it for roughly
85–90 miles. L1 charging is a basic procedure which can be carried out at home
or at any other working place. L1 chargers are capable of powering up most EVs
available. L2 charging demands a special charging station which provides power at
240 V. L2 chargers are commonly found at work environments and open charging
stations and will consume around 3 to 4 h to power up a battery to 85–90 miles of
range. L3 charging, DC fast charging, or just fast charging is currently the fastest
charging method in the EV charging system. DC fast chargers are found at respective
EV charging stations and power up a battery to 100 miles extending around 35 min.
PHEVs can power up the battery via both regenerative braking and supply from the
grid, while basic EVs can travel up to 2–3 miles before the motor turns on which is
running on fuel [8–10]. These EVs can travel up to the range of 10–50 miles before
their fuel motors come into action. HEVs consist of both gas and power. The electric
charge is resupplied by the braking mechanism of the vehicle. Due to this method
of braking, the battery life is extended. The electric engine kicks in as the speed
increases, and fuel motor stops operating [11, 12].
2 Existing System
There are many systems which are used for charging system for certain demands. All
of the charging systems usually accepts AC supply from the grid. This AC supply will
not provide fast charging or will not meet charging requirements all the time. A basic
ideology is about applying a rectifier after receiving the supply from the grid. This
converts the AC input and provides DC output. The major part we commonly snub
here is the ripples; while converting the supply, this also brings too many unwanted
ripples which need to be removed in order to save energy. This is performed by
using a filter along with the rectifier. Now, all we want is to minimize the size of
IOT-Monitored EV Charging Stations Using DC–DC Converter … 661
the new gadget and also improve its functionality. This increases the usage of boost
converter in order to achieve higher voltage without increasing the number of cells.
Boost converters can be of different varieties. Suppose some stations using DC to
DC converters with boosting operations. These have some demerits as they cannot
have additional battery storage system. CHB converter is one of the options which
we have come across in order to come over this demerit. CHB can allow the system
to apply another storage system which can store charge while grid connection is lost.
The boost converter plays a vital role in order to achieve ultra fast charging. Some
other ideas are also in the box as some renewable resources are also used to generate
power and to utilize it for the charging of electric vehicle. There exist less reliability
with such systems. There are some systems which connects six interleaved groups
consisting of two DC–DC converters connected in parallel, performing the charging
process of the battery and fed by DC. But limitation is that we need to concentrate
in total harmonic distortion and power factors. PV panel optimum-sizing algorithm
has been developed. Battery storage system and transformer are connected to the
grid for the charging station surrounded by grid constraints. But the limitation is that
the optimization of loss of load needs to be concentrated. In huge batteries during
charging and discharging rates to avail low charge holding loss, special cathodes
made from nano-structures are used which are bi-continuous. The limitation is that
the battery compatibility needs to be increased.
3 Proposed System
This paper proposes a multiport converter-based idea of preparing a fully moni-

tored UFEVCS. Different converters are utilized in order to test the achievement of
ultrafast charging station. This charging station also seeks a factor of automation in
battery management system. This battery management is accomplished by the use
of IOT. A worldwide control project being designed to control different varieties
of power flowing directions if anything unbalances in the system. The methods are
thus simulated and therefore are being implemented on hardware. By the usage of
CHB converter, the SOC of a system can balance itself. IOT-based monitoring is
performed in order to maintain station (charging) with much less efforts and more
efficiency. The constraints registered from the sensors is current (I), voltage (V) and
power (W). These constraints registered from the sensors are transferred to ThingS-
peak cloud. ThingSpeak cloud can be accessed from anywhere at any moment of
time [1]. Complete battery status is uploaded to the user’s gadgets so that the vendor
gets complete information about the station at any moment of time from any place;
it also does fast charging. If grid fails to supply, still the operation can be executed
(Fig. 1).
Attaining the supply from the grid which charges the battery, in middle, there is a
CHB converter which permits to imply an external battery storage which will store
charge and provide it to the battery when grid connection is OFF. IOT is used in
our model to measure the parameters of the EV battery. The parameter will be then
Fig. 1 Block diagram of proposed system
directly uploaded to the cloud. The vendor can access the cloud to get the information
about the implanted charging station.
4 Circuit Diagram
Left side of this circuit is CHB converter which permits connections of an additional
external storage system, the right side of the circuit is DC–DC converter, and this
permits a multiport charging system. The output voltage is attained at the end of the
operation of DC–DC converter. DC to DC converter: The output DC is at an another
yield level in comparison with the input voltage. They are significant in versatile
electronic gadgets like laptops, getting their power from the batteries. They require
voltage levels not as quite as same as provided by the battery. They supply different
controlled voltages from a solitary variable battery voltage and in this way sparing
space as opposed to utilizing various batteries to supply various pieces of the gadget.
Rectifier: For the time period t = 0 to T /2, the diodes D2 and D3 conduct, whereas
D1 and D4 are in the state of no operation. For the other input cycle D1 and D4
are in conducting state and vice versa, thus resulting in maintaining the polarity of
load. Filter: We are utilizing a LPF to get 0 Hz DC voltage by utilizing capacitor
for a capacitive filter circuit associated at rectifier output resulting in DC across
filtered waveform and eventually is supplied to the load. When the state of charge
of the battery of electrical vehicle is above 80%, the circuit itself performs the buck
operation and converse of this; if it is less than 80%, it will perform boost operation
(Fig. 2).
5 Simulation
Electrical vehicle gains supply from the grid, and then, a CHB converter is placed
in the circuit which gives an option to apply a split storage for battery which can be
used as a backup for the charging station. This simulation works on basic operation
of buck or boost. When the battery percentage is lower than 80%, the boost operation
Fig. 2 Proposed circuit diagram
comes in action and starts charging the battery. If the battery percentage is higher than
80%, then operation takes place is buck operation [4, 5]. Voltage as well as current
can be measured with the help of different functional blocks and can be visualized by
scope. Voltage and current measurement blocks are implemented at different stages
of the simulation, and this gives different measurements of the parameters (Figs. 3
and 4).
Final output is visualized in scope connected to the load (Fig. 5).
This is output of the battery of EV, and X-axis denotes time period and Y-axis
denotes the output voltage of the EV (Fig. 6).
Fig. 3 Simulation diagram of the proposed system
Fig. 4 Simulation of final output part

Fig. 5 Output voltage of battery of EV
Fig. 6 Output of controlled rectifier
This output is of the controlled rectifier in the circuit which is decrease or increase
the output voltage which completely depends on the battery charging operation
(Fig. 7).
This is state of charge of battery which depends on buck or boost operation. If
battery percentage is below 80%, then it will perform boost operation and above 80%
buck operation takes place. In the above figure, battery percentage is 100% (fully
charged). X-axis of the above graph in Fig. 7 is battery percentage, and Y-axis is time
taken to charge the battery (Fig. 8).
As we can observe that battery percentage is below 80%, so the operation is boost
operation and battery gets charged from grid (Fig. 9).
As there is some loss in the battery, that is why graph is slightly going down. X-axis
of the overhead graph is battery percentage, and Y-axis is time period (Fig. 10).
As we can observe that the battery percentage is above 90% so the operation will
be buck operation. The state of charge of the battery is 90% that is why required
operation is of low voltage that is to directly give supply to the electrical vehicle
battery (Fig. 11).
Fig. 7 State of charge of the battery
Fig. 8 Simulation settings for boost operation
Fig. 9 State of charge of boost operation

Fig. 10 Simulation settings of buck operation
Fig. 11 Output of buck operation
Battery is slowly getting charged toward its full potential. X axis of the above
graph is battery percentage and Y-axis is time period.
6 Hardware
We can observe that the battery supply voltage is increased due to the implementation
of CHB converter. This also allows integration of split battery storage. Input AC
transformer is connected to the power-controlled rectifier converts AC to DC, and
then, we are using capacitor to obtain the pure DC. Based on the duty cycle, it acts
as a buck or boost converter. Then, we are connecting the 12 V EV battery which
will get charged. Then, the half-bridge inverter will convert the DC to AC and give it
to isolation transformer or 1:2 transformers which is connected to the output circuit.
The other two are driver circuits being used for isolation and amplifying the input
pulses. We are generating the PWM pulses using PIC controller. The PWM pulses
are given as input to the driver circuits. Optocoupler IC TL250 is used for isolating
the main circuit and control circuit to prevent the main circuit components from
burning due to overvoltage. We are using the Darlington pair transistor to amplify
the input pulses which is connected to each device. Driver circuit: We are using the
driver circuits for impedance matching to provide the main circuit from burning and
provide buffer to the input voltage. PWM pulses are sent to driver circuits to open
the gates for driver circuits. The battery we are using here is lithium-ion battery. The
specialty of this battery is that it is rechargeable. This battery is used in portable
electronics and EVs and is also finding its application from military to aerospace
fields, etc. In lithium ion batteries the positive electrode is made up of lithium and
negative electrode is made up of graphite. During discharging process the ions move
from negative to positive electrodes through electrolyte, whereas in charging process
it is opposite. These batteries consist of high energy densities and low self-discharge
and no memory effect. We are intending of using a 12 V lithium-ion battery for
our project. This battery will be charged by our proposed battery charging system
(Figs. 12 and 13).
Fig. 12 Hardware
connection
Fig. 13 Hardware input voltage
Hardware input voltage is measured with the help of a multi-meter. We have used
four transformers of 15VA rating each. Driver circuits plays an important role in
achieving the target to charge the battery (Figs. 14 and 15).
The name nodeMCU is a combined form of “node” and “MCU (microcontroller
unit)”. It is a prototyping board which can be open sourced. It uses Lua scripting
language. Hardware used is a circuit panel working as a DIP adding a USB with a
smaller panel consisting of antenna and MCU. This nodeMCU will be connected
to the CT and PT which will measure different parameters of the battery [1]. This
will be uploaded to ThingSpeak cloud and can be accessed by anyone via Internet
(Fig. 16).
The output of the battery is uploaded to the cloud via IOT, and this can be accessed
through ThinkSpeak page on your respective browsers. This is accessible worldwide.
Fig. 14 Hardware battery input voltage

Fig. 15 NodeMCU
Fig. 16 Things peak cloud
7 Conclusion
Projects consisting battery controlling methods for the operation generally use more
amount of cells to attain required voltage. As there is a lot of lack of space, there are
restrictions of using such more amount of batteries. By making use of several boost
converters, we can decrease the amount of cells required and also attain the high
voltage for the application. In today’s world, there are plethora of applications which
requires the high voltage uses these converters to boost the voltage. Some of the best
examples are EVs, and there are plenty of EVs launching in the market nowadays as
seeking the brighter and healthier future. There are the results of experiments that
showed that using boost converter the appliances would need less cells and still can
attain higher voltage. Implementation of IOT is also a key role to this project which
makes it easier to maintain the battery system of the EVs. ThingSpeak cloud gives
an excess to the parameters measured by the sensors. The conclusion is that CHB
converter not only give the power supply to the battery but also provide a backup
to the battery charging. These also improve the fast charging. Implementing this
combination with IOT management system provides a complete automated charging
station.’ Without using high-powered chargers, we can charge several EVs simulta-
neously. All conceivable charging station working modes among with the planned
vital control capacities are investigated. The state-of-charge self-adjusting method
of the delta-associated CHB converter is additionally presented.
References
1. Arun Noyal Doss M, Ananthi Christy A, Krishnamoorthy R, Modified hybrid multilevel inverter
with reduced number of switches for PV application with smart IoT system. J Ambient Intell
Humanized Comput. https://doi.org/10.1007/s12652-018-1151-2
2. Haghbin S, Lundmark S, Alaküla M, Carlson O (2013) Grid-connected integrated battery
chargers in vehicle applications: review and new solution. IEEE Trans Ind Electron 60(2):459–
473
3. Lee J-Y, Chae H-J (2014) 6.6-kW onboard charger design using DCM PFC converter with
harmonic modulation technique and two-stage dc/dc converter. IEEE Trans Ind Electron
61(3):1243–1252
4. Amjadi Z, Williamson SS (2010) Power-electronics-based solutions for plug-in hybrid electric
vehicle energy storage and management systems. IEEE Trans Ind Electron 57(2):608–616
5. Yilmaz M, Krein PT (2013) Review of battery charger topologies, charging power levels,
infrastructure for plug-in electric and hybrid vehicles. IEEE Trans Power Electron 28(5):2151–
2169
6. Zhang H, Yu X, Braun PV (2011) Three-dimensional bicontinuous ultrafast-charge and -
discharge bulk battery electrodes. Nature Nanotech 6:277–281. www.nature.com/naturenan
otechnology
7. Battapothula G, Yammani C, Maheswarapu S (2019) Multi-objective optimal planning of
FCSs and DGs in distribution system with future EV load enhancement. IET Electr. Syst.
Transp. 9(3):128–139
8. Zou H, Manzie C, Neˇsi D (2015) PDE battery model simplification for charging strategy
evaluation. c 978-1-4799-7862-5/15/$31.00 © 2015 IEEE
9. Chellaswamy C, Nagaraju V, Muthammal R (2018) Solar and wind energy based charging
station for electric vehicles. Int J Adv Res Electr Electron Instrum Eng 7(1):313–324
10. Kang B, Gerbrand (2009) Battery materials for ultrafast charging and discharging. Ceder1
Nature 458
11. Rishav M, Maity R, Shiva D (2019) Integration of renewable resources such as wind for electric
vehicle and implementation of smart electric vehicle using IOT 8(1s4). ISSN: 2277–3878
12. Vasiladiotis M, Rufer A (2015) A modular multiport power electronic transformer with inte-
grated split battery energy storage for versatile ultrafast EV charging stations. IEEE Trans Ind
Electron
Implementation of TLBO Optimised
PIλ Dμ Controller for LFC of Nonlinear
Reheat Thermal Power System
Nimai Charan Patel, Binod Kumar Sahu, and Ramesh Chandra Khamari
Abstract This article explains the load frequency control (LFC) of dual area inter-
linked reheat type thermal plant including nonlinearities such as generation rate
constraint (GRC) and governor dead band (GDB) by employing fractional order
proportional integral derivative (PIλ Dμ ) controller. A PID controller is also sepa-
rately implemented for performance comparison. The parameters corresponding to
both the controllers are fine-tuned by application of a teaching learning based opti-
misation (TLBO) algorithm using integral time absolute error (ITAE) as the fitness
function. A comparative performance analysis is accomplished between the two
types of controllers by injecting an instantaneous load of 10% in area-1. It is seen
that the TLBO optimised PIλ Dµ controller yields better transient response when
compared with the other controller. Lastly, the robustness of the TLBO optimised
PIλ Dμ controller is verified by applying a randomly varying step load in area-1.
Keywords Load frequency control · Area control error · PIλ Dμ controller ·

Objective function · Teaching learning based optimisation · Transient response
1 Introduction
Frequency control is a key aspect in interconnected power systems. It is imperative to

uphold the system frequency within the prescribed limits so as to preserve the stability
of the power system as well as to maintain the power quality at the consumers’
premises. The frequency of the system is basically a function of the real power
demand on the system [1–3]. Practically, the real power demand is random and
dynamic in nature. Therefore, a suitable secondary controller is used to regulate
the generation of power according to the active power demand so as to uphold the
frequency within a tolerable band and this method of achieving the balance between
demand and generation is known as LFC.
N. C. Patel (B) · R. C. Khamari

Government College of Engineering, Keonjhar, Odisha, India
B. K. Sahu
Siksha ‘O’ Anusandhan University, Bhubaneswar, Odisha, India
https://doi.org/10.1007/978-981-16-4943-1_63
672 N. C. Patel et al.
The first ever work on LFC was carried out by Cohn in 1956 [4]. Since then,
various research works on LFC have been carried out. Elgerd and Fosha proposed
an optimum LFC method for multi area power system in 1970 [5]. Many control
strategies for LFC including the application of various artificial intelligence tech-
niques along with the conventional controllers have been reported in the literature.
Design of PI controller with fuzzy gain scheduling approach is described for LFC
of four area interlinked system [6]. Performance comparison of various classical
controllers for LFC has been outlined in [7]. Implementation of differential evolu-
tionary (DE) algorithm tuned 2-DOFPID controller has been illustrated for LFC of
dual area interlinked thermal plant taking the effect of GDB into account [8]. LFC
of two area solar-thermal system by invasive weed optimisation (IWO) algorithm
tuned multi staged PID controller is explained in [9]. LFC analysis of power systems
having different generation sources has been carried out by implementing whale
optimisation algorithm (WOA) tuned fuzzy integrated PI (FIPI) controller [10]. Ant
lion optimiser (ALO) designed fractional order fuzzy PID (FOFPID) controller is
presented for controlling the frequency of two area interlinked power systems having
different generation sources [11]. PD and FOPI controller are combined together and
optimised by salp swarm algorithm (SSA) for LFC of dual area multi-unit system
[12]. LFC analysis of thermal plant with electric vehicle (EV) aggregators having time
varying delay has been explained in [13]. Cascading of fractional order controller
for LFC study has been illustrated in [14].
In this study, a dual area interlinked nonlinear thermal plant having reheat turbine
is considered for LFC analysis with PIλ Dμ and PID controller. In the first step, the
PID controller is fine-tuned separately by the application of TLBO algorithm to study
the dynamic response of the considered system. In the next step, a TLBO optimised
PIλ Dμ controller is employed for further improvement of system performance and
stability. In all the cases, a step disturbance of 10% is imparted in area-1 to analyse
the transient response of the system. Finally, the robustness of the recommended
TLBO optimised PIλ Dμ (TLBO–PIλ Dμ ) controller is verified by the application of
a random load.
2 System Modelling
The model of the power system is sketched in Fig. 1. It has two control areas that are
interlinked via the tieline. The tieline facilitates the power interchange among the
two control areas. Each control area consists of a thermal unit with reheat turbine.
Effect of generation rate constraint (GRC) and governor dead band (GDB) is also
considered in designing the power system model to make the system more practical
and realistic. In the present work, limiting value of GDB has been taken as 0.036
and the GRC has been taken as 3%/minute. Inclusion of GDB and GRC makes the
power system highly nonlinear and therefore, selection and design of appropriate
controllers are vital to maintain the power system stability.
Area control error (ACE) of every area is given by the following equations.
Implementation of TLBO Optimised PIλ Dμ Controller … 673
Fig. 1 Power system model under investigation
ACE1 = B f 1 + Ptie (1)
ACE2 = B f 2 + Ptie (2)
where, ACE1 and ACE2 denote the area control errors in area-1 and area-2 respec-
tively. B is the frequency bias factor, f 1 and f 2 represent the frequency changes
in area-1 and area-2 respectively and Ptie denotes the tieline power change. ACEs
of every area serve as the inputs to the controller of the respective areas. Parameters
of the model shown in Fig. 1 are depicted in Appendix-1 [13].
3 Proposed Control Approach
The various controllers used in this work are the PID and PIλ Dμ controller. The
description of these controllers is outlined below.
3.1 PID Controller
PID controllers are very popular controllers due to their simple design and robustness.
Also, it delivers better dynamic response and therefore, PID controller has always
been the 1st choice of engineers for many decades. Schematic arrangement of PID
controller is sketched in Fig. 2.
Fig. 2 Configuration of PID

controller
The time-domain output and the transfer function (TF) of a PID controller are
expressed as given below.
t
d
u(t) = A P e(t) + A I e(t)dt + A D e(t) (3)
dt
0
AI
G(s) = A P + + ADs (4)
s
where, A P , A D and A I are the proportionality gain, derivative gain and integral gain
respectively.
3.2 PIλ Dμ Controller
In PID controllers, the order of the derivative and integral (μ and λ) are integers
and often taken as 1. But the order of the derivative and integral (μ and λ) in a
PIλ Dμ controller may take the value of any positive real number. Thus, there are two
extra parameters μ and λ in a PIλ Dμ controller which provides additional flexibility
in adjusting the system dynamics and hence the PIλ Dμ controller delivers enriched
performance than the PID controller. A specific instance of PIλ Dμ controller is the
PID controller, where μ = λ = 1. Configuration of the controller is sketched in
Fig. 3.
The output for this controller is expressed by the Eq. (5) and its transfer function
is given by the Eq. (6).
d−λ dμ
u(t) = A P e(t) + A I e(t) + A D e(t) (5)
dt −λ dt μ
AI
G(s) = A P + + AD sμ (6)
sλ
Fig. 3 Configuration of
FOPID controller
3.3 Optimal Design of Controllers
The time response of the power system is greatly dependent on the parameters of
the controller and hence the parameters of the controller must be properly selected
by using suitable optimisation techniques. Optimisation technique employs a func-
tion known as the objective function or fitness function denoted by J and then the
optimisation problem is formulated by minimisation of this objective function under
the constraints of lower and upper limits of the parameters to be optimised. Various
objective/fitness functions for LFC analysis of the power system are mentioned in
the literature. In the present work, ITAE is selected as the objective/fitness func-
tion. ITAE is expressed by the Eq. (7) and the optimisation problem is structured as
minimisation of J under the restrictions described by the Eqs. (8–10).
ts
J = ITAE = (| f 1 | + | f 2 | + |Ptie |)tdt (7)
0
A P,D,I min ≤ A P,D,I ≤ A P,D,I max (8)
λmin ≤ λ ≤ λmax (9)
μmin ≤ μ ≤ μmax (10)
where, A P,D,I max and A P,D,I min are respectively the upper and lower values of the
proportional, derivative and integral coefficients of the PID and PIλ Dμ controller
and taken as 5 and 0.01, λmax and λmin are respectively the upper and lower values
of the order of the integral in PIλ Dμ controller and taken as 1 and 0.01, μmax and
μmim are respectively the upper and lower values of the order of the derivative in
PIλ Dμ controller and taken as 1 and 0.01. Various optimisation techniques have
been mentioned in the literature for solution to different optimisation problems in
engineering and appropriate selection of the optimisation technique is vital for a given
problem otherwise it may lead to unsatisfactory results. After a careful examination
of the literature, TLBO is chosen in this work to solve the optimisation problem so
as to get the optimum parameters of the controllers.
3.3.1 Teaching Learning Based Optimisation (TLBO):
TLBO was developed by Rao et al. in the year 2011 [15]. It is a metaheuristic popu-
lation oriented search algorithm stimulated by the teaching and learning procedure.
This algorithm simulates the effect of teacher’s impact on the outcome of students
in a particular class. The outcome of a student is assessed by the result or grades
secured by the student. The teacher is considered to be highly educated person than
the students as he shares his experience and knowledge with the students and the
quality of the teacher considerably influences the output of the students. A quality
teacher imparts training on the students in such a way that the outcome of the students
is improved. The students can further enhance their knowledge and produce good
results by interaction among themselves. The working method of TLBO is split into
two phases. The first phase is the teacher phase which comprises of learning from
the teacher and the second phase is the learner phase which consists of learning by
interactions between the learners or students. Various steps of TLBO algorithm are
described below.
(i) Initialisation: Initially a population with size [NP× D] is generated randomly,
where NP represents the number of all learners and D represents dimension
of search space denoting the number of various subjects provided.
(ii) Teacher phase: Each teacher puts his best effort to enhance the mean class
result in his assigned subject. The teacher is considered to be the highly
educated person as he trains the learners and therefore, he is taken to be
the best learner. Hence, the best solution (X best ) is found out and designtated
as teacher. The average or mean value of the marks secured in each subject
by different students is computed as:
Md = [m 1 , m 2 , ..., m D ] (11)
The difference between mean value of the grades in a given subject and the
grade of respective teacher is determined by:
Mdiff = rand × [X best − TF Md ] (12)
where, TF denotes the teaching factor, rand denotes the random number within
0 and 1. The value of TF maybe 1 or 2 and randomly decided using the
equations given below.
TF = round(1 + rand) (13)
where, rand denotes a positive real random number of value less than 1.
The existing population gets updated by using the following equations.
X new = Mdiff + X (14)
Elements of X new are retained if J (X i,new ) < J (X i ) otherwise elements

of X new are replaced by corresponding elements of X . Here, J denotes the
objective function as defined by Eq. (7).
(iii) Learner phase: In this step, a learner randomly selects another learner and
tries to enhance his knowledge through mutual interactions with the other
learner. The learner enhances his knowledge through mutual interaction when
the other learner is more knowledgeable than him. The procedure of learning
in this step is described by the equations given below.
X new = rand(X i − X j ) + X i , if J (X i < X j ) (15)
Otherwise,
X new = rand(X j − X i ) + X i (16)
where, X i and X j are two arbitrarily selected learners and i = j.

If X new performs better than it is kept. Steps (ii) and (iii) will be repeated till
the stopping criterion is arrived and lastly, the elements with the best solution are
retained.
4 Result Analysis
The considered power system model is depicted in Fig. 1. Different model parameters
are depicted in Appendix-1. The power system is modelled and simulated in the
MATLAB and Simulink environment. The controllers employed for the LFC analysis
of the considered power system are PID and PIλ Dμ controllers. The control action
takes place in the governor of the thermal plant of each area. TLBO algorithm is
scripted in them file and integrated with the model under investigation for optimal
tuning of the controllers. The numbers of population are taken as 50 (i.e. NP = 50)
and the numbers of iteration are taken as 500 in each case.
4.1 Transient Performance Analysis
An abrupt disruption of 0.1 p.u. is imparted in area-1 and the transient system
response is examined independently by implementation of TLBO tuned PID and
PIλ Dμ controllers. The tuned parameters of both the controllers obtained by the
injection of an abrupt disturbance of 10% in area-1 are given in Table 1. The tran-
sient response of tieline power oscillations and frequency oscillations in both areas
with both controllers are depicted in Figs. 4, 5 and 6 and the corresponding transient
response specifications with ITAE values for a simulation time of 50 s are depicted
in Table 2.
By careful examination of ITAE value and transient performance specifications
given in Table 2 and transient responses shown in Figs. 4, 5 and 6, it can be inferred that
the TLBO optimised PIλ Dμ (TLBO-PIλ Dμ ) controller delivers superior performance
than the TLBO designed conventional PID (TLBO-PID) controller. Comparing the
performance of both the controllers, it is evident that, apart from a slight increase in
peak undershoot (Ush ) in tieline power oscillation and a slight increase in overshoot
(Osh ) in area-1 frequency oscillation, the TLBO-PIλ Dμ controller exhibits better
transient performance than the TLBO-PID controller. The disturbance injected into
the system is 10% which is significantly large. The system considered in this work
is extremely nonlinear due to the inclusion of GDB and GRC and yet the magni-
tude of the disturbance injected is considerably large. Considering the above facts,
the settling time of the system response is crucial and it is important to observe
that the settling time (TS ) in case of TLBO-PIλ Dμ controller is considerably less
as compared to the TLBO-PID controller. Further, considering the effect of large
injected disturbance and system nonlinearities, the slight increase in peak under-
shoot in tieline power oscillation and the slight increase in overshoot in area-1
frequency oscillation in the case of TLBO-PIλ Dμ controller can be ignored because
finally the TLBO-PIλ Dμ controller gives very less ITAE value than the TLBO-PID
controller. Hence, it is concluded that the recommended TLBO-PIλ Dμ controller
delivers superior performance than the TLBO-PID controller.
4.2 Robustness Analysis
The robustness of the recommended TLBO-PIλ Dμ controller is verified with the

existing controller parameters as depicted in Table 1 by varying the different time
constants of the system over a range of −25 to 25% of the nominal value. The time
constants varied are (i) time constant of the governor (Tg ), (ii) time constant of the
turbine (TC ) and (iii) time constant of reheat (Tr ). The frequency deviations in both
areas ( f 1 and f 2 ) along with the tieline power oscillation (Ptie ) due to variations
of different time constants are depicted in Figs. 7, 8, 9, 10, 11, 12, 13, 14 and 15. It is
obvious from Figs. 7, 8, 9, 10, 11, 12, 13, 14 and 15 that the controller is proficient
in retaining the system stability even with the variations of different time constants
over a wide range.
Table 1 Optimum parameters of the TLBO optimised controllers
Controllers Controller parameters of area-1 Controller parameters of area-2
A P1 AI 1 A D1 λ1 μ1 A P2 AI 2 A D2 λ2 μ2
TLBO-PID 3.4573 0.9626 0.1521 – - – - 0.2911 0.0100 0.0100 – - – -
TLBO-PIλ Dμ 4.6583 5.0000 5.0000 0.9556 0.1591 4.4126 1.3766 1.3006 0.2420 1.0000
Implementation of TLBO Optimised PIλ Dμ Controller …
679
Fig. 4 Frequency oscillations in area-1
Fig. 5 Frequency oscillations in area-2
5 Conclusion
In this paper, a dual area reheat type nonlinear thermal plant is modelled and PID
controller along with PIλ Dμ controller are implemented for load frequency control.
Fig. 6 Tieline power oscillations
The designed system model includes the effect of nonlinearities like GDB and GRC.
The controller parameters are optimised by employing TLBO algorithm. A step
disruption of 10% is applied in area-1 to examine the transient system response with
various controllers. It is found that the proposed TLBO-PIλ Dμ controller gives the
least ITAE value and delivers supreme performance when compared to the TLBO-
PID controller. Lastly, the robustness of the proposed controller is verified by varying
the different time constants of the system over a wide range.
682
Table 2 Transient performance specifications due to application of abrupt load in area 1

Controllers f1 f2 Ptie ITAE
Ush × 10−3 Osh × 10−3 TS Ush × 10−3 Osh × 10−3 TS Ush × 10−3 Osh × 10−3 TS
TLBO-PID −12.3170 3.1293 27.8259 −9.6740 5.3249 33.7908 −45.6303 7.6706 45.3922 3.2452
TLBO-PIλ Dμ −12.0043 8.8329 14.8258 −3.8581 1.3620 17.9315 −50.8009 1.7980 31.1909 1.5222
N. C. Patel et al.
Fig. 7 Frequency oscillations in area-1 due to variation of Tg
Fig. 8 Frequency oscillations in area-2 due to variation of Tg

Fig. 9 Tieline power oscillations due to variation of Tg
Fig. 10 Frequency oscillations in area-1 due to variation of TC

Fig. 11 Frequency oscillations in area-2 due to variation of TC
Fig. 12 Tieline power oscillations due to variation of TC

Fig. 13 Frequency oscillations in area-1 due to variation of Tr
Fig. 14 Frequency oscillations in area-2 due to variation of Tr

Fig. 15 Tieline power oscillations due to variation of Tr
Appendix-1
Investigated system parameters are given below.

B = 21, R = 1 11, FP = 1 6, Tr = 12, TC = 0.3, Tg = 0.2, D = 1, M = 8.8 .
References
1. Kundur P, Balu NJ, Lauby MG (1994) Power system stability and control, vol 7. McGraw-Hill,
New York
2. Elgard OI (1982) Electric energy systems theory. McGraw-Hill, New York, pp 299–362
3. Saadat H (1999) Power system analysis. McGraw-Hill
4. Cohn N (1956) Some aspects of tie-line bias control on interconnected power systems. Trans
Am İnst Electric Eng. Part III: Power Appar Syst 75(3):1415–1436
5. Elgerd OI, Fosha CE (1970) Optimum megawatt-frequency control of multiarea electric energy
systems. IEEE Trans Power Appar Syst 4:556–563
6. Chang CS, Fu W (1997) Area load frequency control using fuzzy gain scheduling of PI
controllers. Electric Power Syst Res 42(2):145–152
7. Nanda J, Saikia LC (2008) Comparison of performances of several types of classical
controller in automatic generation control for an interconnected multi-area thermal system.
In: Australasian universities power engineering conference. IEEE, pp 1–6
8. Panigrahi TK, Behera A, Sahoo AK (2017) Novel approach to automatic generation control with
various non-linearities using 2-degree-of-freedom PID controller. Energy Procedia 138:464–
469
9. Patel NC, Debnath MK, Sahu BK, Dash SS, Bayindir R (2019) Application of invasive weed
optimization algorithm to optimally design multi-staged PID controller for LFC analysis. Int J
Renew Energy Res 9(1):470–479
10. Patel NC, Debnath MK (2019) Whale optimization algorithm tuned fuzzy integrated PI
controller for LFC problem in thermal-hydro-wind interconnected system. Applications of
computing automation and wireless systems in electrical engineering. Springer, Singapore, pp
67–77
11. Patel NC, Sahu BK, Bagarty DP, Das P, Debnath MK (2019) A novel application of ALO-
based fractional order fuzzy PID controller for AGC of power system with diverse sources of
generation. Int J Electric Eng Educ 2019:0020720919829710
12. Patel NC, Debnath MK, Sahu BK, Dash SS (2020) Automatic generation control using novel
PD Plus FOPI controller tuned by salp swarm algorithm. Artificial ıntelligence and evolutionary
computations in engineering systems. Springer, Singapore, pp 1–13
13. Patel NC, Sahu BK, Debnath MK (2019) Automatic generation control analysis of power
system with nonlinearities and electric vehicle aggregators with time-varying delay imple-
menting a novel control strategy. Turk J Electr Eng Comput Sci 27(4):3040–3054
14. Jena NK, Patel NC, Sahoo S, Sahu BK, Dash SS, Bayndir R (2020) Application of fractional
order cascaded controller for AGC study in power system integrated with renewable sources.
Int J Renew Energy Res 10(1):89–100
15. Rao RV, Savsani VJ, Vakharia DP (2012) Teaching–learning-based optimization: an optimiza-
tion method for continuous non-linear large scale problems. Inf Sci 183(1):1–5
Author Index
A C
Aashique Roshan, B., 93 Chandar, M. Krishna, 377
Abarna, R., 407 Chandra Babu, P., 107
Abinayalakshmi, B., 331 Charan Raj, U., 479
Adebiyi, Abayomi A., 431 Cheren, S. Elam, 507
Ahilan, T., 507 Cisse, El Hadji Ibrahima, 193
Aijaz, Md., 467 Cisse, Mouhamadou Lamine, 123
Ajay Daniel, J., 517
Anand, G., 93
Ananthi Christy, A., 659 D
Anitha, M., 457 Davidson, Innocent E., 417
Deepika, T. J., 291
Anjuka, G., 407
Deva Priya, K., 407
Arul Stephen, C., 397
Devaraj, D., 137
Arumbu, V. P., 247
Devineni, Gireesh Kumar, 81
Aruna Priya, P., 31 Dey, Aindrila, 625
Aruna, S. B., 527 Dhanalakshmi, Samiappan, 71
Arun Noyal Doss, M., 659 Dhang, Moumita, 625
Ashmin Sugaji, J., 341 Dhayalini, K., 603
Avirajamanjula, P., 491, 571 Dieng, Biram, 123, 193
Divya Bharathi, M., 303
Divya Navamani, J., 183, 387
Divya, P., 247
B
Dominic Savio, A., 237, 603
Balaji, B., 213
Durga Shree, S., 227
Balaji, C., 237
Balaji, S., 517
Balamurugan, P., 635 E
Balan, Vinoth Kumar, 491, 571 Ebenezer Abishek, B., 397
Bharanidharan, R., 11 Eswaran, T., 613
Bharatiraja, Chokkalingam, 49 Ezhilarasan, G., 613
Bhatia, Kushagra, 387
Bhoopal, Neerudi, 81
Boopathi, C. S., 151, 165, 173 G
Brindha, R., 659 Ganesan, S., 213
Buraimoh, Elutunji, 417 Ganesh, Aman, 81
© The Editor(s) (if applicable) and The Author(s), under exclusive license 689
to Springer Nature Singapore Pte Ltd. 2022
https://doi.org/10.1007/978-981-16-4943-1
690 Author Index
Ganesh, Viswanathan, 659 M

Ganesh, V. N., 517, 559 Maheshwari, Utkarsh, 369
Gayathri, M., 593 Maheswari, R. V., 11, 303, 341
Geetha, A., 151, 165, 173, 387 Malathy, C., 593
Gnanavadivel, J., 331 Manivannan, S., 559
Goel, Piyush, 369 Mariammal, K., 353
Gopalakrishnan, R., 603 Marimuthu, Gajendran, 537
Goyal, Saksham, 369 Marimuthu, Marikannu, 21
Mathesh, R., 397
Mukherjee, Utsahan, 445
H Muralidharan, S., 331
Hannah Pauline, S., 71 Muthamizhan, T., 467, 613
Hari Akhilesh, C., 593
Harishankar, S., 549
Hemakesavulu, O., 237 N
Narayan, M. Adhitya, 353
Navaneethan, Nithya Rani, 603
I Nirmala, M., 227
Iruthayarajan, M. Willjuice, 377 Nithya, S., 257
Iswarya, S., 407
O
J Ojo, Evans E., 431
Jagadeesh Kumar, M., 613
Jagan Mugesh, M. S., 479
Jane Tracy, T., 323 P
Jayachandran, M., 205 Padhi, Tanmay, 183
Jayakumar, J., 291 Padmini, S., 313
Jeyaraman, T., 377 Palanisamy, R., 247
Joelpraveenkumar, D., 377 Paranthagan, B., 21
Josephine, R. L., 549 Parimala Devi, M., 257
Joy, Helen K., 1 Patage, Vinay Vasanth, 369
Patel, Nimai Charan, 671
Penchalaiah, Gajana, 499
K Prasanth, B. Venkata, 107
Kalaiarasy, C., 205 Preetha Roselyn, J., 137
Kaliraj, M., 377 Premalatha, L., 283
Kanakaraj, P., 579 Priyadharsini, S., 39
Kannan, C., 39 Puri, Muskan, 537
Karthikeyan, B., 247
Karthikeyan, D., 247
Khamari, Ramesh Chandra, 671 Q
Kounte, Manjunath R., 1 Queen, Hephzibah Jose, 291
Krishnakumar, C., 407
Kumar Chinnaiyan, V., 61
Kumari, Aditi, 183 R
Kumar, M. Yoogesh, 549 Raghavendran, C. R., 137
Kumar, R., 71 Ragupathy, U. S., 61
Kumudham, R., 479 Rajendran, V., 479
Rajeshwar, L., 93
Raju, Aishwarya, 445
L Ramesh Babu, P., 271
Lavanya, A., 183, 387 Ramesh, L., 579
Lazarus, Ian J., 431 Ramprasath, S., 407
Author Index 691
Ramya, G., 507, 517 T

Ramya, R., 151, 499 Tanweer Ahamed, S., 271
Rao, D. S. N. M., 81 Thakur, Ashay Kumar, 537
Rathinam, P. Subash, 353 Thamizh Thentral, T. M., 165, 173
Rathina Prabha, N., 323 Thentral, T. M. Thamizh, 151
Rathnavel, P., 613 Thiam, Ababacar, 193
Ravi Eswar, Kodumur Meesala, 49 Thyagarajan, T., 93
Ravindran, M., 341 Tiwari, Sourabh, 369
Rayaguru, N. K., 647
Ray, Rangit, 625
Robert, Femi, 537 U
Udaya Simha, R., 271
Usha, S., 151, 165, 173
Uthirasamy, R., 61
S Uthra, R Annie, 369
Sadees, M., 445
Saha, Akshay K., 431
Sahu, Binod Kumar, 671 V
Sakthi, S., 237 Vaithilingam, C., 625, 635
Santha Kumar, J., 165 Vasantha Kumar, S. K., 21
Sarr, Marie Pascaline, 193 Vengatesh, S., 271
Sathish Kumar, P., 479 Venkatesh, Banala, 313
Savio, A. Dominic, 571 Venkateswarlu, T., 467
Sekar, S., 647 Venkat, L., 397
Venugopal, R., 21
Selvarasu, Ranganathan, 39
Vignesh Babu, A., 517
Senthil Kumar, N., 635
Vigneshwar, P., 271
Senthil Kumar, S., 303
Vijayakumar, K., 173, 387, 445
Senthil Nayagam, V., 283
Vijayalakshmi, A., 397
Sewchurran, Sanjeeth, 417 Vijayalakshmi, K., 257
Shiferaw, Dagne Alemayehu, 39 Vijayalakshmi, Subramanian, 21
Shyam Balaji, R., 93 Vijayavelu, S. S., 353
Sivaperumal, P., 237 Vijay, V., 271
Soundarajan, R., 21 Vimala, C., 31
Srinithi, S., 21 Vinoth, Jayakumar, 49
Subramani, C., 151, 165 Vinothkumar, B., 237
Suchitra, D., 527 Vishnu Kumar, S., 61
Suganya, R., 457
Sujatha, P., 107
Sundararaju, K., 247 Y
Suresh, P., 507 Yuvaraj, B., 21

Proceedings of International Conference On Power Electronics and Renewable Energy Systems

Uploaded by

Copyright:

Available Formats

Proceedings of International Conference On Power Electronics and Renewable Energy Systems

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Proceedings of International Conference On Power Electronics and Renewable Energy Systems

Uploaded by

Copyright:

Available Formats

Lecture Notes in Electrical Engineering 795

More information about this series at https://link.springer.com/bookseries/7818

Brayima Dakyo Subhransu Sekhar Dash

ISSN 1876-1100 ISSN 1876-1119 (electronic)

Kattankulathur, India Prof. K. Vijayakumar

Dr. T. R. Paarivendhar, Founder Chancellor

Dr. C. Muthamizhchelvan, Vice Chancellor, SRMIST

Dr. Ramazan Bayindir, Gazi University, Turkey

Dr. Valentina Emilia Balas, UOA, Romania

Prof. K. Vijayakumar, SRMIST

Dr. A. Rathinam, SRMIST

Dr. Y. Jeyashree, SRMIST

International Advisory Committee

Dr. Akhtar Kalam, Victoria University, Australia

Dr. C. A. Vaithilingam, Taylor University, Malaysia

National Advisory Committee

Dr. Kumar Pradhan, IIT Kharagpur

Paper Selection and Publication Process

Dr. A. Geetha, SRMIST

Dr. T. M. Thamizh Thentral, SRMIST

Deep CNN Depth Decision in Intra Prediction . . . . . . . . . . . . . . . . . . . . . . . . 1

Structural Optimization of LMS Adaptive Filter Using Multi-stage

Power-Domain NOMA for Massive Connectivity in Smart Grid

Design of Quadratic Boost Converter for Renewable Applications . . . . . . 331

An Intelligent Fuzzy Controlled Microgrid Fed by Distribution

Performance Optimization of Stepper Motor Using MATLAB . . . . . . . . . 603

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689

Dr. C. Subramani received Master of Engineering

Prof. K. Vijayakumar has received B.E. degree in

Ph.D. and numerous postgraduate students. He has been

Prof. Brayima Dakyo received Engineer Dipl. (1984)

Dr. Subhransu Sekhar Dash is presently Professor

He is also one of the editors of few of the books titled

Helen K. Joy and Manjunath R. Kounte

Keywords Deep learning · Video compression · HEVC · CNN

Various compression methods became so prominent when the transfer of data at a

1.1 Evolution of Video Compression Development

“Motion JPEG (M-JPEG or MJPEG) is an initial video compression technique in

1.2 Evolution of Video Compression Development

1.3 Evolution of Video Compression Development by SMPTE

“SMPTE 421 M technology was developed with contributions from a number of

2 Intra Prediction in Video Compression

building a model, Here classification algorithm with logistic regression. Lu et al.

3 Data Set and CTU Splitting

3.1 Idea of CTU Splitting

Fig. 1 Concept depth decision algorithm using CNN

H3(x i,j) with i,j = 0–3 and can be mathematically represented as

3.2 Deep Depth Decision Algorithm Using Deep Learning

Fig. 2 Deep depth decision algorithm with CNN

Fig. 3 Bit rate comparison

Fig. 4 Time for encoding

Table 1 Bitrate comparison table for various resolutions

R. Bharanidharan and R. V. Maheswari

Abstract In the power system, the occurrence of a short circuit in transmission

Keywords 11 kV polymeric insulator · Finite element method · Electric field

R. Bharanidharan (B) · R. V. Maheswari

2 Finite Element Method of Analysis

Electric field = −∇u. (1)

The Cartesian system of coordinates of two-dimensional function F(V ) is