First, you need to download language models for simplified and traditional Chinese.

fanti.bin

jianti.bin

Alternately, you can build your own language models.

Install KenLM on your system and prepare Chinese corpora in simplified and traditional Chinese. (ex. jianti.txt, fanti.txt)

Then build the language models (names must be jianti.bin and fanti.bin):

$ perl -Mutf8 -CSD -F/\\t/ -nale 'tr/ /\x{e000}/; print join(" ", split(//, $_))' jianti.txt > jianti_space.txt
$ lmplz -o4 --prune 0 5 10 15 < jianti_space.txt > jianti.arpa
$ build_binary jianti.arpa jianti.bin

Now you can perform conversion:

$ echo "干燥 干部 干涉" | ./jianfan.sh
乾燥 幹部 干涉

$ echo "乾燥 乾坤" | ./fanjian.sh
干燥 乾坤

The results may vary depending on the corpora.