Abstract: In response to issues such as the inability to guarantee the security of commercial machine translation engines, low translation accuracy in professionalization fields, and single translation service method, this paper designed a rail transit machine translation system - "Tieyitong" based on the Transformer model, through technologies such as neural network models, knowledge distillation methods, and professional corpus construction, to implement diversified functions such as text translation, document translation, Office plugin translation, and browser plugin translation. The application results show that the system can provide professionalization, diversified, customized, and strong security machine translation services for industry users.