Aiming at the bottleneck of system performance optimization in Machine translation system, long-distance dependence, difficulty in processing low-frequency words and rare words, a Natural language processing Machine translation system based on Seq2Seq model is designed, which effectively realizes multilingual translation and language communication, realizes accurate translation of multiple languages, and improves the efficiency of language communication. Based on the Natural language processing technology of Seq2Seq model, this paper quickly recognizes and translates the source language text, outputs the target language text, and through training the model, selects the task of French translation into English for experiment. The attention weight shows that the system performs well in the translation task.