Abstract: AC contactor is widely used in various low-voltage control lines, thus the residual life prediction can greatly improve the operation stability of power control system. In order to solve the problem that the residual life of AC contactor is determined by the current state and the previous state, and the relationship between the state before and after the degradation process cannot be effectively used, a residual life prediction method of AC contactor based on long-short term memory neural network (LSTM) is proposed. Firstly, the degradation data of AC contactor in its whole life cycle are obtained through a life cycle test platform, and the characteristic parameters which can reflect its operation state are extracted. Secondly, gray correlation analysis (GRA) and Pearson correlation coefficient (PCC) are used to eliminate the redundant information of multi-dimensional parameters, and feature selection is used as the input sample of the prediction model. Finally, the LSTM prediction model is trained. The experimental results show that, compared with the traditional recurrent neural network (RNN), the residual life prediction model based on the improved LSTM can make full use of the correlation information of the whole life cycle time series data, and has higher accuracy for the residual life prediction of AC contactor.