As an important part of aero-generator, rectifier circuit has many problems, such as frequent faults and difficult maintenance. In order to realize the fault diagnosis of doubly salient electro-magnetic generator (DSEG) rectifier circuit, a fault diagnosis method based on longshort term memory (LSTM) network is studied. Firstly, the three-phase armature current signals of the generator under various fault modes are collected. Secondly, different signal processing methods are used to process fault signals to obtain fault characteristic information. Then, the obtained fault characteristic data are divided into training and test samples and input to LSTM network for fault classification. Finally, the diagnosis results are calculated and analyzed. The simulation and experimental results show that the proposed method has a good fault diagnosis effect.