Asynchronous motor fault diagnosis method based on BiLSTM network

Abstract Asynchronous motor stator short-circuit faults can pose a great threat to the production and operation of asynchronous motors. To identify the short-circuit faults and recover the losses in time, this paper establishes a model of a three-phase asynchronous motor in Simulink by using MATLAB, and simulations are carried out under various fault conditions. The time and frequency domain characterization of signals are extracted by fast Fourier transforms (FFT) and input into the bidirectional long short-term memory (BiLSTM), for training and learning. The experiments show that the overall accuracy of the method in terms of fault identification rate is 96.19%, and it provides an effective solution for the diagnosis of this fault.