PMSM Interturn Short Fault Diagnosis Based on WPT and GWO-PNN

Real-time and accurate fault diagnosis for permanent magnet synchronous motor (PMSM) is crucial because interturn short fault (ISF) can disrupt its normal operation. This paper establishes different levels of ISF in Maxwell software to obtain PMSM’s operation data under various working conditions. Wavelet packet transform (WPT) is employed to denoise the original signal, and the preprocessed data is trained using a probabilistic neural network (PNN). By optimizing the PNN with grey wolf optimizer (GWO), a GWO-PNN model is obtained. The experiment shows that the GWO-PNN has higher diagnostic accuracy than PNN.