Permanent Magnet Synchronous Motor Demagnetization Fault Diagnosis and Localization

This study proposes a demagnetization fault diagnosis method using fast Fourier transform (FFT), improved sparrow search algorithm optimized probabilistic neural network (ISSA-PNN), and a fault localization method based on cosine similarity (CS) for permanent magnet synchronous motor (PMSM) with a slot-pole ratio of 3/2 and its integer multiples. First, the reasons for choosing branch currents as diagnostic and localization signals are explained. Second, the FFT is utilized to extract the 0.25th, 0.75th, 1st, 1.25th, and 1.75th harmonics of the branch currents and combined with ISSA-PNN to diagnose demagnetization. The residual values of the branch currents are used to establish a sample bank of demagnetization fault localization data features. Calculate the CS between the PMSM real-time branch current residual values and the sample bank to locate faults. Finally, experiments and finite element analysis (FEA) have demonstrated that the method achieves 100% fault diagnosis accuracy and 97.22% fault localization accuracy.