Uniform demagnetizing fault detection in permanent magnet synchronous motorbased on variational mode decomposition and machine learning

With the application of Permanent Magnet Synchronous Motor (PMSM) more and more widely, people pay more attention to its safety and stability. The operation of PMSM is complicated and it is more prone to failure, among which demagnetization fault is the unique fault type of PMSM. In this paper, the demagnetization fault detection strategy of PMSM based on variational mode decomposition (VMD) and machine learning decision tree is proposed. The variational mode decomposition (VMD) algorithm is used to decompose fault characteristic signals of different degrees. The characteristic values under normal operation of the motor are taken as thresholds, and the characteristic values and thresholds under different fault degrees are compared to judge the motor faults. The data set obtained after VMD decomposition is trained by machine learning, so as to diagnose and classify demagnetization faults of different degrees. And compared with the diagnostic results of the models using empirical mode decomposition (EMD) for data preprocessing and the models without data preprocessing, it is proved that the VMD algorithm can extract the fault characteristic values more accurately, and ensure the accurate identification of demagnetization faults of PMSM .