Diagnosis of local demagnetization faults in PMSM based on CWT and radial leakage magnetic flux

This study proposes a method for diagnosing local demagnetization faults in Permanent Magnet Synchronous Motors (PMSMs) based on Continuous Wavelet Transform (CWT) and radial leakage magnetic signals. The approach involves utilizing CWT to process collected radial leakage magnetic flux from the faulty motor surface, extracting fault features effectively. Leveraging the advantages of the leakage magnetic signals, this method accurately identifies the magnitude of local demagnetization in the motor, providing a reference for assessing the severity of demagnetization. Finally, the ResNet50 convolutional neural network is employed for feature extraction from the Continuous Wavelet Transform images, enabling precise classification of different demagnetization magnitudes.