The Classification Method of Electrical Faults in Permanent Magnet Synchronous Motor Based on Deep Learning

An accurate classification method is highly required in the development of a fault detection system. Various deep-learning techniques have recently been used for fault classification. However, optimally training deeper networks such as convolutional neural networks (CNN) on relatively few and non-uniform experimental data of electric machines is extremely difficult. The proposed classification method is based on wavelet and pre-trained convolution neural networks. This approach ensures the correctness of the diagnostic outcome while streamlining the testing procedure and does not require the collection of a lot of data. The continuous wavelet transform function converted the three-phase stator current signals to RGB images. It does not necessitate sophisticated signal processing methods or additional hardware detection equipment. This research will be useful in the development of an online fault detection system for engineering applications.