Demagnetization Fault Diagnosis Based on Feature Extraction and Convolutional Neural Network for Permanent Magnet Generator

During the operation of permanent magnet wind turbines, magnetic steel demagnetization faults may occur, directly affecting the normal operation of wind turbines and having adverse effects on wind power generation. This article proposes a demagnetization fault diagnosis method for permanent magnet generators based on feature extraction and convolutional neural network. A permanent magnet generator with a power of 25kW was used for demagnetization fault simulation experiments. The current signal of the generator was collected, and time-domain feature extraction, frequency-domain feature extraction, and singular value decomposition were performed on it. The different features extracted were trained through convolutional neural network to achieve pattern recognition of demagnetization faults and determine the operating status of the generator, achieving demagnetization fault diagnosis of permanent magnet generators.