Demagnetization Fault Diagnosis of the Permanent Magnet Motor for Electric Vehicles Based on Temperature Characteristic Quantity

• Published in: IEEE transactions on transportation electrification Vol. 9; no. 1; pp. 759 - 770
• Main Authors: Meiwei, Zhang, Weili, Li, Haoyue, Tang
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.03.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Air gaps; Back propagation (BP) neural network; Back propagation networks; Demagnetization; Eddy currents; Electric vehicles; Fault diagnosis; Finite element method; Motors; Neural networks; permanent magnet motor; Permanent magnet motors; Permanent magnets; Prediction models; Rotors; Simulation models; Synchronous motors; temperature; Temperature measurement; Thermodynamic properties; Torque
• ISSN: 2332-7782; 2577-4212; 2332-7782
• DOI: 10.1109/TTE.2022.3200927

Permanent magnet motors are widely used in the driving system of electric vehicles because of their high power density and small size. However, the permanent magnet motor is prone to demagnetization due to the complex operating environment of electric vehicles and characteristics of permanent magnet motors. To ensure the normal operation of vehicles, the demagnetization fault diagnosis of permanent magnets should develop. In this study, the thermal behavior of a permanent magnet motor under local demagnetization fault is discussed. Results show that the demagnetization of permanent magnets has a significant effect on the temperature of the motor; thus, the temperature is added into input signals of the permanent magnet demagnetization fault diagnosis. The healthy state of the permanent magnet is predicted by the back propagation (BP) neural network to realize the demagnetization fault diagnosis. First, the demagnetization test platform is built to test the performance of the motor, and the current, torque, and temperature of the motor are measured according to different demagnetization degrees. Second, the demagnetization simulation model of the motor is established by the finite-element method, and the air gap magnetic field, loss, and temperature of the motor are analyzed. Lastly, the temperature, current, speed, and torque signals are selected as input signals of the neural network model, and the demagnetization rate of the rotor is taking as the output signal. The neural network prediction model is established and trained, and the good generalization ability is obtained.