Stator and Rotor Temperature Prediction of Permanent Magnet Synchronous Motor Based on BP Neural Network

The performance of motors is largely limited by temperature. Under conditions such as rapid acceleration conditions and low-speed cruising on long ramps, the windings in the stator assembly are susceptible to over-temperature and the permanent magnets are susceptible to demagnetization at high temperatures. Therefore, stator winding and rotor temperatures need to be monitored or predicted. In this paper, we propose a stator and rotor temperature prediction method for the permanent magnet synchronous motor (PMSM) based on back propagation (BP) neural network, firstly, a computational model of the PMSM is established based on the main parameters of the PMSM, and the PMSM losses are calculated and analyzed. Then, a three-dimensional field computational model of the PMSM is established, the temperature field simulation boundary conditions are solved, and the temperature field analysis is carried out. Moreover, the computational fluid dynamics (CFD) verifies the accuracy of the one-dimensional thermal network model. Finally, the technique uses the one-dimensional thermal network model to construct a training set and builds the PMSM stator and rotor temperature prediction model based on BP neural network training.