Parameter Identification of Permanent Magnet Synchronous Motor Based on Physics-Informed Neural Network

In this paper, the physical model of permanent magnet synchronous motor (PMSM) is established by considering the influence of nonlinear factors such as magnetic saturation and cross-coupling, and the relationship between $d q$-axis flux and current is established by polynomial fitting, so as to establish a universal motor high-precision model. Then the parameters in the motor model are identified by the physics-informed neural network (PINN). In this new framework, the complexity of the motor high-precision model can be changed according to the actual situation of the motor. At the same time, the excellent fitting ability of neural network is fully utilized in PINN, and the accuracy of parameter identification is effectively enhanced by combining with the physical model of the motor. Finally, the identification accuracy of PINN for motor high-precision model are verified by simulation and experiment.