Intelligent Parameter Identification of PMSM Based on BPNN Fitting Nonlinear Relationships

In this paper, an intelligent parameter identification algorithm of the permanent magnet synchronous motor (PMSM) is proposed. Firstly, a high-fidelity motor parameter model considering cross coupling effects is constructed using a polynomial model to accurately describe the magnetic saturation scenario, which is difficult to achieve with traditional motor equations. Secondly, an improved- backpropagation neural network (BPNN) is used to identify the coefficients in this model by fitting explicit time-domain relationships, effectively avoiding a large amount of direct calculations. In this framework, the accurate identification of multi-dimensional parameters is realized by changing the definition of loss function, combining backpropagation and making full use of the nonlinear fitting ability of neural network. Finally, Simulations and experiments verify the accuracy and convergence speed of parameter identification.