Fast parameter identification of permanent magnet synchronous motor for electric vehicles

Abstract

This paper designs a recurrent neural network (RNN) that can be used for both permanent magnet synchronous motor (IPMSM) parameter identification and optimal current (OC) solution for IPMSM operating efficiency optimization for electric vehicles (EVs). Firstly, the problem of parameter identification of IPMSM is modeled as a regression problem, and the least absolute deviation method (LAD) is used to estimate the parameters. Then the optimization theory and variational theory are adopted to convert it into a variational problem, and the projection dynamic equation (PDS) is utilized to obtain the solution. Finally, the RNN corresponding to the PDS is designed which can be multiplexed for the optimal solution, aims at achieving the motor parameter identification in parallel. This paper theoretically proves the convergence of the proposed projection dynamic equation, the convergence value and the identity of the PMSM parameter value to be estimated. The IPMSM drive system is built and simulated. The simulation results show that the proposed method identifies the motor parameters quickly and accurately, and it verifies the rationality and effectiveness of the proposed method.