Parameter identification of permanent magnet synchronous AC motor based on spectral method

Abstract

This paper establishes a mathematical model with random terms for a permanent magnet synchronous AC motor, which is a nonlinear system using d - axis current, q - axis current and the rotor electric angular speed as state variables. The numerical solution obtained by the Euler-Maruyama method is used as the measurement data. Aiming at the parameter identification of the system, a step-by-step identification method based on the spectral method discretize the system first and then using the least squares is proposed. This method is used to identify multiple parameters of the motor in the same model. In step-by-step identification, firstly by fixing the motor speed, the system is transformed into a linear system, which is used to estimate the resistance, inductance and flux linkage. After that, the speed is not fixed, for the electrical parameters are known, we can identify damping and inertia by using mechanical equations. Finally, the experimental results show that the relative errors of the parameters identified by the proposed method are smaller, which shows the effectiveness of this method for multi-parameter identification.