Identification For PMSM Rotor Speed Based On Optimized Extended Kalman Filter And Load Torque Observer

Abstract

The performance of a double closed-loop vector control system for permanent magnet synchronous motor (PMSM) relies on the accurate estimation of rotor positions or motor speed. The extended Kalman filter (EKF) algorithm is used for estimating rotor speed in this paper. Aiming at the problems of the traditional EKF method such as time-consuming computation of covariance matrices, poor system robustness, and low precision of speed estimation when load torque disturbances exist, genetic algorithm (GA) is used to optimize the selection process of EKF covariance parameters. Also, a load state observer is introduced to produce compensation input which is used to accelerate the response of current loop combined with the speed regulator. The experiment results indicate that the method proposed shortens the selection time of covariance parameters and improves the accuracy of speed identification and ability against load disturbances.