关键词: 永磁同步电机, 扩展卡尔曼滤波, 转动惯量辨识, 负载转矩辨识

Abstract: In response to the contradiction between convergence speed and filtering effect in the online identification of
mechanical parameters of permanent magnet synchronous motors using the extended Kalman filter ( EKF) , a new adaptive
mechanism was designed by analyzing the influence of the system noise covariance matrix Q on the identification effect.
This mechanism could detect whether there was a significant change in the current working condition based on the estimation
error at the previous moment and dynamically adjust the values of the Q matrix and system state transition matrix in real
time. The adaptive adjustment of the Q matrix and state transition matrix was achieved, which quickly adapts to the current
working condition, increases the system’ s dynamic response speed and robustness. Simulation experiments had verified that
compared with the traditional EKF observer, the new adaptive Kalman observer maintains good filtering performance while
having a faster convergence speed. The identification accuracy of moment of inertia has been improved by 30%, the
accuracy of load torque identification has been improved by 35%, and the convergence speed has been improved by 36%.

Key words: permanent magnet synchronous motor ( PMSM ), extended Kalman filter ( EKF ), moment of inertia identification, load torque identification