A new method named Windowed Least Square (WLS) to test main parameters of Permanent Magnet Synchronous Motor (PMSM) is proposed in this paper. Compared with Extended Kalman Filter (EKF) & Elman neural network and Recursive Least Square (RLS), WLS guarantees identification accuracy and excellent timeliness, and the issue of data saturation of RLS can be avoided. The PMSM model is built combining on-line parameter identification with Active Disturbance Rejection Control (ADRC) to improve the control performance of PMSM. The simulation results demonstrate that the performance of ADRC system using online estimation strategy is better than that of the system using PID method.
Parameter Identification of Heffron-Phillips Model Considering AVR Using On Line Measurements Data