Abstract:
Abstract: To address the problems of locked-rotor and startup failure caused by excessive initial position estimation errors during the sensorless startup of permanent magnet synchronous motors (PMSMs), a startup state monitoring and compensation strategy based on the extended Kalman filter (EKF) was proposed. First, the EKF was employed to estimate the currents, speed, and rotor position. An electromagnetic torque model and a mechanical motion model considering the position error angle were established, and the startup states were classified into six types based on the torque balance relationship to reveal the mechanism of the undesired zero-speed equilibrium. Then, a compensation term proportional to the q-axis control voltage was introduced into the q-axis current prediction equation of the EKF to break the undesired equilibrium and enable normal startup. Simulation results show that at initial error angles of 80°, 160°, and 240°, startup fails in all cases without compensation, whereas successful startup is achieved after compensation is applied. The maximum speed rise time is 0.25 s, and the absolute values of the steady-state speed error and the final position error angle are kept within 4.9 rad/s and 9.2°, respectively. Under variable-speed conditions as well as sudden load application and removal, the system exhibits good robustness, with a maximum absolute steady-state speed error of only 4.36 rad/s. Compared with the existing q-axis current compensation strategy, the proposed strategy achieves a lower steady-state speed error. By integrating state estimation with startup abnormality suppression, the proposed strategy eliminates the need for additional hardware sensors, significantly improves startup reliability under large initial position errors, and provides new insights for the state monitoring of industrial drive systems.