Vold–Kalman Filtering Order Tracking Based Rotor Demagnetization Detection in PMSM

Abstract

Rotor magnet condition is important to maintain a stable permanent magnet synchronous motor (PMSM) operation. In this paper, Vold–Kalman filtering order tracking (VKF-OT) and dynamic Bayesian network (DBN) are employed for the real-time rotor demagnetization detection from the torque ripple. First, a torque ripple model of the PMSM considering electromagnetic noise is proposed, and the torque variation is studied to determine the effect of the demagnetization on the torque ripple, which indicates that it is feasible to detect the magnet demagnetization by analyzing the torque ripple. Then the torque is processed by wavelet transform to eliminate the electromagnetic disturbances. Second, the VKF-OT is introduced to track the order of the torque ripple of the PMSM to extract the torque ripple characteristics as the feature reflecting changes in magnet status. Third, the feature is employed to train the DBN for the rotor magnet demagnetization detection and prediction during motor operation. The proposed approach is a noninvasive and an online method that can be embedded in the physical motor controller. The validation results demonstrate that this method can detect the uniform demagnetization over a wide motor speed range.