Unbalance compensation based on speed fault-tolerance and fusion strategy for magnetically suspended PMSM

High-reliability rotor speed estimation method is essential for implementing unbalance compensation of magnetically suspended permanent magnet synchronous motors (PMSM) without speed sensors. However, the current investigations rarely focus on the fault-tolerance of speed estimation methods, making it challenging to achieve stable and reliable operation of magnetically suspended PMSM upon the impacts represented by DC offset and shock excitation. Herein, a novel speed fault-tolerance and fusion strategy (SFTFS) is proposed to improve the reliability and accuracy of rotor speed estimation. Making use of this strategy which involves in the acquisition of two independent yet redundant speed information from the vibration displacement of rotor and sensorless vector control systems of PMSM, it is feasible to realize automatic diagnosis and isolation of abnormal speed information using the innovation noise characteristics of local Kalman filters when one speed estimation method fails, thus ensuring the validity of speed information. Moreover, when both speed estimation methods are functioning normally, the strategy solves for the optimal fusion matrix weights based on the noise characteristics of two speed information, thereby adaptively optimizing the fusion and enhancing the accuracy of speed information. Finally, the proposed strategy was experimentally validated on a magnetically suspended turbo molecular pump, highlighting its effectiveness and superiority.