Multi-Sensor Fusion-Based Permanent Magnet Demagnetization Detection in Permanent Magnet Synchronous Machines

When permanent magnet (PM) demagnetization in a PM synchronous machine (PMSM) occurs, the demagnetization characteristic information is usually coupled with other characteristics of the PMSM which may demonstrate similar features. Therefore, it is difficult to accurately detect the demagnetization by monitoring only one parameter. This paper explores the use of multi-sensor information, namely, acoustic noise and torque, for demagnetization detection through the information fusion technique. The acoustic noise and torque information are processed and analyzed using wavelet transforms for filtering and extracting features. Bayesian network-based multi-sensor information fusion is then proposed to detect the demagnetization ratio from the extracted features. The proposed approach is experimentally verified on a laboratory PMSM and compared with single-sensor dection methods.