LAUSR

This article focuses on the local demagnetization fault recognition research of permanent magnet synchronous linear motor (PMSLM) and realizes the accurate identification of the position and degree of demagnetized permanent magnets. A fault recognition system based on S-transform (ST) and particle swarm optimization–least squares support vector machine (PSO–LSSVM) is proposed. The ST makes the induced electromotive force (EMF) signal with stronger signal characteristic expression ability, and the PSO–LSSVM model achieves better generalization ability and higher accuracy in the small sample state of PMSLM faults. In the process of fault identification: 1) the induced EMF analytical model for PMSLM under local demagnetization fault is presented; 2) the induced EMF signal is analyzed by ST, and the characteristic parameters are extracted from time–frequency curves. Then a characteristic vector is established by comparing the standard deviation values and similarities of different parameters; 3) a PSO–LSSVM classification model is established to realize the recognition of PMSLM faults. Prototype and finite element simulation experimental results confirm that the method can recognize the PMSLM faults accurately with a recognition rate of 100%.