基于ALE-MRAS磁链观测器的永磁同步直线电机均匀退磁诊断研究

Researches on Uniform Demagnetization Diagnosis of Permanent Magnet Synchronous Linear Motors Based on ALE-MRAS Flux Observers

设计了自适应线性神经元?模型参考自适应(adaptive linear element-model reference adaptive system,ALE-MRAS)磁链观测器,以解决永磁同步直线电机(permanentmagnet synchronouslinearmotor,PMSLM)多工况下均匀退磁磁链观测问题。首先,建立PMSLM正常、微弱、轻度、中度以及重度退磁故障下的有限元模型,仿真计算不同退磁程度下磁链的精确值;其次,建立模型参考自适应(model reference adaptive, MRAS)磁链观测器,采用自适应线性神经元(adaptivelinear element,ADALINE)算法在线修正MRAS的自适应律,得到ALE-MRAS磁链观测器,实现多工况下准确地在线磁链观测;然后,选取磁链幅值为故障特征值,结合ALE-MRAS磁链观测器观测结果,与有限元故障模型计算得到的磁链精确值进行对比,诊断出故障及故障程度;最后,制作退磁样机,搭建实验平台,对不同工况下的退磁故障进行诊断,精确度高达98%以上。通过与滑模磁链观测器和MRAS磁链观测器的对比,验证了文中诊断方法的正确性和先进性。

A adaptive linear element-model reference adaptive system(ALE-MRAS) flux observer was designed to solve the problem of observing the uniform demagnetization linkage of permanent magnet synchronous linear motors(PMSLM) under multi-working conditions. The finite element models of PMSLM under normal, weak, mild, moderate and severe demagnetization faults were established, and the precise values of flux linkage under different demagnetization degrees were simulated and calculated. Then, a model reference adaptive(MRAS) flux observer was established, and the adaptive law of MRAS was modified online by using the adaptive linear neuron(ADALINE) algorithm. The ALE-MRAS flux observer was obtained to realize accurate on-line flux observation under multiple operating conditions. Moreover, selecting the magnitude of the flux linkage as the fault eigenvalue, Comparing the observation results of the ALE-MRAS flux observers with the exact value of the flux linkage calculated by the finite element fault model to diagnose the fault and the degree of the fault. Finally, a demagnetization prototype was made and an experimental platform was built to diagnose the demagnetization fault under different working conditions. The accuracy is over 98%. Compared with the sliding mode flux observer and the MRAS flux observer, the correctness and advancement of the diagnosis method in this paper were verified.