基于mixup-LSTM的永磁同步电机故障诊断方法  

Fault Diagnosis of Permanent Magnet Synchronous Motor Based on Mixup-LSTM

作　　者：张立松 杨明发[1] ZHANG Li-song;YANG Ming-fa(College of Electrical&Automation Engineering,Fuzhou University,Fuzhou 350116,China)

机构地区：[1]福州大学电气工程与自动化学院,福建福州350116

出　　处：《电气开关》2022年第5期58-62,共5页Electric Switchgear

摘　　要：本文针对永磁同步电机匝间短路和失磁故障进行研究,提出了一种基于mixup数据增强和机器学习分类器的故障诊断方法。该方法提取通过小波包分解提取定子电流信号中的故障特征建立故障诊断样本,结合mixup实现样本扩张,避免小样本带来的过拟合问题。最后将扩张样本输入长短时记忆网络(long short-term memory,LSTM)进行分类。结果表明,该方法能够高效地实现永磁同步电机故障诊断,且具有较高的准确度和较强的抗噪性能。In this paper,a fault diagnosis method based on mixup data augmentation and machine learning classifier is proposed to study the inter-turn short-circuit and loss-of-excitation faults of permanent magnet synchronous motors.The method extracts the features in the stator current signal through wavelet packet decomposition to establish fault detection samples,and combines the mixup to realize the sample expansion,so as to avoid the problem of over-fitting caused by small samples.Finally,the dilated samples are input into a long short-term memory(LSTM)for classification.The results show that the method can efficiently realize the fault diagnosis of permanent magnet synchronous motor,and has high accuracy and strong anti-noise performance.

关 键 词：永磁同步电机 故障诊断 数据增强 长短时记忆网络 

分 类 号：TM33[电气工程—电机]