基于EEMD和PSO-SVM的电机气隙偏心故障诊断

Motor Air-Gap Eccentricity Fault Diagnosis Based on EEMD and PSO-SVM

针对电机气隙偏心故障如何通过振动信号进行有效诊断、如何选取合适故障特征等系列问题,提出了基于集合经验模态分解(EEMD)的Hilbert时频谱能量特征表达和粒子群参数优化的支持向量机(PSO-SVM)的故障诊断方法。首先对振动信号进行EEMD分解,并通过相关系数法选择有效的IMF分量;其次,对有效的IMF分量提取Hilbert时频谱能量作为特征向量;最后,利用PSO-SVM对提取的特征进行故障的识别。实验结果表明:利用该方法可以对电机偏心故障进行准确诊断。通过与其他传统故障特征在PSO-SVM下进行的比较,验证了Hilbert时频谱能量特征可以获得更高的诊断准确率。

Aiming at the series of problems of how to effectively diagnosing air-gap eccentricity fault based on vibration signals,and how to select appropriate fault characteristics,a method based on Hilbert time-frequency spectrum energy of ensemble empirical mode decomposition(EEMD)and support vector machine is optimized by particle swarm optimization(PSO-SVM).The vibration signal was first decomposed into IMF by EEMD,and then the effective IMF was filtered by the correlation coefficient.The Hilbert time-frequency spectrum energy of effective IMF was extracted as feature vector,and feature vectors were then used to train PSO-SVM for fault diagnosis.The experimental results show that using this method can accurately diagnose the eccentric fault of the motor,and by comparing with other traditional fault characteristics under PSO-SVM,it is verified that the characteristics of Hilbert time-frequency spectrum energy can obtain higher diagnostic accuracy.