基于多域特征的滚动轴承故障检测和状态识别方法

Fault detection and state recognition of rolling bearing based on multi domain feature

为提高滚动轴承故障检测与识别效果,提出一种基于局部均值分解(LMD)和共空间模式(CSP)的时-频-空多域特征提取方法。首先采用LMD将滚动轴承信号分解为多个乘积分量(PF)并提取时-频熵特征,基于支持向量数据描述(SVDD)分类器实现正常和故障轴承的分类;然后利用CSP对故障轴承信号进行分解并提取空域熵特征;最后利用K-均值聚类算法进行聚类,实现对外圈故障、内圈故障和滚柱故障的区分。实验结果表明,所提方法可以获得优于80%的正确分类性能,明显优于传统单一维度特征。

In order to improve the effect of fault detection and recognition of rolling bearing,a time-frequency-space multi domain feature extraction method based on local mean decomposition(LMD)and common spatial pattern(CSP)is proposed.Firstly,the rolling bearing signal is decomposed into multiple product functions(PF)by LMD,and the time-frequency entropy features is extracted to realize the classification of normal and fault bearings based on support vector data description(SVDD)classifier,then CSP is used to decompose the fault bearing signal and extract the spatial entropy features.Finally,K-means algorithm is used for clustering to distinguish outer ring fault,inner ring fault and roller fault.Experimental results show that the proposed method can achieve better than 80%the correct classification performance,remarkably better than the traditional single dimension features.