基于分段线性分类器的滚动轴承的故障识别

为了解决滚动轴承的特征提取和故障特征的模式分类问题,提出了一种应用小波包变换和线性分类器相结合的滚动轴承故障诊断的识别方法。根据轴承振动信号的频域变化特征,首先对滚动轴承振动信号进行三层小波包分解,提取第三层各个终节点系数的能量作为特征向量,然后将特征向量输入由线性判别式构成的分段线性分类器中进行故障的模式分类和识别,最后在滚动轴承试验台上实测故障。试验表明,分段线性分类器可以有效地识别轴承的故障模式。

Fault Pattern Recognition of Rolling Bearing Based on Modified Linear Classifier

The method of fault recognition is presented based on wavelet packet transform and modified linear classifier,in order to solve feature extracting and feature classifying of rolling bearing diagnosis.According to frequency domain feature of vibration signal,the signal of rolling bearing is decomposed into three-layer by wavelet packet.The energy coefficient and entropy coefficient of the third layer node are extracted and deemed as characteristic vector;then,fault pattern of rolling bearing is recognized by using modified linear classifier constructed with linear discriminant functions;lastly,the fault of rolling bearing is simulated.The shows that the method is available to accurately recognize the fault pattern of rolling bearing.