基于小波包最优熵与RVM的滚动轴承故障诊断方法

Fault Diagnosis Method for Rolling Bearings Based on Wavelet Packet Optimal Entropy and RVM

为解决滚动轴承振动信号信噪比低和故障分类准确性不高的问题,提出了小波包最优熵和相关向量机相结合的故障诊断方法。首先采用小波包对采集到的信号进行信噪分离,寻找分解后信号的最优小波包节点熵;然后提取最优节点能量作为训练样本,对相关向量机的多故障分类器进行训练,实现轴承的智能诊断。试验表明,该方法可简单有效地分离噪声,并具有良好的分类能力,可以很好地应用于轴承故障诊断。

To solve the low signal -to -noise ratio of vibration signals for rolling bearings and the accuracy of fault clas-sification,a fault diagnosis method is presented,which combines wavelet packet optimal entropy and relevance vector machine（RVM）.Firstly the acquisition signals are separated by wavelet packet,the optimal wavelet packet node entro-py of decomposed signal is searched.Then the energy of optimal node is extracted as training samples,the multi -fault classifier for RVMare trained,and the intelligent diagnosis for bearings is realized.The test shows that the method can simply and effectively separate noise,the classification ability is good,which can be good for fault diagnosis of bear-ings.