基于EMD和样本熵的滚动轴承故障SVM识别

SVM Recognition Method Based on EMD and Sample Entropy in Rolling Bearing Fault Diagnosis

针对滚动轴承振动信号的非平稳特性和在现实条件下难以获取大量故障样本的实际情况,提出一种经验模态分解、非线性动力学方法—样本熵和支持向量机相结合的故障诊断方法。运用经验模态分解方法对其去噪信号进行分析,利用互相关系数准则对固有模式分量进行筛选,再计算所选分量的样本熵以组成故障特征向量,并将其作为支持向量机的输入以识别滚动轴承的状态。利用实际滚动轴承试验数据的诊断与对比试验验证了该方法的有效性和泛化能力。

According to the non-stationarity characteristics of the vibration signals from rolling beating and the situation that it＇s hard to obtain enough fault samples,a comprehensive fault diagnosis method based on Empirical Mode Decomposition （EMD）,sample entropy, a nonlinear dynamic method, and Support Vector Machine（SVM） was proposed.Firstly,the denoised vibration signals were decomposed into a finite number of Intrinsic Mode Functions （IMF）,then choosed some IMF components with the criteria of mutual correlation coefficient between IMF components and denoised signal.Thirdly the sample entropy of each 1MF component was calculated as fault eigenveetor and served as input of SVM classifier so that the faults of rolling bearing are classified.Practical rolling bearing experimental data is used to verify this method.The diagnosis results and comparative tests fully validate its effectiveness and generalization abilities.