基于CEEMD能量熵与SVM的低速轴承故障声发射诊断

摘要

Aiming at the problem of fault diagnosis of low-speed bearing,an acoustic emission diagnosis method based on the combination of complementary ensemble empirical mode decomposition (CEEMD)energy entropy and support vector machine (SVM)is proposed.Firstly,the acoustic emission signals of bearing with different damage states are decomposed by CEEMD,thus an adaptive intrinsic mode component (IMF)is obtained.Afterwards,the combination of the variance contribution rate and IMF component mutual correlation coefficient is used to remove the false component and to sift out effective component for signal reconstruction.Due to the different energy distributions of different damage bearing,the damage state of the bearing can be characterized by the change of energy entropy.The energy entropy of the extracted effective IMF components is calculated as the feature vector of different fault bearing.The feature vector is input to the support vector machine to classify and identify the different faults.The experimental results show that the correlation coefficient and variance contribution rate can be selected with the main fault information of the IMF component.At the same time,it is proven that SVM is better than BP neural network in identifying different fault types of low speed bearings.%针对低速轴承故障诊断难的问题,将互补总体平均经验模态分解(CEEMD)能量熵与支持向量机相结合对低速轴承故障进行了声发射诊断.采集不同缺陷状态的轴承声发射信号进行CEEMD分解,得到自适应的本征模态分量(IMF);结合IMF分量的方差贡献率和互相关系数对虚假分量进行剔除,筛选出有效IMF分量.对提取的有效IMF分量计算能量熵,作为不同故障轴承的特征向量.将该特征向量输入到支持向量机(SVM),对不同故障的低速轴承进行分类识别.试验结果表明,通过方差贡献率和互相关系数能够筛选出含主要故障信息的IMF分量,同时验证了SVM相比BP神经网络对低速轴承不同故障类型的识别效果更好.