基于特征量融合和支持向量机的轴承故障诊断  

Bearing fault diagnosis based on feature fusion and support vector machine

作　　者：史庆军[1] 郭晓振 刘德胜[1] Shi Qingjun;Guo Xiaozhen;Liu Desheng(College of Information and Electronic Technology,Jiamusi University,Jiamusi 154007,China)

机构地区：[1]佳木斯大学信息电子技术学院

出　　处：《电子测量与仪器学报》2019年第10期104-111,共8页Journal of Electronic Measurement and Instrumentation

基　　金：黑龙江省留学归国人员基金(LC2017027);佳木斯大学科技创新团队建设项目(CXTDPY-2016-3)资助

摘　　要：为了准确地检测轴承故障,提出了基于经验模态分解(EMD)和局部均值分解(LMD)轴承振动信号相结合构成特征量矩阵的方法。首先对轴承振动信号进行EMD分解得到前三阶本征模态函数(IMF)分量的上下包络值矩阵的奇异值,通过LMD分解,得到各PF分量的能量炳和,然后将奇异值和能量爛融合成一个特征向量矩阵,最后用支持向量机多分类算法进行分类。经过实验仿真验证,滚动轴承滚珠、内圈和外圈故障识别的准确率为90%,与单一特征量作为支持向量机的诊断输入来比较,该方法能更加精准地识别出了轴承故障。In order to accurately detect bearing faults,a method based on EMD and LMD decomposition bearing vibration signals was proposed to form a feature quantity matrix.Firstly,the EMD decomposition of the bearing vibration signal was used to obtain the singular value of the upper and lower envelope matrix of the first three orders of IMF components.The energy entropy of each PF component was obtained by LMD decomposition,and then the singular value and the energy entropy were merged into one eigenvector.The matrix was finally classified by the support vector machine multi-classification algorithm.The experimental simulation results show that the accuracy of fault identification of rolling element,inner ring and outer ring of rolling bearing is 90%.Compared with the single eigenvalue as the diagnostic input of SVM,this method can identify the bearing faults more accurately.

关 键 词：轴承故障 经验模态分解 能量爛 支持向量机 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]