EEMD能量熵与优化LS-SVM的滚动轴承故障诊断  

The Roller Bearing Fault Diagnosis Based on EEMD Energy Entropy and LS-SVM

作　　者：陈法法[1,2] 李冕[2] 陈保家[2] 陈从平[2] 

机构地区：[1]三峡大学新能源微电网湖北省协同创新中心,湖北宜昌443002 [2]三峡大学水电机械设备设计与维护湖北省重点实验室,湖北宜昌443002

出　　处：《组合机床与自动化加工技术》2016年第12期71-75,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金(51405264;51205230);三峡大学人才启动基金(KJ2014B007);湖北省教育厅项目(B2015248)

摘　　要：针对滚动轴承振动故障信号非平稳、非线性难以有效诊断的问题,提出基于集成经验模式分解(ensemble empirical mode decomposition,EEMD)能量熵与优化最小二乘支持向量机(least square support vector machine,LS-SVM)的滚动轴承故障诊断方法。首先利用EEMD对滚动轴承的振动故障信号进行分解,得到各阶的内禀模态函数分量(IMF)并计算其能量构造成特征向量矩阵,随后将该特征向量矩阵输入给优化的LS-SVM进行故障模式的分类辨识。通过实验验证了该方法的有效性和可行性,结果表明,基于EEMD能量熵特征与优化LS-SVM的滚动轴承故障诊断方法能够有效的诊断滚动轴承的实际运行工况。Aimed at the roller bearing fault vibration signal are non-stationary and nonlinear that are difficult to effectively diagnose, a roller bearing fault diagnosis method based on ensemble empirical mode decompo-sition （ EEMD） and least square support vector machine （ LS-SVM） is proposed in this paper. Firstly, the roller bearing fault vibration signal is decomposed by EEMD. Then, each intrinsic mode function compo-nents （ IMF） is got and these energy are calculated to construct the features vector matrix. Finally, the fea-ture matrix is input into the LS-SVM for the fault mode identification. The validity and feasibility of this method is verified by experiments. The results show that this method based on EEMD energy characteristic and LS-SVM can be more effective for the roller bearing fault diagnosis.

关 键 词：集成经验模式分解 最小二乘支持向量机 滚动轴承 故障诊断 

分 类 号：TH166[机械工程—机械制造及自动化] TG659[金属学及工艺—金属切削加工及机床]