改进LMD和排列熵的滚动轴承故障诊断  

Fault Diagnosis of Rolling Bearing Based on the Improved LMD and Permutation Entropy

作　　者：李巧艺[1] 单奇[1] 陈跃威 叶运广 LI Qiao-yi;SHAN Qi;CHEN Yue-wei;YE Yun-guang(School of Mechanical Engineering,Southwest Jiaotong University,Sichuan Chengdu 610031,China)

机构地区：[1]西南交通大学机械工程学院,四川成都610031

出　　处：《机械设计与制造》2018年第4期51-53,57,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金项目(51275426);国家自然科学基金资助项目(51475386)

摘　　要：针对滚动轴承故障振动信号的复杂特性和局部均值分解(Local Mean Decomposition,LMD)方法存在的端点效应问题,提出了基于振动信号自相似性对左右端点两侧延拓来抑制端点效应问题的改进LMD、排列熵(Permutation Entropy,PE)及优化K-均值聚类算法相结合的轴承故障诊断方法。首先通过改进LMD将非线性、非平稳的原始故障振动信号分解出一系列的乘积函数(Production Function,PF)分量,对包含主要故障信息的PF分量提取PE值作为故障特征分量,在提取特征量的基础上,最后采用优化后的K-均值聚类算法对故障类型进行识别分类。将该方法应用在滚动轴承实验数据,实验结果表明该方法可以准确、有效的实现滚动轴承的故障诊断。In the processing of fault vibration signals of rolling bearings,a novel signal fault feature extraction method based on the improved local mean decomposition(LM D),which is based on the self-similarity of vibration signal extending the right and left sides of the original signal to suppress its edge effect,permutation entropy(PE)and the optimized K-means clustering algorithm is proposed in consideration of the complexity for rolling bearing’s fault vibration signals and edge effect of LM D.First,vibration signal in each state are decomposed into a series of PF components with the improved LM D method,and the PF components which contain the main information of fault state are chosen.Permutation entropy of the selected PF components is calculated,and the optimized K-means algorithm is used to cluster analysis as a pattern recognition approach.Finally,the experiment results show the proposed method is effectively to fault extraction and diagnosis for rolling bearing.

关 键 词：改进的局部均值分解 排列熵 端点效应 自相似性 滚动轴承 故障诊断 

分 类 号：TH16[机械工程—机械制造及自动化] TH165.3