基于主元分析与动态时间弯曲的故障诊断方法及应用研究  

Fault Diagnosis Method and Applied Research Based on Principal Component Analysis and Dynamic Time Warping

作　　者：邱立鹏[1] 姚海妮 王珍[1] 陈建国[1] 杨铎[1] 

机构地区：[1]大连大学机械工程学院,大连116622

出　　处：《计算机测量与控制》2016年第1期92-94,104,共4页Computer Measurement &Control

基　　金：国家自然科学基金(51405053);辽宁省教育厅一般项目(L2012446)

摘　　要：轴承是工程实际中常用而又极易损坏的部件,特别是对其早期微弱响应的辨识,具有重要的社会价值和意义;为提高运转轴承的安全可靠性和可维护性,提出了基于主元分析与动态时间弯曲距离的故障诊断方法,它可以准确对早期微弱动态响应辨识、诊断;该方法首先将典型故障样本信号与待测信号小波去噪并EMD分解,并对若干固有模态分量主元分析求取主元,然后对主元分量进行分析,获得相关特征值组成特征向量,计算待测信号与已知故障样本信号特征向量的弯曲距离,弯曲距离越小表明两信号越相似,从而辨识故障;此外,还可将其应用于转子、碰磨、齿轮故障诊断中,工程应用实例表明该方法可以准确故障分类,高效故障诊断。To improve the operation of the mechanical safety and reliability and maintainability, proposed Fault diagnosis method and applied research based on principal component analysis and dynamic time warping,It can accurately identify the dynamic response of the weak early and diagnosis.First fault samples and measured signal de-noising of wavelet and EMD decomposition, Several IMF components strike PCA ,And PCA component was analyzed to obtain the relevant eigenvectors composed by eigenvalues composition,Calculate the signal being measured and the known sample signal of fault feature vectors bending distance, the smaller the distance, the more similar bent two signals.Application examples of engineering show that the method can accurately classify faults, efficient troubleshooting.

关 键 词：EMD PCA 动态时间弯曲 故障诊断 相似性度量 

分 类 号：TH133.3[机械工程—机械制造及自动化] TH165