基于经验模态分析的故障诊断方法及应用研究  

Fault Diagnosis Method and Application Research Based on Empirical Mode Decomposition

作　　者：姚海妮 王珍[1] 邱立鹏[1] 杨铎[1] 

机构地区：[1]大连大学机械工程学院,辽宁大连116622

出　　处：《计算机测量与控制》2015年第12期3980-3982,3986,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：为实现对微弱动态响应的准确辨识及故障状态的早期诊断,提出了基于经验模态分析的故障诊断方法,将模态分解、互信息熵与主元分析结合,故障特征更凸显,方法更有效;首先模态分解,得到一系列固有模态分量,利用互信息熵判断所有固有模态分量的高低频分界点并对高频分量自适应阈值去噪;将去噪后的所有高频分量和低频分量主元分析,计算各主元的峭度值,选取峭度值大的分量求时频谱得故障频率,从而确定故障;将该方法应用到含有高频环境噪声的轴承故障信号中诊断可靠、准确。To achieve early diagnosis and accurate, fault identification dynamic response of a weak state, proposed fault diagnosis method and application research based on empirical mode decomposition. EMD, mutual information entropy are combined with PCA, fault features more prominent, more effective. First EMD decomposition, Intrinsic mode components then use mutual information entropy determine high and low frequency cut off point for all IMFcomponents and high frequency components of the adaptive thresholding. All high--frequency components and low frequency components denoised PCA principal component analysis, Calculate kurtosis value of each principal compo nent, select kurtosis large component for determining a time--frequency spectrum obtained the fault frequency. The method is applied to the bearing, fault signal contains high frequency ambient noise in diagnostic accuracy.

关 键 词：经验模态分解(EMD) 互信息熵(IM) 主成成分分析(PCA) 故障诊断 时频谱 

分 类 号：TH165.3[机械工程—机械制造及自动化] TH133.3