基于EMD的改进马田系统的滚动轴承故障诊断

摘要

为了提高滚动轴承的可靠性、及时发现其潜在的故障,提出了一种基于改进马田系统(MMTS)的滚动轴承故障诊断方法.首先利用经验模态分解(EMD)方法对原振动信号进行分解,得到了多个本征模态分量(IMF)并计算基本模式分量的统计特征集.然后,在此基础上构建基准空间(马氏空间),针对马田系统在筛选特征变量时效果不佳、基准空间数据的差异性问题,引入粗糙集(RS)筛选有效特征变量改进马田系统,大幅降低特征向量的维数.最后,计算待诊断信号到基准空间的马氏距离,从而完成滚动轴承的故障诊断.利用滚动轴承振动数据对该模型进行了测试,结果表明,该模型与实际相符,可以准确、有效地识别滚动轴承的故障类型.%In order to improve reliability of rolling bearings and find their potential faults,a method of rolling bearings fault diagnosis based on Modified Mahalanobis-Taguchi System was proposed.Firstly,the original vibration signal was decomposed into several intrinsic mode functions by means of the empirical model decomposition (EMD) and the statistical characteristics of the basic mode components were calculated.Then effective feature variables were screened with rough set aiming at shortages of Mahalanobis-Taguchi system in screening feature variables and the problem of difference in the reference space (Mahalanobis space) data.Mahalanobis-Taguchi System was improved and the number of dimension of the feature vector was obviously reduced.Finally,Mahalanobis distance from a signal to be diagnosed to the reference space was calculated and the fault diagnosis of a rolling bearing was completed.This model was verified using vibration data of rolling bearings.The results showed that this model agees well with actuality and can identify fault types correctly and effectively.