基于自适应局部迭代滤波和模糊C均值聚类的滚动轴承故障诊断方法  

Fault Diagnosis Method for Rolling Bearings Based on ALIF and KFCM Clustering

作　　者：张超[1] 何闯进 何玉灵[1] ZHANG Chao;HE Chuangjin;HE Yuling(Department of Mechanical Engineering, North China Electricity Power University, Baoding 071003, China)

机构地区：[1]华北电力大学机械工程系,河北保定071003

出　　处：《轴承》2021年第5期50-55,62,共7页Bearing

基　　金：国家自然科学基金项目(51777074);河北省自然科学基金项目(E2020502032);河北省第三批青年拔尖人才支持计划([2018]-27);中央高校基本科研业务费专项资金资助项目(2017MS152)。

摘　　要：为准确提取滚动轴承振动信号的故障特征,并对不同状态信号进行划分,提出了一种基于自适应局部迭代滤波(ALIF)和模糊C均值(KFCM)聚类的滚动轴承故障诊断方法。首先,将多模态信号自适应分解为多阶单一模态分量;然后,结合相关系数提取出含有最多故障特征信息的最优分量,计算其近似熵值并构建特征向量矩阵;最后,将得到的特征向量输入KFCM得到聚类结果。试验结果表明,与基于EMD,EEMD和KFCM聚类,以及ALIF和FCM聚类的方法相比,ALIF和KFCM方法的分类系数更接近1,平均模糊熵更接近0,聚类效果更好,对滚动轴承各类故障信号具有很高的识别度和良好的分类效果。In order to extract the fault features of rolling bearing vibration signals accurately and classify the signals under different states,a fault diagnosis method for rolling bearings based on adaptive local iterative filtering(ALIF)and kernelized fuzzy C-means(KFCM)clustering is proposed.Firstly,the multimodal signal is decomposed into multi-order single IMF component adaptively,and then combining with correlation coefficient,the optimal IMF components containing most fault feature information are extracted and its approximate entropy value is calculated to construct the eigenvector matrix.Finally,the obtained eigenvectors are input into KFCM to obtain the clustering results.The experimental results show that compared with the methods based on EMD,EEMD and KFCM clustering,as well as the methods based on ALIF and FCM clustering,the classification coefficient of ALIF and KFCM methods is closer to 1,the average fuzzy entropy is closer to 0,and the clustering effect is better.The ALIF and KFCM methods have high recognition degree and good classification effect for various fault signals of rolling bearings.

关 键 词：滚动轴承 故障诊断 自适应局部迭代滤波 模糊C均值聚类 近似熵 

分 类 号：TH133.33[机械工程—机械制造及自动化] TP181[自动化与计算机技术—控制理论与控制工程]