基于TSMLZC和SVM的滚动轴承故障检测方法  

Fault detection of rolling bearing based on TSMLZC and SVM

作　　者：卢艳静 许艳英 包宋建[4] LU Yanjing;XU Yanying;BAO Songjian(College of Electronic and Electrical Engineering,Henan Normal University,Xinxiang 453007,China;College of Information Engineering,Sanmenxia College of Social Administration,Sanmenxia 472000,China;College of Artificial Intelligence,Chongqing Creation Vocational College,Chongqing 402160,China;School of Electronic Information and Electrical Engineering,Chongqing University of Arts and Sciences,Chongqing 402160)

机构地区：[1]河南师范大学电子与电气工程学院,河南新乡453007 [2]三门峡社会管理职业学院信息工程学院,河南三门峡472000 [3]重庆科创职业学院人工智能学院,重庆402160 [4]重庆文理学院电子信息与电气工程学院,重庆402160

出　　处：《机电工程》2023年第9期1326-1335,共10页Journal of Mechanical & Electrical Engineering

基　　金：重庆市高校创新研究群体项目(CXQT21039);重庆市教委科学技术研究计划项目(KJZD-K201905401)。

摘　　要：滚动轴承故障检测过程中存在故障特征提取困难和分类准确率较低的问题,为此,提出了一种基于时移多尺度Lempel-Ziv复杂度(TSMLZC)和灰狼优化器-支持向量机(GWO-SVM)的滚动轴承故障诊断方法。首先,基于Lempel-Ziv复杂度,提出了一种称为TSMLZC的测量信号复杂度的工具,其能够无参数地测量时间序列的复杂度;然后,利用该工具提取了滚动轴承振动信号的TSMLZC值作为故障特征,以表征滚动轴承的不同故障状态;最后,将故障特征输入到基于灰狼优化器的支持向量机分类模型中,对不同滚动轴承的状态进行了准确识别;再将该方法应用于2种实际滚动轴承故障案例,并对5种故障检测方法的诊断结果进行了分析和对比。研究结果表明:与传统的故障诊断模型——多尺度Lempel-ziv复杂度(MLZC)、多尺度熵(MSE)、多尺度模糊熵(MFE)、多尺度排列熵(MPE)相比,TSMLZC+GWO-SVM在2种数据集下分别取得了98.8%和94.4%的诊断准确率,证明了该模型能够适用于滚动轴承的故障诊断;而在诊断滚动轴承的不同负载和工况时,所建立的模型也获得了良好的分类准确率,这表明TSMLZC+GWO-SVM具有较强的竞争力。Aiming at the difficulties of feature extraction and low classification accuracy in rolling bearing fault detection,a rolling bearing fault diagnosis method based on time-shifted multi-scale Lempel-Ziv complexity(TSMLZC)and grey wolf optimizer-support vector machine(GWO-SVM)was proposed.Firstly,based on Lempel-Ziv complexity,a tool called TSMLZC for measuring signal complexity was proposed,which could measure the complexity of time series without parameters.Then,the TSMLZC value of the rolling bearing vibration signal was extracted as the fault feature to represent the different fault states of the rolling bearing by using this tool.Finally,the fault features were inputted into the support vector machine classification model based on the grey wolf optimizer to realize the accurate recognition of different rolling bearing status.The method was applied to two actual rolling bearing fault cases,and the diagnostic results of five fault detection methods were analyzed and compared.The research results show that compared with the traditional fault diagnosis models—multi-scale Lempel-Ziv complexity(MLZC),multi-scale sample entropy(MSE),multi-scale fuzzy entropy(MFE),and multi-scale permutation entropy(MPE),TSMLZC+GWO-SVM has respectively achieved 98.8%and 94.4%diagnostic accuracy under the two data sets,which proves that the model can be applied to the fault diagnosis of rolling bearings.When diagnosing different loads and working conditions of rolling bearings,the established model also achieves good classification accuracy,which shows that TSMLZC+GWO-SVM has strong competitiveness.

关 键 词：故障特征提取 故障分类 故障检测准确率 时移多尺度Lempel-Ziv复杂度 灰狼优化器-支持向量机 

分 类 号：TH133.3[机械工程—机械制造及自动化]