多特征评估筛选的滚动轴承故障诊断算法  

Rolling Bearing Fault Diagnosis Algorithm Based on Multi Feature Evaluation and Selection

作　　者：徐国权 罗倩[1] 郭鹏飞 XU Guo-qu.an;LUO Qian;GUO Peng-fei(School of Information and Communication Engineering,Beijing Information Science and Technology University, Beijing 100101,China)

机构地区：[1]北京信息科技大学信息与通信工程学院,北京100101

出　　处：《计算机仿真》2018年第12期446-450,455,共6页Computer Simulation

基　　金：国家自然科学基金资助项目(61271198);北京市科技提升计划项目(PXM2016_014224_000021);促高效内涵发展-研究生科技创新项目(5111624111)

摘　　要：针对滚动轴承故障的复杂性,现代信号处理技术和特征提取技术的多样性提供了多特征故障诊断技术的思路,但是特征集的特征过多不仅会增加计算量,而且有时还会降低故障诊断正确率。提出基于故障类内、类间标准差的多特征评估筛选方法,先对信号进行经验模式分解得到本证模式分量,再对原始信号和本证模式分量提取时域特征,使用该方法对提取到的特征进行敏感度评估和排序,筛选出最优特征集,使用贝叶斯判别分析对故障特征进行诊断分类。仿真结果表明,该方法能对故障进行有效、准确的诊断,并且在样本量较少的情况下,对现实中的滚动轴承故障数据也有较好的诊断率。In view of the complexity of rolling bearing faults,the diversity of modem signal processing technology and feature extraction technology provides the idea of multi feature fault diagnosis technology.However,if the feature set is too large,it will not only increase the computational complexity,but also sometimes reduce the accuracy of fault diagnosis.A method of multi feature assessment and screening based on intra class and inter class standard deviations was proposed to carry out sensitivity evaluation and sorting of features.Firstly,empirical mode decomposition was used to obtain the intrinsic mode component.Secondly,the time domain features was extracted from the original signal and the intrinsic mode component,and the sensitivity of the extracted features was evaluated and sorted by using this method to screen the optimal feature set.Finally,Bayesian,discriminant analysis was used to diagnose and classify the fault characteristics.The simulation results show that the method can diagnose faults effectively and accurately,and in the case of a small sample size,it has good diagnostic effect.

关 键 词：滚动轴承 多特征 标准差 敏感度 贝叶斯判别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] B[自动化与计算机技术—计算机科学与技术]