主减速器故障的增强多尺度微分符号熵和优化SVM诊断

Main Reducer Fault Diagnosis Based on Enhanced Multi-scale Differential Symbolic Entropy and Optimized SVM

为准确提取出能够表征车辆主减速器故障的故障特征,同时针对多尺度微分符号熵(Multi-scale Differential Symbolic Entropy,MDSE)粗粒化过程中存在的问题,提出增强多尺度微分符号熵(Enhanced Multi-scale Differential Symbolic Entropy,EMDSE)的概念,并结合蝴蝶算法(Butterfly Optimization Algorithm,BOA)优化的支持向量机(Support Vector Machine,SVM),提出主减速器故障诊断的EMDSE和BOA-SVM方法。EMDSE可解决MDSE粗粒化过程中存在的信息泄露和计算结果不稳定的不足,能够更加有效地利用信号中存在的故障信息。主减速器故障诊断实例结果表明,相比于MDSE,EMDSE的计算结果更稳定,对主减速器不同故障状态的可区分性更强,BOA-SVM得到的诊断精度更高。

In order to accurately extract the fault features that can represent the fault of the vehicle′s main reducer,and to solve the problems existing in coarse granulating process of the multi-scale differential symbolic entropy(MDSE),the concept of enhanced multi-scale differential symbolic entropy(EMDSE)was proposed.Combined with the support vector machine(SVM)optimized by butterfly optimization algorithm(BOA),a main reducer fault diagnosis method based on EMDSE and BOA-SVM was presented.The EMDSE solves the problems of information leakage and unstable calculation results in the process of MDSE coarse granulating,and can make more effective use of the fault information in the signal.The example results of main reducer fault diagnosis show that,compared with MDSE,the calculation results of EMDSE are more stable and can distinguish different fault states of the main reducer more precisely,and the diagnostic accuracy obtained by BOA-SVM is higher.