基于类间可分性度量和SVM的多故障分类算法  

Multi-Fault Classification Algorithm Based on Inter-Class Separability and Support Vecton Machine

作　　者：李敏[1,2] 杨洁明[1] 张晓平[1] 

机构地区：[1]太原理工大学机械电子工程研究所,太原030021 [2]山西省科技厅,太原030024

出　　处：《振动．测试与诊断》2009年第1期83-85,共3页Journal of Vibration,Measurement & Diagnosis

基　　金：国家自然科学基金资助项目(编号:50335030);山西省自然科学基金资助项目(编号:2006011056)

摘　　要：由于基于支持向量机(Support Vector Mechine,简称SVM)的多类分类算法的分类器结构对故障分类效果有直接影响,首先提出一种在高维特征空间基于核函数的Jb'/Jw'类间可分性度量准则,然后根据可分性度量结果用min-max原则来构造支持向量机分类器结构。通过对滚动轴承4类故障测试样本的分类试验结果表明,采用该方法构建的分类器其分类效果明显优于任意确定的二叉树结构分类器,有更高的故障识别率。The multi-classification algorithm structure of the support vector machine(SVM) has a decisive effect on the classification ability.Based on Kernel function mapping,this paper presents an inter-class separability method in high dimensional feature space called Jb/Jw.Then the classifier structure of SVM is designed with the min-max rule.Finally,classifying experiment of 4 types of rolling bearing faults is done,and the results show that the multi-fault classifier based on the Jb/Jw method and the min-max rule can identify the fault pattern more accurately and effectively.

关 键 词：支持向量机 类间可分性 故障分类器 滚动轴承 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]