基于支持向量机的机械故障多类分类研究

Research on Machine Fault Pattern Classification Based on Support Vector Machine

对基于支持向量机的多类分类故障诊断方法进行了研究。采用9频段幅值谱作为分类器的特征输入,比较了现有常用的几种支持向量机多分类方法:"一对一"法、"一对多"法、导向无环图法,试验结果表明导向无环图法耗时短、分类精度更高,更适合应用于机械多分类故障诊断研究。

This paper studies multi-class problem of machine fault diagnosis based on vector machine. 9 segments of frequency spectrum is used as the input of SVM multiple classifier Three existing multi-class methods of SVM： ＂one-vs-one＂, ＂one-vs-all＂, Directed Acyclic Graph （DAG） is compared to test. Cross validation is use to optimize the parameters of SVM. The result indicates that DAGSVM has fewer time resume and higher classification precision than the other methods.