基于决策树和相关向量机的智能故障诊断方法

Intelligent fault diagnosis method based on decision tree and RVM

针对故障诊断面临的故障样本少、非线性强、多故障处理等问题以及传统智能诊断方法存在的不足,提出了一种基于决策树(DT)和相关向量机(RVM)的智能故障诊断方法。通过构造决策二叉树,将多类分类问题分解成多个二类分类问题;在各个决策节点,利用RVM进行二类分类,从而实现RVM的多类分类。理论分析及仿真结果表明,相比支持向量机,新方法在保持高诊断正确率的同时具有更高的稀疏性和诊断效率,并且能够提供概率式输出,更具实用价值;相比OAR-RVM和OAO-RVM方法,新方法节省了训练时间,具有更高的训练效率。

In view of the problems in fault diagnosis, such as small samples, nonlinear, multiple faults processing, and the defects of traditional intelligent methods, an intelligent fault diagnosis method based on Decision Tree（DT） and Relevance Vector Machine（RVM） is proposed. The DT is constructed, and the multi-class classification problem is divided into many binary classi- fication problems. RVM is used to make binary classification at every node, and then the multi-class classification of RVM is achieved. The theoretical analysis and results of application show that the proposed method has better performance in sparsity and diagnosis efficiency while keeping high accuracy compared with the traditional SVM methods, which makes it more practi- cal; and that the proposed method has a better training efficiency compared with OAR-RVM and OAO-RVM.