基于支持向量机的功率变换器开关管开路故障诊断  

Fault Diagnosis of Open-switch Damage in Power Converter based on Support Vector Machine

作　　者：王德成[1] 林辉[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《火力与指挥控制》2010年第1期126-128,共3页Fire Control & Command Control

基　　金：航空基金项目(2007ZC53036)

摘　　要：提出一种基于支持向量机的功率变换器开关管开路故障诊断方法。利用支持向量机建立分类模型,模型以相电流功率谱为输入量,7种故障状态为输出量,选用高斯核,使用基于二叉树分类器的分类算法。将诊断结果与基于BP神经网络的诊断结果进行比较,结果表明基于支持向量机的分类器在功率变换器开关管开路故障诊断中具有更高的准确率和更好的泛化能力。This paper provides a fault diagnosis method based on support vector machine algorithm for open-switch damage in power converter. A model using support vector machine algorithm is built to implement classification. It takes power spectrum of motor phase current as its inputs and seven fault states as its outputs, selects the guass kernel, apply binary tree classification algorithms. A mass of fault samples are analyzed and the result is compared with those obtained by methods of BP neural networks. The result shows that the support vector machine algorithm has higher precision and better generalization capacity in fault diagnosis for open fault of power converter open-switch.

关 键 词：功率变换器 故障诊断 支持向量机 功率谱 

分 类 号：TM307.1[电气工程—电机]