基于重分类与支持向量机的交流电路故障诊断  

AC Circuit Fault Diagnosis Based on Re-classification and SVM

作　　者：黄亮[1] 侯建军[1] 魏学业[1] 宋伟[1] 李赵红[1] 

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《信号处理》2009年第9期1484-1487,共4页Journal of Signal Processing

基　　金：国家863计划项目(编号:2007AA01Z241-2)

摘　　要：分析了交流电路故障诊断中的故障类重叠问题。针对在该情况下故障诊断正确率低的问题,采用了一种新的故障分类方法。首先用MATLAB软件编写电路程序,生成样本,然后用故障样本对支持向量机进行训练,最后用训练好的模型文件对测试样本进行多故障分类。通过一个交流电路的故障诊断实例表明,该方法具有算法简单、泛化能力强等优点,重分类后的故障诊断正确率高于重分类前。The fault overlap circumstances in AC circuit fault diagnosis are analyzed. In order to solve the problems of low accuracy of fault diagnosis under the above circumstances, a fault re-classification method is proposed. Firstly, the circuit program is developed using MATLAB software, samples are obtained. Secondly, the fault samples are sent into SVM to classify the multiple fault of analog circuits. Finally, a fault diagnosis example of AC circuit is provided, the result shows that this method has the advantages of simple algorithm,high efficiency, and great capability in generalization and classification, the fault diagnosis accuracy of re-classification is improved.

关 键 词：支持向量机 故障诊断 交流电路 重分类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP206.3[自动化与计算机技术—控制科学与工程]