基于最小二乘支持向量机的轨道电路故障诊断方法  

Track Circuit Fault Diagnosis Method Based on Least Squares Support Vector Machine

作　　者：王彤[1] 

机构地区：[1]兰州交通大学自动化与电气工程学院,兰州730070

出　　处：《铁道标准设计》2014年第2期89-93,共5页Railway Standard Design

基　　金：国家自然科学基金资助项目(61074139);国家支撑计划项目(2009BAG11B02)

摘　　要：为提高轨道电路故障处理效率和正确率,对轨道电路的多故障诊断方法进行研究。建立基于最小二乘支持向量机的轨道电路故障诊断模型,用某轨道电路实测数据进行训练和测试,选择基于BP神经网络的故障诊断方法进行对比。结果表明:基于最小二乘支持向量机的轨道电路故障诊断方法能有效实现轨道电路5种故障的诊断,且具有更快的运算速度。与BP神经网络故障诊断方法比较,故障诊断正确率提高了17.14%,运算时间减少2/3。In order to improve the troubleshooting efficiency and accuracy of track circuit, the multi-fauh diagnosis method of track circuit was researched in this paper. The fault diagnosis model of track circuit was established based on least squares support vector machine, and then the data measured from an actual track circuit were employed to verify the feasibility of this model. Finally, this model was compared with the fault diagnosis method based on BP neural network. The research result shows that this track circuit fault diagnosis model based on least squares support vector machine can effectively diagnose five kinds of track circuit faults with much faster computing speed. Compared with fault diagnosis methods based on BP neural network, the accuracy was improved by 17.14% , and the computing time was reduced by two-thirds.

关 键 词：ZPW-2000轨道电路 故障诊断 最小二乘支持向量机 多故障分类 

分 类 号：U284.2[交通运输工程—交通信息工程及控制]