基于ReliefF-Adaboost的无绝缘轨道电路故障诊断模型  

Fault Diagnosis Model of Jointless Track Circuit Based on ReliefF-Adaboost

作　　者：刘瑞丹 董昱[1] LIU Rui-dan;DONG Yu(School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)

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：《兰州交通大学学报》2023年第3期61-67,78,共8页Journal of Lanzhou Jiaotong University

基　　金：国家自然科学基金(61763023);甘肃省科技计划项目(重点研发计划)(20YF8GA037)。

摘　　要：现阶段对ZPW-2000A型无绝缘轨道电路的故障判别方式主要是对微机监测数据进行人工分析,但这种方式存在主观性强、判别效率低等问题。为解决上述问题,提出了ReliefF-Adaboost故障诊断模型对轨道电路进行故障识别。首先,通过分析轨道电路工作原理,初步选取10个特征参数;然后,利用ReliefF算法得到各个特征参数的权重,并按照权重由大到小依次输入到Adaboost故障分类器中,筛选出对轨道电路故障诊断最有效的特征参数;最后,将其再次输入到Adaboost故障分类器中完成故障识别。仿真结果表明:ReliefF-Adaboost模型故障诊断准确率为96.67%,相较于无特征参数筛选的Adaboost模型提升了2.50%。该模型有效提升了ZPW-2000A型无绝缘轨道电路的故障诊断性能。At the present stage,the main way of fault identification for ZPW-2000A jointless track circuit is to manually analyze the microcomputer monitoring data,which has the problems of strong subjectivity and low identification efficiency.To solve the these problems,ReliefF-Adaboost fault diagnosis model is proposed for track circuit fault identification.Firstly,10 characteristic parameters are preliminarily selected by analyzing the working principle of track circuit.Then,the ReliefF algorithm is used to obtain the weight of each feature parameter then input into Adaboost fault classifier in descending order of weight.The most effective feature parameters for fault diagnosis of track circuit are screened out.Finally,they are input again into the Adaboost fault classifier to complete fault identification.Simulation results show that:the accuracy of fault diagnosis of ReliefF-Adaboost model is 96.67%,which is 2.50%higher than that of Adaboost model without feature parameter screening,which can effectively improve the fault diagnosis performance of ZPW-2000A jointless track circuit.

关 键 词：故障诊断 ZPW-2000A无绝缘轨道电路 RELIEFF算法 ADABOOST算法 

分 类 号：U284.28[交通运输工程—交通信息工程及控制]