基于粒子群支持向量机的轨道电路分路不良预测方法  

Prediction of Shunt Malfunction of Track Circuit Based on PSO-SVM

作　　者：张梦琪[1] 赵会兵[1] 孙上鹏[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：《铁道学报》2015年第10期68-74,共7页Journal of the China Railway Society

基　　金：中国铁路总公司科技研究开发计划(2013X012-A-1)

摘　　要：轨道电路分路不良是影响铁路运输效率和运营安全的重要因素。本文基于ZPW-2000型轨道电路理论模型分析了分路不良对机车感应电流信号的影响,采用小波分解与重构算法对正常和分路不良情况下分路电流信号的细节分量进行分析与比较,在此基础上提取了分路不良预测所需的特征参量,通过粒子群参数优化的支持向量机模型实现分路不良的预测。通过对比实验结果表明:本方法能够有效预测轨道电路分路不良现象,预测正确率可达99.5%,高于其他方法。Shunt malfunction of track circuit is an important factor that affects the efficiency and safety of railway transportation.In this paper,the influences of shunt malfunction on induced current signal of cab signal have been analyzed based on ZPW-2000 track circuit theoretical model.The wavelet decomposition and reconstruction algorithm was used to compare and analyze the detail signals of the shunt current both under the normal condition and shunt malfunction condition,whereby characteristic parameters required by shunt malfunction prediction were extracted to predict the shunt malfunction by using the PSO-SVM model.Experimental results showed that this method can effectively predict the shunt malfunction of track circuit with prediction accuracy rate as high as 99.5%,which is higher than other methods.

关 键 词：轨道电路 分路不良 机车信号 粒子群优化算法 支持向量机 

分 类 号：U284.2[交通运输工程—交通信息工程及控制]