基于变分模态分解和能量谱的轨道电路分路不良故障监测

在传输线理论和四端口网络的基础上,建立ZPW-2000A型轨道电路机车感应信号电流模型,仿真得到正常和分路不良逐渐严重的5类信号,并对比分析正常和分路不良信号的曲线特征。为实现轨道电路分路不良故障监测,运用双指标法确定变分模态分解的模态数K;用变分模态分解算法将5类信号分解成K个固有模态,并提取能量谱特征和传统时域特征;将提取的故障信号特征输入到粒子群支持向量机(PSO-SVM)中,实现分路不良的故障监测,且使用能量谱特征的监测精度大于使用传统的时域特征。实验结果表明:使用变分模态分解算法能够有效分解轨道电路正常和分路不良的信号,便于分路不良故障特征的提取;能量谱特征集较于传统时域特征集,更利于故障分类。

Prediction of Shunt Malfunction of Track Circuit Based on Variational Mode Decomposition and Energy Spectrum

Based on the transmission line theory and two-terminal pair network,this paper builds an induced current signal model,gets five types of signals by simulating the normal condition and shunting malfunction of track circuit, and compares the signal of normal state with the signal of shunting malfunction. In order to realize the monitoring of shunting malfunction, this paper firstly uses variational mode decomposition todecompose signals into K intrinsic modes based on mode K of variational mode decomposition determined by double index method. Then,characteristic parameters of energy spectrum and time domain are extracted. Finally,PSO-SVM is used to monitor the fault type,and the accuracy by using characteristic parameters of energy spectrum is higher than that by using characteristic parameters of time domain. Test results show that VMD can effectively decompose the signals of normal track circuit and shunting malfunction so as to extract the characteristics of shunting malfunction. The characteristic parameters of the energy spectrum are more effective than those of the time domain in faults classification.