基于改进新陈代谢GM(1,1)的ZPW-2000A型轨道电路故障预测  

Fault Prediction of ZPW-2000A Track Circuit Based on Improved Metabolic GM(1,1)

作　　者：夏荣斌[1] 肖金山 吴永成[2] XIA Rong-bin;XIAO Jin-shan;WU Yong-cheng(Changzhou Research Institute,Lanzhou Jiaotong University,Changzhou 213000,Jiangsu,China;Gansu Industrial Traffic Automation Engineering Technology Research Center,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]兰州交通大学常州研究院,江苏常州213000 [2]兰州交通大学甘肃省工业交通自动化工程技术研究中心,兰州730070

出　　处：《兰州交通大学学报》2019年第5期67-73,共7页Journal of Lanzhou Jiaotong University

基　　金：江苏省创新能力建设计划产学研联合重大创新载体项目(BM2016004);中国铁路总公司科技研究开发计划重点项目(2016X003-H)

摘　　要：ZPW-2000A型无绝缘移频轨道电路在我国铁路线路上应用广泛,随着铁路线路向高速化、重载化的方向发展,传统的电务“故障修”及“定时修”在保证线路安全、提高运营效率及经济性等方面劣态日显.引入PHM理论,通过改进GM(1,1)模型实现ZPW-2000A轨道电路的故障预测.首先,为选取更优的预测模型,分别对传统GM(1,1)进行两次改进,即新陈代谢GM(1,1)模型与改进新陈代谢GM(1,1)模型.其次,引入预测步长思想,进一步提高模型预测精度.基于兰州西站轨道电路测试数据的算法验证表明改进新陈代谢GM(1,1)间接多步预测模型的预测结果最优.最后,通过实例验证了该模型在轨道电路故障预测中的有效性.ZPW-2000A frequency-shift track circuit is widely used in railway line.With the development of high-speed and heavy-duty railway line,traditional electrical "fault repair" and "timing repair" have become increasingly weak in ensuring safety and improving operation efficiency and economy.By introducing PHM theory,fault prediction of ZPW-2000A track circuit is realized by improving GM(1,1) model.First,two improvements were made to select better prediction model,namely the metabolic GM(1,1) model and the improved metabolic GM(1,1) model.Secondly,the idea of predicting step size is introduced to further improve the prediction accuracy of the model.The algorithm verification based on the test data of track circuit of Lanzhou West Railway Station shows that the indirect multi-step prediction model of GM(1,1) with improved metabolism has the best prediction effect.Finally,the effectiveness of the model in the track circuit fault prediction is verified by examples.

关 键 词：轨道电路 改进GM(1 1)模型 故障预测 多步预测 

分 类 号：U283.2[交通运输工程—交通信息工程及控制] U284.7[交通运输工程—道路与铁道工程]