基于MK-RVM的地铁列车继电器剩余寿命预测  

Prediction of remaining useful life of metro train relays based on MK-RVM

作　　者：马垚 胡新杨 刘志强[2] 魏秀琨[2] MA Yao;HU Xinyang;LIU Zhiqiang;WEI Xiukun(Beijing MTR Operation Co.,Ltd.,Operation No.2 Branch,Beijing 100043,China;State Key Laboratory of Advanced Rail Autonomous Operation,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京市地铁运营有限公司运营二分公司,北京100043 [2]北京交通大学先进轨道交通自主运行全国重点实验室,北京100044

出　　处：《现代城市轨道交通》2025年第3期71-79,共9页Modern Urban Transit

基　　金：自然科学横向项目(I23L00270)——继电器可靠性评估及剩余寿命预测研究。

摘　　要：继电器作为地铁列车电气系统的关键部件,其可靠性直接影响地铁列车的安全运行。因此,对地铁列车继电器的剩余寿命(RUL)进行精准预测尤为重要。文章提出一种基于多核相关向量机(MK-RVM)的继电器剩余寿命预测模型,并通过贝叶斯优化算法对相关超参数进行优化。该模型能够基于预测得到的RUL值和方差获得累积分布函数,进而估算在当前开合次数下继电器至少还能运行一段时间的概率。同时以A型号继电器为例,通过搭建的继电器特性参数测试实验台采集10种时间特征参数,并基于筛选的关键特征参数进行剩余寿命预测实验。实验结果显示,A型号继电器两对触点RUL预测的平均均方根误差为48966,即预测误差约为4个月,充分证明MK-RVM方法在继电器剩余寿命预测中的有效性。As a critical component of the metro train electrical system,the reliability of electromagnetic relays can directly impact the safe operation of metro trains.Therefore,accurately predicting the remaining useful life(RUL)of metro train relays is of paramount importance.This article proposes a relay RUL prediction model based on a multi-kernel relevance vector machine(MK-RVM)and optimizes the relevant hyperparameters using a Bayesian optimization algorithm.This model can derive the cumulative distribution function(CDF)based on the predicted RUL value and variance,and estimate the probability that the relay will continue to operate for a certain period given the current number of switching cycles.At the sametime,using a type A relay as an example,ten of its time characteristic parameters were collected by a relay characteristic parameter test rig,and RUL prediction experiments were conducted based on the selected key characteristic parameters.The experimental results show that the average root mean square error(RMSE)of the RUL predictions for the two pairs of contacts of type A relay was 48966,indicating a prediction error of approximately four months,which fully demonstrates the effectiveness of the MK-RVM method in the RUL prediction of electromagnetic relays.

关 键 词：地铁 列车继电器 剩余寿命预测 相关向量机 MK-RVM 

分 类 号：U284.71[交通运输工程—交通信息工程及控制]