摘要
为了更加科学地评价地铁车辆的可靠性状态,并依据评价结果指导车辆检修策略的调整,使上线车辆的可靠性处于可控状态,提出采用灰色综合评估和粒子群算法优化反向传播神经网络(PSO-BP)预测相结合,综合评价地铁车辆当前的可靠性状态。依据地铁车辆安全评价相关标准,确定地铁车辆可靠性状态评定等级、取值范围及对应状态。基于车辆各子系统当前故障数据分析获取各子系统可靠度及评分值,同时利用层次分析法确定各子系统所占车辆的权重,根据各子系统评分值及权重采用灰色综合评估法确定车辆不同可靠性等级的比重,对车辆可靠性状态进行预分析。利用BP神经网络和PSO-BP神经网络可靠度预测模型,根据历史故障数据对车辆各子系统的可靠度进行预测,并对比2种模型的预测精度。根据灰色综合评估法预分析结果和PSO-BP神经网络可靠度预测结果,综合评价车辆当前的可靠性状态。以上海地铁某车型为例,依据各子系统的故障数据进行算例分析及验证。研究结果表明:PSO-BP神经网络相比于BP神经网络预测的相对误差降低了4.39%,具有较好的预测精度。将灰色综合评估和PSO-BP预测相结合,可以更客观准确地评价地铁车辆当前可靠性状态,为深入开展轨道交通车辆的可靠性评价体系研究提供新方法与案例支持。
In order to evaluate the reliability status of subway vehicles more scientifically,and to guide the adjustment of vehicle maintenance strategies based on the evaluation results,which can make the reliability of online vehicles controllable,a grey comprehensive evaluation with Back Propagation Neuron Network optimized by Particle Swarm Optimization method(PSO-BP)was proposed to comprehensively evaluate the current reliability status of subway vehicles.Firstly,according to the relevant standards of subway vehicle safety evaluation,determine the reliability status evaluation grade,value range and corresponding status of subway vehicles.Secondly,the reliability and score value of each subsystem are obtained under the current failure data analysis of each vehicle subsystem,and the weight of the vehicle occupied by each subsystem were calculated by the analytic hierarchy process.The gray comprehensive evaluation method was adopted to determine the proportion of different reliability vehicle levels based on subsystem score values and weight to conduct a preanalysis of the reliability state of the vehicle.Then,compare the accuracy of BP neural network and PSO-BP neural network reliability prediction model for the reliability of vehicle subsystems based on historical fault data.Finally,the current reliability status of the vehicle was comprehensively evaluated according to the pre-analysis results of the grey comprehensive evaluation method and the PSO-BP neural network reliability prediction results.This paper used the certain model of Shanghai Metro as an example to analyze and verify the case based on the fault data of each subsystem.The results show that the relative error of PSO-BP neural network prediction is reduced by 4.39%compared with BP neural network,and it has better prediction accuracy.The combination of grey comprehensive evaluation and PSO-BP prediction can evaluate the current reliability status of subway vehicles more objectively and accurately,and provides new methods and case support for in-depth resea
作者
范乔
师蔚
廖爱华
胡定玉
FAN Qiao;SHI Wei;LIAO Aihua;HU Dingyu(School of Urban Rail Transit,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《铁道科学与工程学报》
CAS
CSCD
北大核心
2022年第1期239-247,共9页
Journal of Railway Science and Engineering
基金
国家“十三五”科技部重点研发项目(2016YFB0100700)。
