密级： 公开本校编号：
中图分类号： U283.2UDC:
蔺州交通大学
硕士学位论文
研究生姓名： 董国庆 学号：
朱爱红 高工学校指导教师姓名： 职称：申请学位等级： 工学硕士 专业： 交通信息工程及控制2023.06.12论文答辩日期：论文提交日期： 2023.05.27
硕士学位论文
列控中心系统可靠性及安全风险 评估方法研究
Research onReliability andSafetyRiskEvaluationMethod ofOn-boardTrainControl CenterSystem
作者姓名： 董国庆学科、专业： 交通信息工程及控制研究方向： 列控系统可靠性评估学 号： 11200409指导教师： 朱爱红完成日期： 2023.06.12
摘要
列控中心（TrainControlCenter，TCC）作为中国列车运行控制系统（ChineseTrainControlSystem，CTCS）的地面核心设备，具备轨道电路编码、控制区间信号机点灯等功能.但由于列控中心的设备设置繁多且结构复杂，极大地增加了系统可靠运行和设备维护维修方面的难度.因此，为了提高列控中心的维护水平、改进后续设计以及降低故障诊断成本，对列控中心进行可靠性和安全风险评估具有重要研究意义.
由于列控中心故障数据缺乏，且对列控中心的认知水平有限，导致较难获取到准确的系统失效机理和精确的设备失效率，这使得在可靠性评估时无法避免地存在认知不确定性：列控中心在实际运行中会受到电磁、温度、噪声等环境因素的影响，导致系统的各个设备发生不同程度的性能退化，即设备的状态除了正常运行和完全失效外，会存在多个中间状态：此外，在对列控中心进行安全风险评估时，利用专家经验来获取风险数据的方法会存在较强的主观不确定性，以及在计算风险因子权重时多采用单一赋权法，使得评估结果的可参考性较低.针对以上问题，本文进行了以下工作：
(1)针对列控中心可靠性评估中的认知不确定性、动态性、共因失效、以及恢复机制等问题，本文在分析列控中心各单元功能逻辑关系的基础上，提出将证据理论和动态贝叶斯网络相结合，建立了动态证据网络(DynamicEvidential Network DEN）可靠性分析模型.通过DEN正向推理，并结合信任测度和似然测度得到列控中心的可靠度区间以及讨论共因失效对系统可靠度的影响：通过DEN反向推理得到列控中心的薄弱环节：最后，求解重要度的大小，探究底事件可靠度变化和认知不确定性对系统可靠度的影响.
(2)针对列控中心设备在实际运行过程中产生的性能退化问题，本文建立基于多状态Markov状态转移模型，并将其引入DEN中，完成了列控中心在多状态条件下的可靠性建模：并比较列控中心不考虑维修、考虑完全维修、以及不完全维修对系统可靠性的影响程度，为提高系统可靠性和维修维护提供决策支持和理论依据.
(3)针对列控中心安全风险评估结果可参考性较低的问题，首先，采用失效模式与影响分析法(FailureMode and Effect Analysis FMEA)识别TCC系统接口单元潜在的风险因素，总结故障模式.然后邀请专家对列控中心接口单元各故障模式的风险等级进行打分，在此基础上，利用证据理论对专家初始评估数据进行修正，剔除冲突数据：其次，采用博弈论综合赋权法确定各风险因子的权重，最大程度上降低单一赋权带来的信息损失：最后，对列控中心接口单元的各故障模式进行风险排序.
关键词：安全工程；可靠性评估；证据网络；认知不确定性；博奔论
Abstract
The Train Control Center(TCC) is the ground core equipment of the Chinese Trainlighting and so on. However because the TCC system has a variety of equipments andplicated structure it has greatly increased the difficulty of reliable operation equipmentmaintenance and maintenance of system. In order to improve the maintenance level of theTCC system and improve the subsequent design and reduce the cost of failure it is of greatsignificance to evaluate the reliability and safety risk of the TCC system.
Due to the lack of failure data and the limited epistemic level of the TCC system it isdifficult to obtain accurate system failure mechanism and accurate equipment failure rate.This makes it unavoidable to have epistemic uncertainty in reliability evaluation of system. Inthe actual operation the TCC system is affected by environmental factors such aselectromagnetic temperature and noise leading to different degrees of performancedegradation of various equipment of the system. In other words the system equipment haveseveral intermediate states in addition to normal operation and plete failure. In addition during the safety risk assessment of the TCC system the method of using expert experience toobtain risk data has strong subjective uncertainty. And the single weighting method is oftenused to calculate the weight of risk factors. This makes the evaluation results less referential.In response to above problems this thesis has done the following:
(1) The reliability evaluation of the TCC system have problems of uncertainty dynamicfailure mon cause failure and recovery mechanism. Based on the analysis of theapproach by bining the evidence theory with dynamic Bayesian network. And the functional logical relationship of each unit of the TCC system this thesis puts forward aDynamic Evidential Network(DEN) reliability analysis model is established. By using theforward inference ability of DEN the reliability of the TCC system is obtained by biningthe belief measure with plausibility measure of the evidence theory. The effect of moncause failure on system availability is discussed. By using the backward inference ability ofDEN the weaknesses of the TCC system are recognized. Finally by solving the magnitude ofBirnbaum and importance the influence of basic units with reliability changes and theuncertainty on system reliability are explored.
(2) During the actual operation the equipment of the TCC system will produceperformance degradation. In this thesis multi-state transition model based on Markov isestablished and introduced into DEN to plete the reliability modeling of the TCC systemunder multi-state conditions. Compare the influence degree of the TCC system reliabilitywithout considering maintenance considering plete maintenance and inplete
reliability and maintenance.
(3) The safety risk evaluation results of the TCC system have low reference value.Firstly failure mode and effects analysis(FMEA) is used to identify the risk factors of theTCC system interface unit. Then experts are invited to score the risk levels of each failuremode of the interface unit of the TCC system. On this basis the evidence theory is used toprehensive weighting method of game theory is used to determine the weight of each risk revise the initial evaluation data of experts and eliminate the conflict data. Secondly thefactor so as to minimize the information loss caused by individual weighting. Finally the riskfactors of TCC system interface unit are sorted.
Key Words: Safety Engineering: Reliability Evaluation;Evidence Network;EpistemicUncertainty; Game Theory
