【作者】 李国正；

【导师】 谭南林；

【作者基本信息】 北京交通大学 ， 安全技术及工程， 2013， 博士

【摘要】 地铁列车快速、稳定、便捷、低噪、环保,是最佳的大众交通运输工具。近年来,我国城市地铁的发展进入了飞速发展时期,有30多个城市正在筹建,投资超过万亿,为现代化城市建设提供了坚强的运输保障。地铁列车是运输乘客的载体,是地铁运输的核心组成部分,必须提高可靠性、可用性,降低风险和维修费用。目前,地铁列车的运营管理存在以下突出问题：依靠历史经验制定维修策略,依赖检修人员的主观判断进行故障诊断,诊断系统依靠国外技术引进,缺乏统一平台管理数据和制定检修、维护方案。本文依托“广州地铁故障诊断系统和专家系统研究”项目,根据地铁列车的系统结构,对车载设备的可靠性、可用性、可维护性和安全性(Reliability Availability Maintainability Safety, RAMS)评估理论开展系统的研究,建立和完善相关数学模型,提出决策方法,解决维修方式不当、维修策略不合理、故障定位困难的问题,开发了故障诊断系统和专家系统并已投入实际应用。本文的主要研究内容包括：(1)研究地铁列车的常见故障和各系统的失效组件,从RAMS角度提出影响制定维修策略的9个指标,采用改进型层次分析法(Analytic Hierarchy Process,AHP)和蒙特卡罗法分析相关数据,建立组件关键度定量评估体系,为确定组件的维修方式提供依据。(2)结合现场数据的特点,建立组件可靠性模型,提出定量评估系统可靠性和安全性的方法,为建立系统的平均可靠性模型和制定维修策略提供理论基础。并将提出的方法应用于广州地铁一号线的故障数据分析与处理,得出该线路地铁列车部分关键组件的可靠性模型,准确评估了其子系统的可靠性和安全性。(3)分析地铁列车的寿命周期费用和维修活动的特点,研究不同维修级别与组件可靠性、可用性、安全性和维修费用之间的关系并建立模型,将维修策略的制定转化成为一个多目标的优化问题,提出一种改进型的混沌自适应进化算法,优化计算该问题。改进后的算法与多目标粒子群算法(Multi-Objective Particle Swarm Optimization, MOPSO)和多目标快速非支配排序(Non-Dominated Sorting Genetic Algorithm, NSGA2)算法求得的解集相比,在Pareto前端分布上更为均匀,种群保持多样,全局搜索能力更强。文中以可靠性和维修费用为优化目标,以可用性和安全性为约束条件,对维修策略进行优化,得出维修计划最优解集,为制定合理的维修策略提供参考。(4)在对现有诊断系统中存在问题分析的基础上,设计和开发了广州地铁一号线地铁列车故障诊断系统和专家系统。设计的诊断系统随车记录故障信息和车辆状态；专家系统中建立了故障数据库、案例库、知识库和模型库,数据库规范了RAMS现场数据的记录方式,案例库辅助建立案例模型,为故障诊断积累经验。结合实际“紧急制动故障”案例,分析了待检案例检索和匹配的过程,系统实现指导定位现场故障原因,并给出检修建议,有效提高检修效率。文中设计的故障诊断系统和专家系统在成本、性能、可靠性等多项参数上具有优势,并通过相关国家和欧洲行业标准的型式试验,已投入使用,填补了国内相关设备研制的空白。更多还原

【Abstract】 Metro train is one of the best public traffic modes, which is convenient, stable, fast, low noise and clean. In order to solve urban traffic jams, urban metro has become an important trend. In recent years, the development of China’s urban metro has entered a rapid development period. There are more than30cities planning to build metros and the investment on metro will be more than1trillion, which will improve the city transportation greatly. Metro train, the core part of the metro transportation, is the carrier of passengers, its reliability, availability are must be improved to reduce the maintenance costs and risks. Currently, the metro train operation management exist the following problems:the maintenance strategy for the metro train is formulated based on experiences; the fault diagnosis of metro trains mostly relies on the subjective judgment of the maintenance staff, diagnostic systems rely on foreign technology, as well as, the management platform of data and maintenance program are lacked.This thesis is carried out based on the project:development of Guangzhou Metro fault diagnosis and expert system. According to the metro train system architecture, equipment reliability, availability, maintainability and safety assessment theory are researched, the relevant mathematical models are established and improved. Further more, the problem of improper maintenance and fault location are solved, a fault diagnosis system and expert system are developed which have been put into practical application. The main research contents include:First of all, the common failures of metro trains and the failure components of equipment are counted and classified. Moreover, using the improved AHP (Analytic Hierarchy Process) and Monte Carlo method analysis of related data, establish a quantitative evaluation system of components, provide the basis for determine the component maintenance way.Next, this thesis analyzes the characteristics of the component running data, and further proposes the modeling method of the component reliability and safety. These methods provide a theoretical basis for establishing average reliability model of the system and formulating the maintenance strategies. Method was applied to fault data analysis and processing, some components’ reliability models are obtained. With the established model, the reliability and safety of the equipment can be evaluated accurately. Following, the influences on the reliability, availability, safety and maintenance costs, generated by the maintenance interval and repair modes, are analyzed. And, the multi-parameter dynamic model of the RAMS is proposed in the thesis. More important, the thesis proposes an improved chaotic adaptive evolutionary algorithm, by which, the multi-parameter dynamic model can be solved and the maintenance strategy can be obtained. Compared the solution set which obtained by the improved algorithm, Multi-Objective Particle Swarm Optimization (MOPSO) algorithm and Non-Dominated Sorting Genetic Algorithm (NSGA2), the improved algorithm obtained solution set is better. This thesis made reliability and maintenance cost as the optimization goal, availability and security as constraint conditions, optimize the maintenance strategy. The optimal solution set provide reference for reasonable maintenance strategy.Finally, On the basis of analyzing the problems existing in diagnostic system, the thesis designs and develops a set of fault diagnosis system and expert system for the metro train. Fault diagnosis system is used to record what information and vehicle status; in the expert system, the fault database, case base, knowledge base and model base are established. Database standardizes the RAMS data recording way, case base is to help establish case model, accumulate experience for fault diagnosis. Combined with the "emergency brake fault" case, the case retrieval and matching process are analyzed. The system realizes the fault location, and gives maintenance recommendations, effectively improve the maintenance efficiency. Fault diagnosis and expert system designed in this paper have advantages in low cost, good performance, high reliability and so on, and have been passed through the relevant national and European industry standard type tests, and used in practice. The fault diagnosis and expert systems fill the domestic blank at related equipment.更多还原