【作者】 蒋熙；

【导师】 杨肇夏；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2008， 博士

【摘要】 近年来,我国铁路实现了跨越式发展。从“组织型”向“规划型”转变,既有铁路行车组织模式开始了新的变革,同时,多条客运专线、高速铁路自明年起将先后开通运行,相应的行车组织模式设计也在紧锣密鼓的进行。铁路行车组织计划的制定与优化是铁路行车组织工作的核心内容,行车组织模式的改革与发展在优化目标、制约因素、精细程度、安全可靠及时效性要求等方面为行车组织计划赋予了新的内涵。基于行车组织模式及相应行车组织计划编制方法、技术及应用的研究对铁路行车组织仿真提出了一系列更高、更新的要求。在这一新的背景下,我们认为,面向行车组织模式研究的仿真已提到了议事日程上,特别是对行车组织计划决策过程和方法进行建模与仿真,将为行车组织计划的优化和可靠性研究,分析和实验新的行车组织模式,提供有力的支撑。面向铁路行车组织方式的仿真,在系统性、复杂性、智能性、运算规模等方面将发生巨大变化,同时在环境适应性要求方面也有很大提高。因此,开展这一研究将面临两个方面的困难,一是针对铁路行车组织智能决策活动的协作性和可演化性的仿真建模方法,二是具有高度可重用性和适应性的仿真系统构建方法。以往的铁路行车组织仿真一般将相应计划作为外部输入条件,属于对行车组织某些最终决策结果的仿真评估和检验,相应的研究方法难以解决这两个方面的难题,有必要探索新的途径。本文以复杂适应系统理论作为研究行车组织系统发展与演变基本规律的理论基础,并将建模和系统构建两大关键活动统一在这一理论基础下。对于行车组织计划制定这样一个复杂问题,利用基于协作任务求解方式和多主体建模方法,将行车组织计划系统的决策问题分解为多个子任务,分布到若干决策主体上,通过自主决策和交互作用实现整体决策过程。利用个体间的交互所产生的复杂多变的协作行为,能自然表现不同行车组织方式下的决策过程以支持系统演化。对于在发展演变背景下仿真系统的构建,本文围绕仿真集成,利用SOI以粗粒度松散耦合方式便于支持系统流程重组和优化的优势,在先进仿真技术的代表HLA的研究成果基础上,结合SOA架构的思想实现仿真集成单元间不同层次的互操作与重用,以探索保障系统适应性和开发效率的途径。在此研究思路下,本文主要进行了如下研究工作:1)根据铁路行车组织系统的复杂性和可演化性特征,提出以结构模型构建、单元模型构建和仿真集成三项基础活动构造铁路行车组织仿真应用的方式,建立了基于系统演化的铁路行车组织面向领域建模与仿真框架,从不同层次上实现仿真系统的可演化性,以适应行车组织仿真在发展背景下的研究需求。2)面向铁路行车组织计划制定与优化,提出以分布的运输流主体和资源管理主体通过自主决策方式实现行车组织优化的建模思想,研究了基于协作任务求解的铁路行车组织计划建模方法,建立了基于多主体的铁路行车组织系统模型框架。以“运输流”和“运输资源与资源管理”建模为重点,利用多主体间的协作来实现行车组织计划制定的决策目标。即,以车流联盟的形式实现车流间通过协作来获得合理编组方案的车流组织问题建模;用车流联盟接续网描述了列车任务规划求解中的协作关系,并建立了车流联盟主体与车流主体间的协作求解模型;针对不同类别运行资源协作实现列车任务的要求,提出了以列车任务作为协作框架的基于“惩罚金”的列车运行资源分配协同优化方法。3)在铁路行车组织多层次仿真集成需求下,引入基于服务的思想实现仿真集成单元间粗粒度的松散耦合集成,并结合HLA规范提出了铁路行车组织仿真集成机制,构建了基于MAS的仿真集成框架结构。在仿真集成框架的基础上,结合铁路行车组织仿真的领域共性,构建基于三个支撑功能层次的仿真支撑环境。4)在上述研究基础上,结合路网货车集结仿真评估应用,对定编集结和定点集结两种货车集结策略,建立多层次仿真评估模型,给出了宏观仿真算例,并设计相应的仿真系统,对本文研究方法进行实例分析。本文以探索面向铁路行车组织的仿真建模方法为目标进行研究,提出的建模与仿真框架、仿真建模方法和系统集成方法有助于解决新的发展背景下铁路行车组织仿真建模中的智能协作性、可演化性以及仿真系统构建中的适应性两个方面的难题,探索了一条能适应环境变化的铁路行车组织方式仿真研究的新途径。更多还原

【Abstract】 In the last years, Chinese railway has achieved great development, and the transportation organization pattern has been changing from "dispatching-based" to "planning-based". At the same time, many high-speed railway passenger dedicated line, will be operational in the future and the corresponding transportation organization pattern is also intensely carried out. Planning optimization is the key issue of railway operations, driving organization pattern of reform and development in the optimization objectives, constraints, sophistication, security, reliability and timeliness requirements, and other aspects for the railway traffic organizational plan to a new connotation. Organization pattern and the corresponding plans based scheduling method, technology and application make a series of higher new requirements of the railway transportation organization simulation.In this new context, we believe that the simulation aiming at transportation organization pattern has been mentioned on the agenda. The modeling and simulating decision-making process and methods, will provide strong support for research on the optimization and reliability of the traffic organizational plan, and analysis and organization of the new traffic patterns. Railway traffic organization pattern-oriented simulation will change dramatically in the systematic, complexity, intelligence, and the scale of operation, while the environmental adaptability requirements will be greatly improved. Therefore, this study will mainly face difficulties in two areas, the first is collaborative and evolutional simulation modeling methods which is focused on railway transportation organization intelligent decision-making activities, and the other one is simulation system constructing methods with high reusability and adaptability Simulations of the railway traffic organization in the past treat corresponding plans as output data. So, they are just the assessment of some scheduling results of the decision-making process. The corresponding research methods can not be used to solve the two difficult issues mentioned above, that means we have to explore new efficient ways.In this dissertation, we take complex adaptive systems theory as the theoretical basis for the study of development of railway transportation system, and unify modeling and system constructing on the basis. For such a complex scheduling issue, we realize the entire decision-making process based on the use of method of dividing collaborative tasks multi-agent modeling method, which divides the issue of the decision-making in the system of plans into multiple sub-tasks, and distributes to a number of decision-making entities, through independent interaction and decision-making process. The use of collaborative activities emerging from individual interactions, can naturally perform decision-making process under different traffic organization to support of evolution. Constructing simulation system in the context of the evolution, the dissertation on integrated simulation, uses the advantages of SOI loosely coupled way to coarse-grained support system processes to facilitate the reorganization and optimization, is based on the research results of advanced simulation technology in the HLA representatives, with SOA Simulation of the ideological framework to achieve integration between different levels of the unit interoperability and reuse, the protection system to explore the development of efficient and adaptive way. In this study procedure, the dissertation is organized as follows:1) According to the complexity and adaptability of the railway traffic organization, we presented a railway transportation organization simulation application method based on the structure model, the cell model and simulation integration. We also established the field - oriented modeling and simulation framework based on the systemic evolution of the railway traffic organization, which can achieve the evolution of the organization from the different levels of the simulation system, to meet the simulation research requirements of the traffic patterns in the context of changes.2) We presented a modeling method to realize the railway organization optimization via distributed transport entities and resource entities in self-decision making way. We studied the modeling method based on the decomposition of collaborative tasks and constructed multi-agent based railway transportation organization system framework. "Transport stream" and "transportation resources and resource management" model are treated as the two key points to realize plan scheduling through the collaborations among multi agents. That is to vehicular traffic Union in the form of traffic flow through the collaboration between the grouping to obtain reasonable program on the question of traffic flow modeling; subsequent alliance with the traffic stream train network described in the mission planning for the collaborative relationship and the establishment of the Union the main traffic flow and traffic flow for collaboration among the main model for different transport resources to achieve collaboration trains tasks, to put forward transport resource allocation methods based on the "compensation" synergy. 3) Under a multi-level integration simulation demand of railway traffic organizations, the dissertation introduces the services-based idea to achieve loosely coupled integration of coarse-grained simulation of integrated modules, proposes railway traffic simulation integration organization mechanism combined with HLA norms, and builds Simulation Integration Framework structure based on MAS. On the basis of simulation integration framework, combination of rail travel organizations simulation common areas, it builds simulation support environment which is based on the three levels of support functions.4) Based on the above studies, the dissertation combines the applications of assembly simulation of network truck, establishes a multi-level evaluation simulation model aimed at truck assembly strategy which is pointing assembly and assembly, gives the macro simulation assessment arithmetic, designs the corresponding simulation system, and analyzes the methods of examples.This dissertation studies the organization of the railway traffic-oriented simulation modeling methods, in the text of the framework for modeling and simulation, simulation modeling method and system integration methods help to solve the difficult problem of rail traffic simulation applied research organization in the context of the new development, to support demands of study of railway traffic organization which the changes adapt to the changes of outside world.更多还原