【作者】 熊志华；

【导师】 邵春福；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2006， 博士

【摘要】 国家中长期科学与技术发展规划纲要(2006-2020)，将“发展一个体系，解决三大热点问题”作为交通科技发展战略任务，即发展现代综合交通体系，解决交通能耗与污染、交通安全和大城市交通拥堵三大热点问题。在道路交通数据采集成为现实和人们追求高出行质量的今天，提高交通参与者的出行效率和可靠性、最大限度地利用道路资源，以解决城市交通拥堵问题至关重要。本论文从可靠性角度考察城市道路网的运行质量，利用可靠性理论深入研究道路网的可靠性，建立道路网可靠性模型，以理论和实证分析相结合的方法，定性和定量分析道路网的可靠性问题，为道路交通网络的运营与服务提供理论基础。首先，总结了目前包括道路网在内的一般系统可靠性指标及评价方法，给出基于海量交通数据分析研究行程时间可靠性的可行性。第二，在总结行程时间可靠性分析的理论框架和实现的技术手段的基础上，按照目标层、准则层和指标层的层次关系，建立了行程时间可靠性指标体系，该指标体系适用于满足不同服务对象以及不同路网层次和时间变化的要求，并量化指标。应用该指标体系评价北京市二环东直门桥至朝阳门桥区间的行程时间可靠性，提出了相应的出行策略。第三，综合现有各种随机网络理论方法和实证研究的结果，建立了道路网随机波动服从Beta分布的一般随机道路网可靠性模型。从实用性的角度出发，改进了随机路网可靠性模型参数，并提出了Frank-Wolfe算法和连续平均算法(MSA-Method of Successive Average)结合求解算法。第四，通过各种可靠性试验，研究行程时间可靠性随路段行程时间的波动、用户感知误差、交通需求和路段通行能力随机性等的变化规律，提出了提高可靠性的措施。第五，根据实际交通场景，按有无交通流检测设备将路段分成两类，改进了随机路网的可靠性评价模型，并参考随机路网模型算法进行求解。第六，从路径、OD对、道路网等不同层次，研究了行程时间可靠性与通行能力可靠性之间的关系，建立了交通需求、服务水平和可靠性之间的关联模型。最后，建立了基于Wardrop平衡的混合随机路网模型，结合随机路网模型进行求解，并进行了算例分析。本论文的主要创新性成果如下：1、建立了行程时间可靠性的评价指标体系，并进行了指标度量。本指标体系分为目标层、准则层和指标层三个层次，立足于满足出行者和管理者决策需求，从时空变化角度分析行程时间变化规律和取值范围。2、建立了从海量数据进行行程时间可靠性评价的技术指导。与以往的研究相比，能够充分利用现有资源，并作为路网可靠性评价的校验标准。3、重构了随机路网模型，用于评价行程时间可靠性。根据实证性研究结果，提出了路段行程时间波动服从Beta分布的随机路网理论描述模型。比较模型参数的实用性，改进参数并在此基础上，设计了Frank-Wolfe算法和连续平均算法(MSA-Method of Successive Average)结合的求解算法；通过减少内循环次数，改进了算法，并基于模型进行了参数影响和指标关联分析。4、建立了考虑行程时间可靠性的混合交通流分配模型。将可靠性引入混合交通模型，其中有信息出行者的路径选择标准为行程时间可靠性最大，而无信息出行者则以感知负效用最小为依据。在考虑路段间相关性的基础上，对路段和路径行程时间可靠性进行了数学描述。更多还原

【Abstract】 The national medium-and-long-term planning for development of science and technology (2006-2020) issues supporting polices as strategic task of transportation science and technology. Those are to develop a modern comprehensive transportation system and to solve three crux transportation problems, which are energy dissipation and pollution, traffic safety and traffic jam in metropolis. The efforts trying to ease traffic congestion are highlighted through enhancing trip efficiency of travellers, reliability and maximum utilization road resources in the light of the fact that feasibility of data collection and travellers’ expectation of high travel quality. In this dissertation main theories and methods of travel time reliability were discussed to evaluation the performance of road network. Based on reliability theory, the road network reliability model was proposed and the theory support about performance and service of road network can be obtained by using the method of combining theory with reality, processing with qualitative and quantitative analysis.The dissertation was made up of the following parts:Firstly, based on the previous related literature studies on reliability index and model, the guidance about reliability analysis dealing with large quantities data was proposed. Secondly, in view of travel time reliability lack of effective description, an integrated index system was put forward, including target hierarchy, rule hierarchy and index hierarchy, which were used to satisfy various requirements. The index system took a certain section from Dongzhi Men to Chaoyang Men as example, and some conclusions about trip tactics can be obtained. Thirdly, compared with the previous models and evidence study, stochastic network reliability model where link travel time obeyed beta distribution was proposed. In practicality, the model parameters were improved and the model can be solved by combined Frank-Wolfe method and Method of Successive Average. Fourthly, the effects of multi-factor including link travel time, perceptive error, traffic demand and capacity on travel time reliability can be estimated by performing various reliability trails and measures to improve reliability can be obtained. Fifthly, there were two types of link on the road network according to whether it contained the vehicle detectors or not. The stochastic network model was improved and corresponding algorithm can be obtained. Sixthly, the relation between travel time reliability and capacity reliability was researched from various network hierarchies, involving path, OD pair or the whole network. Finally, hybrid model based on Wardrop principle and corresponding algorithm was enacted. The following innovative conclusions were drawn through the studies on the travel time reliability index system, its reliability model and the application of reliability:1、 The principle of indices was put forward and the index system was set up based on the research into road network. The indices included three-layer infrastructure for travel time reliability assessment, which were target hierarchy, rule hierarchy and index hierarchy respectively. The index system helped reflecting properties of road network where users can find their suitable service from various spatio-temporal frameworks.2、 Previous attempts to estimate travel time reliability were contribute little to deal with numerous detected data. Therefore, the guideline for travel time reliability evaluation was introduced. This was not only redounding to take full advantage of current road resources but also to offer reliability of existing road network as checkout criterion for reliability design.3、 After being compared with previous efforts to model reliability, stochastic reliability model was reconstructed where link travel time obeyed beta distribution from empirical study. The model parameters can be simplified in practically and corresponding algorithm combined Frank-Wolfe method with Method of Successive Average was presented. The efficiency of algorithm can be improved when the inner cycle was reduced. To assess the performance of road network, lots of reliability trails were performed.4、 Hybrid model was proposed which was made up of two parts: when travellers supplied with information, route choice criterion in one model was to minimize the disutility associated with a route; otherwise, to maximize the travel time reliability of a route. The dependence between links was considered to model travel time reliability.更多还原