胡华

入学时间：2006级

答辩时间：2009年

论文题目：城市轨道交通与常规公交协调调度理论与方法

中文摘要

摘要

为优先发展公共交通，我国各大城市正不断加快轨道交通线网及公共交通换乘枢纽的建设，部分特大城市已初步形成“以轨道交通为骨干、以常规公交为主体”的多模式公共交通网络。在传统的公交企业运营模式和调度体制下，轨道交通和常规公交的运营调度各自独立不予协调，枢纽换乘环节成为提高多模式公共交通系统“网络化”运输效率和整体服务水平的瓶颈。城市轨道交通与常规公交协调调度旨在从运行层面优化公交网络客流组织，在实践中它能够提升乘客换乘过程的连续性和顺畅性，并最终提高多模式公共交通系统的吸引力和出行分担率。

论文在分析我国大城市多模式公共交通网络客流组织中存在的问题和发展需求、总结国内外公共交通智能化调度相关成果的基础上，系统解析了轨道交通与常规公交运营协调的内涵、必要性，以及实现条件。通过研究发现，如果轨道列车和常规公交车辆的运行实现同步衔接，则可保障乘客在最短的时间内完成公共交通多方式之间的换乘，并可通过设置合理的车辆到达和出发时间间隔提高换乘协调几率。

常规公交车辆到站时间实时预测方法和换乘枢纽客流OD实时预测技术是轨道交通与常规公交协调调度的两项关键支持技术。本文基于公交车辆AVL（Automatic Vehicle Location，简称AVL）实时数据信息预测公交车辆的路段行程时间。算例分析表明，该方法提高了公交车辆到站时间预测结果的精度。由于已有枢纽客流OD预测方法立足规划层面而难以应用于短时预测，本文立足于运营管理层面，根据枢纽当前时段各条公交线路的上下客人数采用三阶段法预测下一时段枢纽客流OD，并针对第三阶段的枢纽换乘客流OD预测模型设计了改进的BP神经网络方法求解，最后通过算例验证了该方法的可行性。

轨道交通与常规公交行车计划联合优化是提高多模式公交网络整体服务水平的关键。论文以单条轨道交通线路及沿线站点与接驳常规公交线路构成的多模式公共交通网络为研究对象，以乘客和营运者广义费用最小为目标，建立了面向枢纽链的轨道交通与常规公交行车间隔基本模型和协调模型，前者仅优化行车间隔，后者优化行车间隔和松弛时间。论文应用“逐步逼近”的优化逻辑设计了四阶段方法求解这两个模型。算例研究表明，在多模式公共交通网络换乘客流量小、线路发车间隔大的前提下，从枢纽链层面对轨道交通和常规公交进行多线路行车间隔协调优化，与从单个枢纽“点”层面的协调优化相比，能够进一步降低公交网络的总成本。

轨道交通和常规公交动态协调调度是克服公交运行随机性因素影响、提高多模式公共交通系统运行可靠性的关键。在常态调度层面，针对常规公交车辆晚点、客流需求波动等不确定性因素对执行运行协调计划带来的不利影响，论文研究了面向轨道交通换乘枢纽的多线路驻站协调控制方法，给出了驻站控制的逻辑流程，并以乘客和营运者广义费用最小为目标建立优化模型，并将该模型转化为组合优化问题，设计了改进的遗传算法求解。算例分析表明驻站协调控制能够促进轨道交通与常规公交的动态换乘协调。在异常态调度层面，本文针对常规公交车辆运行出现“大间隔”或换乘枢纽客流突增等异常情况，分客流均匀和客流非均匀两种情况建立常规公交车辆实时放车调度模型，并对模型进行解析、给出优化算法。算例分析表明放车调度能够使常规公交线路的行车间隔变得均匀和减少换乘枢纽内乘客的候车时间。文章最后，简要讨论了有待深入研究的内容。

本研究成果从多模式公交网络运行层面初步研究了轨道交通与常规公交协调调度理论与方法，为进一步开发具有我国自主知识产权的公共交通智能化调度决策支持软件系统提供模型和算法支持。

关键词：轨道交通，常规公交，协调调度，多模式公共交通系统，优化

英文摘要

ABSTRACT

In order to give priority to development of public transport system, China’s big cities is continuously speeding up construction of rail transit network and transit transfer stations. Multi-modal transit network with rail transits as backbone and bus lines as main body has been basically formed in some megacities of China. However, under traditional operation modes and dispatching methods adopted by public transport corporations, the operation schedule of rail transit and bus are always independent with no coordination, which leads to too much transfer time and thus reduces operation efficiency and service level of multi-modal transit network. The purpose of rail transit and bus coordination dispatching was to optimize its passenger flow organization on the operation level of public transport network. Its application can make passenger transfer smoothly，which correspondingly improves attractiveness and trip-share ratio of public transport system.

Firstly, this dissertation analyzed existing problems and developing requirements in passenger flow organization of public transport system in China’s big cities, and reviewed domestic and foreign literatures on intelligent public transport dispatching theories and methods, systematically analyzed the meaning, necessities, and conditions of transfer coordination between rail transit and bus operations. The research results demonstrated if transit vehicles on all or most of routes or lines are scheduled to arrive at a transit center simultaneously, transfers among all the routes and lines involve virtually no waiting. Furthermore, to set up interval between arrival time and depart time in the schedule can enhance the transfer coordination probability among all the routes and lines.

Secondly, two key techniques, real-time forecasting methods of bus arrival time and passenger OD (Origin-Destination) matrix supporting dynamically coordinated dispatching method of transit rail and bus operations were studied. Bus arrival time was forecasted based on the Automatic Vehicle Location data, and an example verified the method could increase forecasting accuracy. Considering exiting forecasting methods of passenger OD matrix from planning level, this dissertation put forward a three-stage method from operation level. The method of Back-Propagation network was used to forecast passenger transfer OD matrix, and an example was given to verify its feasibility.

Thirdly, aimed to a multi-modal transit network which is composed of a rail transit line and some feeder bus routes at its transfer stations, planned transfer coordination strategies between rail transit and bus were put forward to improve service level of transit network. Two models, basic model and coordination model, were set up to minimize generalized cost of passenger and public transport corporations. The basic model optimizes headway of bus and rail train, and the coordination model optimizes both headway and slack time of bus. A four-stage algorithm was designed to solve the models. An example demonstrated scheduling coordination among the routes and lines in the multi-modal transit network could reduce the total cost.

Fourthly, in aspect of normal dynamic dispatching, buses holding control method among the routes were studied to reduce the influence of randomness of bus arrival time. Its logical procedure was given and optimization model was set up. An improved Genetic algorithm was designed to solve the model. An example indicated the method of buses holding control could promote dynamic transfer coordination between rail transit and bus. In aspect of abnormal dynamic dispatching, aimed to the phenomenon of large bus headway or sharply increasing demand at a transfer center, bus real-time deadheading model was set up and optimization algorithm was given. An example verified that the model could balance bus headway and reduce passenger waiting time. Finally, the areas for future research were briefly discussed.

The research results of the dissertation provided theoretical supports and basically formed the corresponding public transport intelligent dispatching decision-making methodologies, providing theoretical and methodological guidance for the future development of China’s own public transport intelligent dispatching systems.

Key Words：Rail Transit，Bus，Coordinated Dispatching，Multi-modal Transit Network，Optimization

马莹莹

入学时间：2006级

答辩时间：2009年

论文题目：面向交通小区的交通控制策略研究

中文摘要

摘要

随着城市的发展及人民生活水平的提高，城市交通的供需矛盾加剧，交通拥堵已成为世界性的难题。城市交通信号控制是预防和缓解城市交通拥堵的一个有效手段。交通信号控制的发展经历了单点交通控制、连线协调控制及区域控制的主要阶段。目前应用较为广泛的交通信号控制系统多为递阶结构的区域交通信号控制系统。这些系统通过将大的网络降解成小的控制区域（交通控制小区）并在小区内实施自适应的协调控制以完成整个网络的控制。然而在控制范围较大，交通状态分布不均匀的情况下，各交通小区的相互影响加强，仅考虑小区内部的最优控制往往很难获得网络的最优，此时则需要对网络内各个控制小区进行协调，以避免或缓解局部交通拥堵，达到网络效益最优。

本文研究的主要内容为递阶式交通控制系统中最上层（网络层）要完成的两项主要工作：交通控制小区的划分及小区之间的协调问题。

在交通控制小区划分问题上，本论文首先分析了相邻交叉口协调控制效益与下游交叉口到达交通流参数：流量、周期及不均匀系数（定义见第四章）的关系，建立影响关系模型，并进行仿真验证。分析结果表明，相邻交叉口的协调控制效益与协调周期、不均匀系数成线性关系，与流量成二次曲线关系。该模型将下游交叉口交通流参数与协调控制效益建立直接联系，可以作为划分交通协调控制区域的依据。其次，在论文第五章，提出了基于谱方法的交通控制小区划分方法，并以简单网络和实际网路为案例对该方法进行验证。

在面向小区的协调控制问题上，本文利用交通网络的宏观交通流特性，以整个网络到达终点的车辆数最大为优化目标，小区间转移的交通量为决策变量建立各小区交通协调模型，为下层交通小区内部的协调控制提供战略性的指导。案例表明，该方法可以平衡网络交通需求，提高网络整体效益，减少由于局部交通拥堵造成的网络效益下降。

关键词：城市道路交通网络，交通信号控制系统，交通控制小区，网络交通控制策略，宏观交通流模型

英文摘要

ABSTRACT

With the development of cities and mobility, congestion has become a worldwide problem to be solved. Traffic signal control is one of the most effective management strategies. The development of traffic signal control has experienced isolated control, corridor coordination and regional traffic control. Currently, the regional traffic control systems with hierarchical structure are used widely in the world. These systems group some adjacent intersections with similar traffic flow characteristics into one sub-network and covert a huge network control problem into several sub-network control problems in a hierarchical fashion. However, when the network work is big and some of the sub-networks are extremely congested, the optimal coordination within each sub-network could not get the global optimum of the entire network. At this time, the coordination of sub-networks is very useful to alleviate the local traffic congestion and get the best network benefit.

This doctoral research is aimed at two main contents of the highest level – network coordination level of the signal control system with the hierarchical structure: traffic signal control network partitioning and sub-networks coordination.

For the case with traffic signal network partitioning, the research developed a model to describe the relationship between signal coordination delay improvement of adjacent intersections and downstream traffic flow characteristics that includes traffic volume, cycle and uneven index R(the definition of index R is described in chapter 4) at first. The analysis of the model showed that the relationships between the coordination delay improvement of adjacent intersections and cycle length, delay improvement and index R are linear, while the relationship between delay improvement and traffic volume is quadratic. This model used downstream traffic flow characteristics to describe the coordination delay improvement of two adjacent intersections, which can help to decide whether two intersections should be coordinated and support the network partitioning. Secondly, a traffic signal control network partition method using the spectral methodology was developed in chapter 5 and was tested on two networks, including a computer-generated network and real-world road network. The results showed that this method is a reasonable network partitioning method to support urban traffic signal control.

For the case with sub-network coordination, this research used macroscopic model of traffic network to setup the optimization model, in which the objective is maximize the total number of vehicles who get their destination and the decision variables are transfer volumes among the sub-networks. The solution of this model can give a strategic optimization direction to the middle level of signal control system(signal coordination within each sub-network). Case study showed that the sub-network coordination can blance the traffic demand of the network and improve the efficiency of entire network.

Key Words:urban road networks, traffic signal control systems, traffic signal control network partitioning, sub-network coordination strategy, macroscopic fundamental diagram of traffic road network

李晓丹

入学时间：2006级

答辩时间：2009年

论文题目：城市道路网络交通小区动态划分理论与方法研究

中文摘要

摘要

大城市或特大城市的道路网络，其拓扑结构具有复杂、多维等特性，其网络交通流具有复杂性、实时性、随机性等特点，难以对城市道路网络及其交通流系统进行同一的交通控制与管理。首先，城市道路网络规模庞大，使得无法对整个网络进行控制与管理的建模；其次，网络交通流在时间、空间、构成等上的差异特性，使得无法对整个网络采取同一的控制与管理策略。因此，从网络交通流的时空特性和交通管理的动态、即时特性分析出发，需要对城市道路网络进行合理的时空划分，解耦城市道路网络及其交通管理系统，决定了对城市道路网络进行时空划分的必要性；另外，城市道路网络的拓扑特性和网络交通流的周期相似特性与空间关联特性，也决定了对城市道路网络进行时空划分的可行性。本文以国家高技术研究发展计划（863计划）项目专题“城市道路网络交通小区动态智能划分技术”和“城市交通状态智能预报与管理支持系统”两个课题为依托，围绕城市道路网络交通小区划分的关键理论和技术展开研究。论文通过大量的文献阅读和资料总结，分析研究了国内外在城市交通网络时空分解方面的研究情况，特别就交通诱导子区、交通控制子区的划分方法、模型、算法等方面的研究成果进行综述。总体而言，以往的研究主要为了解决交通控制系统中的交通信号协调控制方面的问题，对交通控制子区的划分原则、指标、方法等相对研究较多，而对交通诱导子区的研究未形成完整的交通小区划分理论和方法体系。本论文在此基础上确定了研究方向，即建立交通小区的通用概念解析及划分理论体系，并侧重于中观和微观的交通小区的划分方法和实现技术进行深入研究。在城市道路网络交通小区概念解析、交通小区划分体系设计、诱导交通小区划分方法、控制交通小区划分方法、交通小区划分系统设计等方面取得了一系列成果，具有重要的理论意义和实用价值。

首先，论文在分析城市道路网络交通流的周期相似特性和空间关联特性的基础上，提出了划分交通小区的可行性；综合考虑交通动态管理的系统需求，提出了交通小区的通用概念及数学描述。并基于交通小区的定义及划分的一般规则，研究了交通小区动态划分的影响因素、原则、依据，以及交通小区动态划分的模式和方法，初步确定了交通小区动态划分的时空规则。

其次，基于交通小区的定义，论文建立了交通小区动态划分的理论体系。在对交通管理体系分析的基础上，提出区域交通管理是城市道路网络交通系统管理的高级形式，从而根据不同的管理策略建立了交通小区的层次体系，并解析了不同层次交通小区间的相互关系；在以往子区划分指标研究的基础上，进一步扩充并建立了交通小区动态划分的指标体系；基于对城市道路网络交通小区动态划分的数据需求分析，分别对城市道路网络时空数据模型、多尺度模型、海量数据存储等方面的数据支撑体系进行了初步研究，为交通小区动态划分系统的建立和实现奠定了数据和模型基础。

再次，论文面向实际的应用需求，分别对面向路径诱导的交通小区（以下简称诱导交通小区）和面向交通信号协调控制的交通小区（以下简称控制交通小区）的划分方法进行了深入研究。基于诱导交通小区在路径诱导中的重要作用分析，确立了诱导交通小区划分的目标，进而提出了基于连线交通相似度的城市道路网络动态分层规则和基于矩形区域算法的两步骤诱导交通小区空间划分方法，并提出了诱导交通小区的周期确定方法。至于控制交通小区，首先基于系统控制目标的研究，确定了控制交通小区的划分步骤，提出了基于关键节点和路径关联度的控制基本交通小区划分方法，并采用空间聚类分析算法加以实现；在控制基本交通小区划分的基础上，进一步提出了基于车队关联度的控制交通小区动态划分和优化过程，重点研究了动态划分和优化的扩张、收缩和裂变规则，以使控制交通小区能够动态适应交通流的时变特性。

最后，论文从实现层面研究了城市道路网络交通小区动态划分系统在城市综合交通管理平台中所处的地位，对交通小区动态划分系统的功能需求、总体设计、数据库设计等方面进行了分析和设计，并基于组件式GIS实现了原型系统的开发，简单介绍了交通小区在部分城市中的实际应用情况。应用研究表明，城市道路网络交通小区动态划分系统能够有效地提高交通控制系统和交通诱导系统的效益，具有科学性和有效性。

关键字：交通小区，诱导交通小区，控制交通小区，空间划分，时间划分，交通相似度

英文摘要

ABSTRACT

For road network of major cities or mega-cities, it’s topology has characteristics of complexity,Multi-dimension; it’s traffic flow has characteristics of complexity, real-time, randomness. It is difficult for urban road network and traffic flow system to be in the same traffic control and management. First of all, the large-scale of urban road network makes it impossible to model the whole network by control and management; Secondly, the differences on region, time and constitution of traffic flow make it impossible to take a same control and management strategies for the whole network. Therefore, by analyzing temporal and spatial characteristics of traffic flow and dynamic and immediate characteristics of traffic management, it is necessary for reasonable regional division of urban road network, and decoupling urban road network and its management system. These determine the necessity of spatial and temporal division for urban road network. In addition, the characteristics of periodical similarity and spatial correlation determine the feasibility of spatial and temporal division for urban road network. This dissertation is mainly based on the research of two national 863 projects. One isDynamic and Intelligent Division Technology of Traffic Zone for Urban Road Network, and the other one isSystem of Intelligent Forecasting and Management Support for Urban Traffic Status, and it focuses on the critical theories and technologies of traffic zone division.According to massive bibliographic data firstly, the dissertation analyzed international research status of space-time decomposition of urban transportation network, and elaborated international research achievements of division methods, models, algorithms and the others of traffic induced and traffic control sub-areas especially. In general, there are many research results of traffic control sub-area division, to solve the problem of coordinated control for traffic control system, while little research on traffic induced and other sub-areas. There was not a complete theory and method system of traffic zone division. Based on the results of previous studies, direction of this dissertation was determined, that is, establishment of general concept analysis and dynamic division theoretical system of traffic zone, as well as deep research on division method and implement technology of traffic meso-zone and micro-zone. The dissertation gained a series of research achievements in some aspects such as concept analysis of traffic zone, design of dynamic division system of traffic zone, division method research of induced traffic zone, division method research of controlling traffic zone, system design of traffic zone division. It has great theoretical significance and practical value.

Firstly, the dissertation analyzed periodical similarity and spatial correlation of traffic flow of urban road network, proposed the feasibility of traffic zone division. Considered of system requirements of dynamic traffic management comprehensively, the dissertation put forward the general concept and mathematical description of the traffic zone. Based on the definition and general dynamic division rule of traffic zone, the factors, principles, basis, patterns, methods and so on, of traffic zone division were studied, and the spatial and temporal rules were initially identified.

Secondly, according to the definition of traffic zone, the theoretical system of traffic zone division was established. After analyzing of traffic management system, it considered that regional traffic management is advanced form of traffic management in urban road network. Thus, according to different management strategies, a level system of traffic zone was established, and the interrelation of different levels between traffic zone was analyzed. In view of previous research of division indicators, the index system of traffic zone was further expanded and established. In addition, based on data demand analysis of traffic zone of urban road network, the data supporting system of urban road network, such as spatial and temporal data models, multi-scale models, mass data storage were studied preliminarily and respectively, which laid data and models foundation for establishment and achievement of traffic zone system.

Thirdly, facing practical needs of application, the dissertation separately studied division method of induced traffic zone and controlling traffic zone deeply. Base on the analysis of important role of induced traffic zone in route guidance system, the dissertation established objective function of induced traffic zone division, and then put forward a two-step spatial division method: dynamic stratifying of urban road network, which is based on traffic similarity, and zone dividing of urban road network, which is based on rectangle algorithm. Then, the cycle of induced traffic zone was determined. Through establishing of the objective function of controlling traffic zone in the traffic signal coordination control system, division steps of controlling traffic zone was determined, and spatial division method of basic controlling traffic zone was proposed, which is based on crucial node and path correlation, and was realized by using the spatial cluster parsing algorithm. Furthermore, the dissertation proposed dynamic division and optimization of controlling traffic zone which is based on the motorcade correlation. Especially research foucs on the rules of dynamic division and optimization, such as expansion,contraction, fission. Which made the controlling traffic zone to adapt dynamic traffic flow.

Lastly, this dissertation studied the position of traffic zone dynamic division system on urban comprehensive traffic management platform from achieving level, analyzed and designed functional requirements, overall design, database design of traffic zone dynamic division system; developed prototype system based on ComGIS, and introduced practical application of traffic zone in some cities briefly. Applied research showed that system of traffic zone of urban road network can effectively improve the effectiveness of traffic control system and route guidance system, possessing scientific and effectiveness.

Keywords: Traffic Zone, Induced Traffic Zone, Controlling Traffic Zone, Spatial Division, Temporal Division, Traffic Similarity

王岩

入学时间：2006级

答辩时间：2009年

论文题目：道路交通安全过程分析理论与方法

中文摘要

摘要

道路交通安全是人类社会发展进程中面临的共同问题。WHO的统计数据表明，全球每年因为道路交通事故造成的伤害超过5000万人。因此，对交通安全问题的过程、机理和对策进行深入研究，努力构建一个安全、和谐的交通系统，具有重要的理论价值和现实意义。本文以国家自然科学基金重点项目和公安部应用创新研究项目为依托，在总结国内外交通安全相关研究的基础上，基于“主动安全思想”，综合运用系统工程、交通工程、交通冲突等理论方法，围绕道路交通安全过程分析中的理论和方法进行了研究，取得了一系列成果。

论文首先建立了交通安全过程研究的理论框架。在界定交通安全的概念后，通过引入“主动安全思想”，建立了交通安全4E过程，并对过程中每个阶段的输入输出特性进行了讨论；在此基础上，构建了交通安全过程研究体系框架。其次，应用该过程理论，讨论了城市道路交叉口Encounter阶段的交通冲突问题。交通冲突可分为常规交通冲突和非常规交通冲突两类。在已知交叉口交通冲突点分布以及冲突点交通到达规律的情况下，交叉口的常规交通冲突数可以采用理论计算进行预测。论文对无控交叉口交通冲突理论计算模型进行了改进，并应用该模型对交通冲突观察值和理论计算值进行了验证，证明无控交叉口在机动车交通到达服从泊松分布的前提下，机动车交通冲突理论计算值是实际观察值的无偏估计。基于交通冲突计算模型，分析了单个无控十字交叉口在不同交通到达下的交通冲突变化特性。选取了三种典型小型无控路网，采用UE分配方法和上述冲突计算模型，得出了不同无控路网形式交通走行时间和交通冲突随路网OD的变化特性和规律。再次，应用安全过程分析理论对交通系统规划中的安全问题进行了探讨。论文提出了宏观交通安全的4维描述方法，应用主动安全思想，构筑了规划阶段交通安全研究的框架，提出了主动交通安全相关的规划-设计推荐流程，对交通规划与交通安全的关系进行了初步讨论，并给出了两个研究实例。最后，对全文进行了总结，指出了论文的创新点，对有待于进一步研究的问题进行了展望。

关键词：道路交通，交通安全，交通事故，主动安全，交通安全过程，无控交叉口，交通冲突，路网交通冲突，交通暴露，交通风险，交通规划。

英文摘要

Abstract

Road safety is the problem confronted by whole world during the process of social and economic development. According to statistics of WHO, more than 50,000,000 people were injured in traffic accidents every year in the world. So there are important theoretical values and realistic meanings to study the road safety process, mechanism and countermeasures to construct a safer and more harmony traffic system. Supported by National Natural Science Foundation of China’s Key Program and Applicable Innovation Program of Public Security Ministry, based on some related road safety researches, this dissertation introducedProactive Safety Thoughtand studied the theories and methods in road safety process by applying knowledge of system engineering, traffic engineering and traffic conflict technologies. Some valuable results were achieved.

Firstly, theoretical framework of road safety process analysis was constructed. After defining what is road safety,Proactive Safety Thoughtswas introduced to construct the road safety 4E process which was composed ofExposingstage,Encounterstage,Evasionstage andEnergy Transferstage, and each stage’s input and output characteristics were discussed. After that, a study framework of road safety process was contructed. Secondly, traffic conflicts of urban intersection inEncounterstage were discussed. Traffic conflicts can be classified intoNormal ConflictsandAbnormal conflicts. With data of traffic flow arrival distribution and conflict spots of intersection, quantities ofNormal Conflicts can be predicted theoretically. The uncontrolled intersection conflicts calculation model was improved and it is proved that, when vehicles arrive at an uncontrolled intersection subjected to Poisson distribution, the vehicle conflicts values calculated by this model is the unbiased estimation of observed conflicts values by verifying the relative errors between calculated and observed data. Then the model was applied to analyze the variation of traffic conflicts in single uncontrolled intersection under different vehicle arrival conditions. After that, variation of trip time and vehicle traffic conflicts of three typical small uncontrolled road networks were studied by usingUser Equilibriumtraffic assignment method and the traffic conflict calculated models. Thirdly, road safety process analysis theory was applied to discuss the safety in traffic planning. After bringing forward the four-dimensional description of macro road safety, based onProactive Safety Thoughts, this dissertation constructed framework in transportation planning stage, recommended planning-design procedures related to road safety, discussed the relationship between traffic planning and road safety and introduced two application examples. Finally, the dissertation is summarized, the creative research achievements and other important research directions were pointed out.

Key Words: Road Traffic, Road Safety, Traffic Accident, Proactive Road Safety, Road Safety Process, Uncontrolled Intersection, Traffic Conflict, Road Network Conflict, Exposure, Traffic Risk, Transportation Planning.

朱彤

入学时间：2006年

答辩时间：2009年

论文题目：信号控制交叉口主动安全信息服务系统基础问题研究

中文摘要

摘要

道路交叉口是各方向交通流合流、分流与交汇的节点，从道路交通事故发生的地点来看，有相当一部分发生在交叉口。本文根据交通事故发生原因，基于车车、车路通信的车路协调技术发展趋势及交通安全理论研究之需要，从主动安全思想出发，提出在道路交叉口范围内，建立智能交通运输系统（ITS）子系统，对交通参与者提供交通安全信息服务。

为不断完善并最终实现信号交叉口主动安全信息服务，本研究提出了未来研究的主要阶段，并在明确各阶段任务的基础上提出了各专业研究者的跨学科合作机制。同时将其中第一阶段任务作为本文研究对象。

鉴于本研究的前沿性，需了解国内外相关研究进展并把握其发展方向。因此，在各国智能交通运输系统研究的大背景下，综述了与本文相关的各国重要研究计划，梳理出宏观发展脉络；再对系统框架与平台、模型与评价方法等核心内容加以回顾，提炼出成功的研究方法和阶段性成果，并总结了以往研究不足之处。

为补充我国智能交通运输系统框架并为后续研究提供基础，本研究在我国交通事故统计信息不完备的情况下，通过分析交通参与者行为，自上而下归纳出系统需求，设计系统功能、逻辑与物理框架。在系统框架之后补充了子系统框架，从而清晰地表述出系统信息流向。

本研究在现有技术条件下，讨论物理技术指标对于系统功能实现的影响，分别实现短程通信、交通信号控制、车载信息采集与人机交互，对各部分加以连接实现了基于车路协调技术的实验平台，弥补国内研究的不足。最后以交通控制信号信息为传输目标，测试系统整体运行情况并考察定位精度和通信质量问题。实验结果表明，该系统能够完成信号灯状态的车载实时显示，仍需要在下一步的研究中改进定位精度与通信效果。该实验系统不仅能为测试物理设备性能与基础模型适用性提供实验平台，还能为信号影响下的驾驶行为特征分析提供数据。

研究了支撑系统的基础模型与算法。考虑到以往采用车辆速度、加速度作为参数难以准确判别驾驶员闯红灯行为，增加速度变化值作为判别参数，而非仅用车速和加速度判断闯红灯行为，提出了基于二项回归的闯红灯行为判别方法，通过视频检测系统采集城市道路进口道车辆运行数据，用实测数据进行了参数估计及检验。研究表明，车辆当前速度与上游30 m处速度差或速度比最宜作为解释变量预测驾驶员停驶行为。采用反向传播神经网络方法判别闯红灯行为，获得了更为理想的判别效果，但二项回归方法相对更易实现。

鉴于以往用时空距离表述风险不符合风险基本概念及不易设置信息发布阈值等问题，提出基于概率计算的交叉口冲突点碰撞风险模型与追尾碰撞风险模型，交叉口冲突点碰撞风险模型获得的风险值与其他模型产生的计算结果相比，风险值随碰撞临近而单调升高，更符合实际交通安全状态且易于设定信息发布阈值。追尾碰撞风险模型能够表现出制动过程中的碰撞风险，使模型更具应用价值。

研究系统评价体系、指标与评价方法。引入了描述判别准确度的ROC（Relative Operating Characteristic）曲线并将其用于评价。提出评价指标不仅应包括原有的误报（Fault）、漏报（Miss）、命中（Hit）与正确（Correct）等，还应增加“所需减速度”、“有效警告率”与“实际事件的危险程度”等指标。接着以交叉口冲突点碰撞警告为例说明了基本仿真评价方法。为弥补以往仿真评价无法体现驾驶员特性、物理系统参数的缺点，设计出能够反应上述问题的仿真方法，使评价更接近实际。并用该方法评价了GPS设备发送频率和精度变化情况下的系统特性，利用微观仿真软件的二次开发模拟常见车辆运行场景实现评价结果提取。最后以闯红灯警示系统评价为例，说明了基于实测数据的仿真评价方法。

关键词：道路交通安全，智能交通运输系统，安全信息服务，系统平台测试，风险模型，系统评价

英文摘要

Abstract

Road intersection is the node for each direction of traffic flow merging distributing or crossing path. Meanwhile, intersections contribute to a relatively high proportion of accidents compared with other road section. According to the analysis of traffic accident causes, impact of vehicle infrastructure integration technology and requirement of traffic safety theory, this dissertation studies basic issue of proactive traffic safety information service system at signalized intersection. The system aims to design and establish a subsystem of intelligent transport systems (ITS) at road intersection to improve traffic safety by providing information for traffic partcipant.

First of all, system objective and scope is presented. Research stages and composition of participants are also analyzed. This dissertation focuses on the first research stage as fundamental of other stage.

After that, related research projects are reviewed to outline development trend. Research content such as architecture, test-bed, information providing and evaluation method are also in retrospect to extract research method and achievements.

In the case of lack and absent accident data for research in China, this study analyzes driver behavior to figure out system demand and design system service, system architecture and situation of processing unit. System framework and flow chart is drawn to depict data flow direction and process.

Equipment specifications are considered and equipments are connected to complete system building. Test-bed based on vehicle infrastructure integration is accomplished including GPS, traffic signal controller, on board unit, road side unit and short-range wireless communication. Performance of positioning accuracy and communication quality are tested by experiment.

After that, risk model and algorithm are discussed. For each subsystem, risk assessment and triggered threshold are discussed. Red light running behavior classification method based on Logistics regression and artificial anural network are proposed, velocity change rate are adopted as a innovating parameterData was collected from urban road. Research show that velocity deference or velocity ratio between detect spot and 30 meters upstream was the parameter which had the most strong correlation to red light running behavior.

Crossing path collision and rear-end collision risk measurement based on probability are put out to substitute for the measurement based on time different. Character of dilemma zone is also analyzed.

Evaluation method and indexes are studied.Relative Operating Characteristicare introduced for evaluating system effect. Indexes not only include Fault, Miss, Hit and Correct, but also include effective warning ratio, required acceleration and actually vehicle arrival time difference. Driver reaction time and GPS data frequency and precision are considered. Naturalistic driving data are used to evaluate red light running alarm system.

Key Words: Road Safety, Intelligent Transportation Systems, Traffic Safety Information Service, System Platform Field Test, Risk Assessment, System Evaluation

曾滢

入学时间：2006级

答辩时间：2009年

论文题目：城市道路信号控制交叉口群交通设计方法研究

中文摘要

摘要

近年来，“交通设计”被实践证明为改善城市交通的有效方法和技术。现有的城市道路交通设计研究通常从局部交通状况的改善来考虑交通问题的解决方法，而缺乏从系统上将微观与宏观交通问题良好结合的理论和方法，使它难以应对以大中型城市中心区和CBD地区为代表的较大范围的交通拥挤问题。为解决这一问题，形成适用于交通需求较大的密集信号控制道路网络的交通设计方法，本文从系统的角度出发，充分研究交叉口与其所在网络的相互影响，进而把交通设计的理念与交叉口群的思想相结合，提出了信号控制交叉口群交通设计的概念，并论述了其理论和方法，将交通设计的研究范围从单个交叉口推广到以交叉口群为代表的道路交通网络。

首先，研究了信号控制交叉口群交通设计基础问题，论述了交叉口群交通设计的基本思想、概念、内容、方法、流程及其与交叉口交通设计的区别和联系，从理论上描述了交叉口群交通设计网络建模问题，研究了它的目标与参数，给出了模型框架，初步构建了交叉口群交通设计方法体系。

然后，分别研究了交叉口群交通设计网络优化模型框架下的两个子问题：1）建立了适用于网络优化的交叉口车道功能与控制方案组合优化模型，通过对实际案例的计算和仿真分析，验证了模型的有效性，同时也证实了车道功能与信号控制之间存在协同作用；2）在交通分配模型中引入实际的信号控制机制和延误模型，考虑了交叉口信号配时的自适应调整，运用变分不等式理论，建立了基于用户均衡的非对称网络交通分配模型，并采用对角化算法求解。

此后，结合以上两个子问题，建立了交叉口群交通设计网络优化的双层规划模型，并用迭代优化算法求解，对比了交叉口群交通设计与传统交叉口交通设计的应用效果，验证了前者对提高网络效益的有效性和可靠性。该模型可以在交通设计优化过程中体现交通网络中用户路径选择行为与交通设计方案的相互影响，并可求解得出使交叉口群交通效益最大化的最优设计方案。

最后，由于交通网络的复杂性和随机性，为寻求信号控制交叉口群交通设计的最佳研究范围，研究了面向交通设计的信号控制交叉口群范围划分问题。考虑交叉口在网络交通流分布中的相互影响，提出了交通设计关联性的概念，并对它进行了量化，进而提出了面向交通设计的交叉口群范围划分方法，同时，结合交通分配的原理，对该方法进行了验证。

本论文的研究丰富和完善了交通流理论和交通设计理论，并为城市道路交通规划、管理、设计和评价提供了理论和技术支持。

关键词：信号控制交叉口群，交通设计，混合网络设计，非对称交通分配，车道功能，交叉口关联性，网络区域划分，双层规划

英文摘要

ABSTRACT

In recent years, "Transportation Design" has been proven as an effective method to improve the operation of urban transportation. The existing researches of urban road traffic design usually try to solve the traffic problems from a local point of view, but lack of a good combination of the microscopic and macroscopic traffic problems from a systematic view both in theories and methods, making them difficult to cope with large-scale traffic congestions that may occur in the center and CBD areas of large and medium cities. In order to solve this problem and construct an effective traffic design method suitable for high demand road networks with high density signals, the interaction between an intersection and the road network of it are fully investigated from a systematic point of view. With the combination of the ideas of traffic design and signal group, the concept of signal group based traffic design (SGTD) for urban intersections is proposed. In this dissertation, the theories and methods of SGTD are presented and the research scope of traffic design is extended from a single signal to road networks which take signal groups as representatives.

Firstly, the basic problems of SGTD are investigated. The basic idea, concept, main contents, methods, procedures of SGTD are presented as well as the comparison between SGTD and traditional traffic design for intersections. Then, the network modeling of SGTD are described theoretically with the study of its optimization objectives and parameters. The framework of the network analysis model of SGTD is proposed and the methodology of SGTD is constructed.

Secondly, two subproblems under the framework are studied respectively: 1) an integrative model of lane use and signal control at intersections, which is suitable for the network analysis, is proposed and analyzed with both mathematical calculation and microscopic simulation. The results show that the model is effective and there is synergetic effect between lane use and signal control; 2) more realistic control policy and delay models of signalized intersection are introduced in the traffic assignment model. Considering the responsive signal control, an asymmetric network traffic assignment model based on user equilibrium is proposed with the theory of variational inequality and solved with a diagonalization algorithm.

Thirdly, a bi-level programming model for the network analysis of SGTD is proposed and the two subproblems mentioned above are integrated into it. The model is solved with an iterative optimization algorithm and compared with the traditional intersection design method in a case study. The results show that the former is efficient and reliable in improving the traffic operation of the network. The model reported in this dissertation can reflect the interactivity between the traffic design plans and the travelers’ behavior of traffic routing in the process of traffic design plan optimization. The local traffic design plan that maximizes the performance of the system can be figured out by solving the model.

Because of the complexity and randomness of transportation network, it is necessary to find the best size of research area for SGTD. Finally, the partition problem of the scope of signal groups for traffic design is studied. Considering the interaction between intersections on the distribution of the network traffic flow, the concept of traffic design correlation between intersections is proposed and quantified. A method for partitioning the scope of signal groups is proposed based on the traffic design correlation. With the principle of traffic assignment, some procedures are designed to verify the method.

The findings of this research can enrich and improve the traffic flow theory and traffic design theory, and provide theoretical and technical supports for urban traffic planning, management, design and evaluation.

Key Words:Signal group, Traffic design, Mixed network design, Asymmetric traffic assignment, Lane use assignment, Intersection correlation, Network partition, Bi-level programming