林瑜
入学时间：2003年
答辩时间：2006年
论文题目：信号控制交叉口群交通阻塞机理解析方法
中文摘要
摘要
城市道路信号控制交叉口群是关联性较强的若干信号控制交叉口的集合，是交通管理及协调控制的最小网络，也是研究复杂交通系统的基本单元，其交通阻塞对于诱发大范围区域交通拥堵具有关键性作用。因此，以信号控制交叉口群为对象，解析交通阻塞现象及其影响因素，探索交通拥堵状态演化规律，具有非常重要的意义。城市道路交通阻塞现象的解析属复杂性问题，所以首先研究其解析方法是非常必要的。
本论文首先研究了信号控制交叉口群的边界范围的界定方法，建立了信号控制交叉口群关联的有向图结构模型，并提出了综合考虑交通流路径关联和道路网络拓扑结构的交叉口群范围划分流程。在此基础上，归纳研究了信号控制交叉口群的主要构成元素：交通信号控制、交叉口与道路路段，并应用离散事件动态系统（DEDS）建模理论，建立了信号控制交叉口群的混合petri网模型，该模型具有既能够全面分析交通阻塞现象又便于计算机软件程序实现的优点。
论文针对间断交通流阻塞的量化问题，提出了阻塞度的概念，并应用模糊推理方法建立了阻塞度的量化模型。同时，深入研究了信号控制交叉口群交通阻塞的影响因素：交通流条件、几何条件、交通控制条件等，并应用正交试验方法对影响因素进行显著性分析。
最后，本论文基于自动控制理论的传递函数思想，研究了阻塞度随交通需求的变化规律，并从中微观层面的交通供给条件改善的角度出发，应用交通工程的基本方法论，阐述了运用交通组织、交通设计、交通信号控制等措施缓解信号控制交叉口群交通阻塞的思路。
论文较全面地研究了信号控制交叉口群交通阻塞机理的解析方法，特别在信号控制交叉口群的划分方法、交叉口群混合petri网建模方法、间断交通流阻塞度的量化方法、阻塞因素分析方法以及阻塞度随交通需求变化的研究方法方面具有创新性。
关键词：信号控制交叉口群，阻塞机理,关联度，混合petri网模型，阻塞度，正交试验，传递函数，交通阻塞改善方法
邵海鹏
入学时间：2003年
答辩时间：2006年
论文题目：交通语言系统基础问题研究
中文摘要
摘要
随着人们出行活动的日益频繁和出行环境的日益复杂，人们的出行对信息的依赖越来越强；同时交通管理者也需要将交通管理信息和出行服务信息更有效地提供给出行者。由于交通信息的提供不当，人们出行过程中不能有效地获取有用信息而引起无效的绕行和不稳定的驾驶行为，不但大大降低了出行效率和安全性，而且产生的无效交通会加剧交通的阻塞。因此，客观上交通系统需要一个平台以实现交通管理和服务信息的传递，使其易于被出行者获取和理解，并通过交通信息的管理和导向作用调整和均衡交通需求。另外，面向标志、标线等设施的传统的研究方法难以准确和深刻地描述交通系统中的信息流现象并解析其对交通系统的作用机理。因此，本论文提出一种新的研究思路，以交通管理和服务信息的传递工具为研究对象，探索交通信息的表达、传递和作用机理。
论文首先基于自然语言向艺术、工程科学等应用领域转移的规律，结合对标识系统的应用研究，提出交通语言的概念。借助语言系统的结构化和规则化的方法，通过形式类比，将交通系统中的信息的传递、交通管理者与出行者之间的交互作用以语言交流的形式来比拟。给出了交通语言系统的学术定义：交通语言系统是交通管理者和出行者之间进行信息交流的工具，是以颜色、形状、符号和文字等为基础的元素体系和规则体系的集合。
信息是交通语言的本质属性。根据信息论的基本原理，出行者通过接收信息而减少或消除出行过程中对路径选择、驾驶操作等的不确定性，并作出决策。因此，将广义的信息传递模型应用于交通语言的信息传递过程中，并基于信息的定义提出了交通语言信息的度量方法。在对出行者的认知、记忆特性和信息接收规律研究的基础上，分析了驾驶员的信息处理能力，并结合论文提出的信息密度的概念，给出了信息过载的判断方法。
出行者的信息需求特征很大程度上取决于其采取的交通方式，因此论文按交通方式对出行者的信息需求进行研究。对于机动车交通方式，分为出行前、出行中和出行结束时刻三个阶段，论文着重研究了出行末端的目标定位和搜索行为，并用概率论的方法建立了基于道路名称和门牌号的选择模型；对于公交出行方式，将管理和服务信息分为外部导向和内部指引两类，并通过调查分析研究了其信息表达方式和需求规律；对于慢行交通和公共建筑内部的步行交通，提出了导向信息的三要素（目标、距离和方向），并在慢行交通出行过程分析的基础上建立了出行时耗模型。
接着，论文对目前国内外交通语言设施应用状况及存在的问题进行深入剖析，总结了各种功能的交通语言设施的组成规律，提出了交通语言系统的六大基本元素：颜色、符号/图形、数字、文字、位置和尺寸等，并分别从各元素自身属性出发分析了其对出行者心理、生理方面的影响。根据交通语言所表达的交通管理含义，提出了交通语言约束强度的概念。
最后，基于系统构成的理论和方法，提出交通语言系统结构框架。从空间上，参考道路等级划分方法和各等级道路衔接规律，按照交通方式将交通语言系统层次划分为；从逻辑上，分为内容、规则/关联以及表现三个层次。对应于语言系统的基本概念，提出了交通语言系统语法单位及其组合和聚合规则。提出了三种组合规则：并列、顺接和主从组合。基于设施间信息关联的思想，提出了交通语言关联性和关联度的概念，并建立了交通语言连续性信息模型和互补信息模型。
论文从一个全新的视角对交通系统中信息流的运动规律进行探索，在提出交通语言系统的定义，交通信息需求和交通语言系统结构的构筑方面具有创新性。论文的成果可以为交通语言设施的规划和设计提供基础的理论支持和方法指导。
关键字：交通语言；交通管理；出行者信息；信息传递；信息需求；交通行为；交通心理；规则
英文摘要
ABSTRACT
With the increase of travel demand and the complexity of traffic environment, travelers depend more and more on traveler information. And also, traffic manager need to provide information to travelers more efficiently. Due to improperly provided information, travelers can not obtain useful information and redundant traffic as well as insecure drive behavior comes out. It reduced the efficiency and security of travel and contributes to congestion. So a platform for management and service information transferring is needed, which makes information easer to be obtained by travelers, and regulate traffic demand. Traditional research method based on traffic control devices cannot accurately describe the phenomenon of information flow in traffic system, and means by which information flow influences traffic system is also hard to analyze. The thesis proposed a new research method, which take information transferring tool as object, and explore the mechanism of information expressing, transferring and functioning.
Firstly, the transferring of nature language to related application fields was studied, with the research of signing system being considered, the concept of Traffic Language System was proposed. Analogize from the structural and regular characteristics of natural language, the transmission of information as well as the interaction between traffic managers and travelers were described by means of language communication. The definition of Traffic Language was proposed: Traffic Language System is a tool through which traffic managers and travelers communicate; it is the set of elements and regulations based on colors, figures, symbols and words.
Information is the essence of Traffic Language. According to the principle of information theory, travelers receive information and reduce their uncertainty in decision making for route choice and drive tasks. So, the general information transmission model was applied into the transmission of traffic language information, and the measurement method for traffic language was proposed based on the definition of information. The cognitive, memory and information reception mechanism was studied, and driver’s ability in information processing was analyzed, with information density being considered, the estimation method for information overload was proposed.
Traveler information demand depends to a great extent on the means by which thy travel, research of travel demand was conducted respectively. For vehicles, pre-trip, en route, and the end-trip information was analyzed. The Object search behavior at the end of trip was studied in detail. And probability choice model was established based on street names and doorplate numbers. For public transit users, traffic management and service information was classified to Outer guide and inner induce, the information express method and information demand was studied though survey. For pedestrian and bicycle traffic including in public buildings, three elements (Object, Direction and Distance) were proposed as part of guide information, and an time spending model was established based on trip analysis.
Than, the state of practice of traffic language devices and the problems were analyzed, the combination mechanism of traffic language device was summarized. Six fundamental elements of traffic language system were extracted from a great amount of devices: Color, Symbol/Figure, number, words, location and dimension.. Psychological and physiological influences of the elements on travelers were studied. According to the meanings of information that being expressed, the restriction of traffic language on traffic behavior was proposed.
Finally, the framework of Traffic Language System was established based on the method of system construction. The system was spatially divided into layers according to road network characteristics, and logically divided from the combination mechanism into three layers: content, relationship and behavior. Corresponding to language system, two regulations were proposed: Syntagmatic Regulation and Paradigmatic Regulation. Three combination regulations was proposed: Paratactic, Sequent and Subordinate. Based on the thought of information relevancy, traffic language relevancy and the relevant index were proposed. Continuity model and supplementary model of traffic language information was developed.
The thesis explored the mechanism of information flow in traffic system. The thesis innovated in defining of Traffic Language System, analyzing of traveler information demand and establishing of the framework of Traffic Language System. The production of the thesis can serve as theoretical base and technical guidance for planning and design of traffic language system and devices.
Keywords：Traffic Language System; Traffic management; Traveler information; Information transmission; Information demand；Traffic behavior；Traffic psychology；Regulation; Continuity
杨晓芳
入学时间：2003年
答辩时间：2006年
论文题目：道路交通连续流与间断流整合控制机理及方法
中文摘要
摘要
城市高速道路的入口匝道和出口匝道处一般易成为高速道路的物理瓶颈，由于其重建或改建的困难性，采用管理与控制将是改善其交通阻塞的关键措施。高速道路出入口匝道处出现的交通阻塞主要表现为“进不去，出不来”。所谓“进不去”又有两种情况即：（1）入口匝道通行能力有限不能满足流入的交通需求，导致流入交通阻塞；（2）入口匝道尚有相当的通行能力，但因流入交通受控于衔接交叉口，入口匝道通行能力不能得到有效利用。“出不来”是由于衔接交叉口的疏散能力和出口匝道需求不匹配，造成出口匝道车辆超长排队延伸至高速道路主线。
本学位论文针对高速道路匝道处经常出现的“进不去，出不来”的交通拥堵现象，深入解析高速道路与普通道路衔接处连续流与间断流相互作用的机理，分析拥堵产生的原因。
针对高速道路入口匝道“进不去”的问题，提出在入口匝道处采用ALINEA控制方法，保证高速道路主线的畅通，并利用衔接信号控制交叉口来控制入口匝道的交通需求，提出在入口匝道存在超长排队时，提前结束衔接交叉口关联相位的控制策略，而在普通道路压力大且高速道路主线存在剩余通行能力时提前激活衔接交叉口关联相位的整合控制策略。
针对高速道路出口匝道处“出不去”的问题，在出口匝道存在超长排队时，且交叉口关联相为当前非执行相位时，提出在保证其他流向损失最小的条件下提前激活关联相位绿灯的控制策略；而在交叉口关联相为当前正在执行相位时，在保证其他流向损失最小的条件下延长关联相位绿灯控制策略。在采用合理的控制方法，仍然不能解决匝道处的拥堵时，提出采用VMS诱导车辆选择其他路径。
文章采用CTM模型实现匝道及衔接道路交叉口的整合控制，并对该模型在入口匝道处和信号交叉口处进行了改进。
本研究成果适用于目前交通信息采集和处理技术条件下的城市高速道路与关联普通道路的整合控制，并且能够随着相关技术的发展而得到进一步完善。
关键词：城市高速道路，入口匝道控制，连续流，间断流，阻塞机理，整合控制
英文摘要
Abstract
On-ramp and off-ramp, the main traffic bottleneck of urban expressway, are not easy to be rebuilt. Traffic control and management are thus the key measures to alleviate traffic jam and improve the operation of expressway. The main phenomenon of Traffic jam at on-ramp and off-ramp is that vehicles cannot enter the expressway or exit from it. The reason that vehicles cannot enter the expressway lies in two aspects: the first, the capacity of the main lanes is limited and can not match the demand; the second, the surplus capacity of the main lane is enough while there is no vehicle result from the surface street control. Traffic flow at the off-ramp cann’t enters into the surface street because outflow is more than the surface street capacity.
Aiming at the congestion and jam at on-ramp and off-ramp, its mechanism and reason is analyzed in this paper. Aiming at the traffic jam at on-ramp, this paper put forward that an extension of the feedback local ramp metering strategy ALINEA is proposed which allows the automatic tracking of the critical occupancy to help maximize the mainstream and control the surface street demand which enter into on-ramp. There are two methods to control the demand of the surface street, one is green truncation, and another is red truncation. If traffic jam occurs even if the intergrated control strategy is adopted, road guidance and VMS control are activated. Aiming at the traffic congestion at off-ramp, the thesis put forward that the surface street capacity is increased, which has two ways, one is green extension, another is red truncation, the intergrated control strategy consider other traffic. If traffic jam occurs even if the intergrated control strategy is adopted, road guidance and VMS control are activated.
Cell transmission model was adopted in studying intergrated control of urban expressway and the surface street, and the model at on-ramp andtraffic signal modelwere ameliorated.
The results are fit for urban freeway control under the condition of existing technologies of traffic information collection and processing, and can be improved with the progress of relevant technologies.
Keywords: UrbanFreeway, On-ramp Metering, uninterrupted flow, interrupted flow mechanism of congestion and jam, Intergrated control
彭春露
入学时间：2003年
答辩时间：2006年
论文题目：城市突发事件紧急救援管理可靠性问题研究
中文摘要
摘要
突发事件的发生具有高度随机性。城市作为社会经济财富的聚集地，一旦有突发事件出现，其社会危险和财产损失可能无法估量，因此有必要建立科学高效的紧急救援管理系统，实现事件发生前的良好监测和预警，事件发生后快捷有效的救援，从而将财产损失和人员伤亡控制到最低水平。基于此，各国都在不断的开发和制定适合于自身特征的突发事件紧急救援管理系统以及相应的法律法规体系；研究领域里紧急救援管理方法和理论方面的研究成果也层出不穷。事实上，对突发事件进行良好的预警和快捷高效的救援很大程度上也依赖于紧急救援管理系统运行的高度稳定性和可靠性。特别是随着城市化进程的加快和城市规模的不断扩大，突发事件紧急救援管理系统的运行可靠性变得尤为重要。本研究将剖析城市突发事件紧急救援管理系统运行期间各环节发生故障的可能性，进而为提高其运行稳定性提供依据。
尽管可靠性分析在不少领域有广泛应用，如供水系统、通讯系统、电力系统等，但突发事件紧急救援管理可靠性分析方面的研究成果较为鲜见，因此本文以城市突发事件紧急救援管理系统为研究对象，在对突发事件发生前后事件预警、事件救援等活动特征进行分析的基础上，借鉴系统可靠性理论对城市突发事件紧急救援管理可靠性分析体系以及建模方法进行了一定的研究。
根据系统质量的定义，系统质量是其性能特性、专门特性、经济性、时间性以及适应性等方面，是系统满足使用要求的特性总和。本文研究的可靠性属于城市突发事件紧急救援管理系统的质量特性范畴，本研究需要基于救援系统功能特性、经济性等方面的研究成果进行。因此，本文首先对城市突发事件紧急救援管理理论和技术的研究成果以及各国紧急救援管理系统发展趋势进行了分析，从而明确了本研究对象为动态复杂系统；在此基础上本文通过分析系统可靠性理论的适用性，明确了本文需要采用适合于大型复杂动态系统可靠性研究方法，因此本文首先运用故障树模型建立了紧急救援管理可靠性分析体系，在此基础上运用Petri网模型分析了紧急救援组织可靠性，引入道路时空资源分析理念，运用模糊综合评价方法建立了救援车辆畅通可靠性评价模型。
本文首先对现行的城市突发事件紧急救援管理系统特征进行了分析，结果表明城市突发事件救援活动具有明显的阶段性，且功能模块的界限比较清晰，因此这里运用故障树模型以及层次分析法构建了突发事件紧急救援管理可靠度分析框架体系。在此基础上，本文发现城市突发事件紧急救援管理可靠性主要受到两类因素的影响，依据分析对象的区别，分为内部要素和外部要素两类，分别对应于紧急救援组织和城市道路交通网络。
本文通过分析紧急救援组织机制和特征，建立了城市突发事件紧急救援组织结构的随机Petri网模型，对突发事件紧急救援组织的静态和动态特性进行了模拟和描述，在此基础上分析了紧急救援组织结构可靠度以及效率可靠度的建模方法。在外部可靠性评价指标的分析中，本文首先运用Monte Carlo方法对道路网络的连通可靠性进行了建模分析，进而运用宽度优先搜索技术，提出了BFS算法对多出行起讫对以及节点失效情况下路网连通性进行了研究。在此基础上，本文提出了救援车辆畅通可靠度概念，并运用多级模糊综合评价法构造了救援车辆在路网OD对之间出行的畅通可靠度评价模型，实现了对救援管理系统可靠性外部影响因素——交通流可靠性的分析。
本论文首次将可靠性理论运用到突发事件救援管理系统的分析评价中，研究了系统运行过程中各环节发生状态突变的可能性。研究结果表明，这些研究方法具备较强的适应性。该研究成果为城市紧急救援管理系统的规划和管理提供了良好的评价依据，并且在紧急救援组织管理方面的研究成果也可为其他运输系统管理效率分析优化提供一定的借鉴。
关键词：系统可靠性，突发事件，紧急救援，组织可靠性，畅通可靠性
英文摘要
ABSTRACT
Incidents are stochastic and unpredictable, and they will bring huge loss and danger in the city, so setting up the effective defending system and taking some measures after incident occur becomes very necessary. Most countries in the world are trying to develop effective urban incident management systems and the relevant laws according to the situations of themselves. Furthermore, research on the urban incident management methods has begun for a long time, and the fruit also emerges in endlessly. In reality, the exact prediction and effective rescue for incidents also depends on the well and steady running of incident management systems. Especially the speedup of urbanization and the expanding city scale would cause incident happen more frequently, so the urban incident management system running with high reliability becomes very important. The reliability analysis of urban incident management system tries to study the likelihood of the failure occurrence in the system, which helps to improve its running stability.
Though reliability analysis has been extensively employed in many fields, such as water supply systems, communication networks, power systems etc., very little attention has been given the reliability analysis on incident management system. Thus this paper tries to set the reliability analysis method for the urban incident management system on the basis of analyzing its running characteristics.
According to the quality control method, the quality of one system includes a good many attributes such as its performance, technicality, economization, timeliness and adaptability etc., and is the summation of its traits which meet the anticipant application requirement. Reliability of the incident management system is part of its technicality, and this study should base on its other attributes such as performance, economization etc. So firstly this paper looks back and sums up the correlative methods on the urban incident management system and reliability research field, then analyzes the developing direction of these fields, by which the study object, research base and research method is cleared.
On the basis of analyzing the running characters of urban incident management system, this paper sets up its reliability analyzing system with FTA (Failure Tree analysis) and AHP(Analytic Hierarchy Process)method. On this base, the key factors which impact the running reliability of incident management system are analyzed, and the reliability analyzing models are set up for them. According to their characters, the factors are differentiated as inner and exterior reliability factors, corresponding to the reliability evaluation of the incident rescue management system and the road traffic network in turn.
On the basis of analyzing the rescue organizing mechanism, this paper sets up the stochastic Petri net model for analyzing the reliability of the incident rescue organizing process. When studying the exterior reliability of the incident management system, this paper models the road network connectivity reliability with Monte Carlo method, then the BFS algorithm is put out for studying the reliability of the multi Origin and Destination network and the network with some nodes being of no effect. On this basis, the travel reliability model is built for analyzing the impact of traffic flow on the emergency vehicles on the road.
For the first time, this paper applies the reliability analysis method for analyzing the incident management system, and sets up the reliability analyzing method for the urban incident management system. The study can help the management departments to evaluate and plan emergency management systems more scientifically, and to promote the rescue process more efficiently.
Key Words:system reliability, incident, rescue, organization reliability, travel reliability
张存保
入学时间：2003年
答辩时间：2006年
论文题目：基于浮动车的交通信息采集与处理理论及方法研究
中文摘要
摘要
交通信息是智能交通系统的重要基础。目前交通信息主要通过固定式检测器(如线圈检测器、超声波检测器、视频检测器等)进行采集，存在安装和维护成本高、覆盖范围小、仅能检测固定位置数据等不足，无法满足智能交通系统对信息的需求。随着GPS、GIS和无线通信技术的广泛应用，利用装备GPS的浮动车采集动态交通信息，具有建设周期短、实时性强、覆盖范围广、数据精度高等优点，是交通信息采集技术的重要发展方向。因此，从理论和技术层面对浮动车采集方式进行全面、深入地研究，具有重要的理论意义和实用价值。
本论文在国内外相关研究基础上，对浮动车数据采集、数据处理、多源交通信息融合等方面进行了研究，建立了相应的理论方法和数学模型，解决了目前浮动车技术应用中存在的一些难点问题，并提出了下一步的研究方向。
论文首先阐述了基于浮动车的交通信息采集系统总体设计，总结分析了国内外相关研究现状，指出了现有研究成果存在的不足，并在此基础上确定了本论文研究的切入点。
针对浮动车采样周期主要凭主观经验确定的问题，本文将浮动车瞬时速度当作随机信号，利用傅立叶变换对其进行频域分析，然后根据香农采样定理确定浮动车的最优采样频率和采样周期，并用时域分析方法（均值和方差）对最优采样周期进行了验证，从而建立了浮动车采样周期优化的理论方法。
对浮动车样本数量优化问题，分析了现有基于路段和基于路网的浮动车样本数量优化模型存在的不足，利用数理统计方法，将基于路段和基于路网的样本数量优化模型统一起来，在考虑计算时间间隔、浮动车运行速度、交通流密度分布状况、数据采集精度要求等因素基础上，首次提出了浮动车样本数量优化的两步骤模型，该模型适用于较大规模道路网络的浮动车样本数量优化问题。
在浮动车数据处理方面，建立了道路区间平均速度、路径行程时间和动态OD矩阵的估计方法和模型。当浮动车样本数量较少时，为了提高区间平均速度估计精度，提出了自适应权重指数平滑法。另外，对动态OD矩阵估计问题，在考虑车辆在途时间基础上，提出了一种新的动态OD矩阵估计方法，即先根据浮动车数据估计到达OD矩阵，然后利用到达OD矩阵估计出发OD矩阵，克服了现有动态OD矩阵估计方法存在的不足。
在道路交通事件判别方面，根据发生事件时车辆速度变化特征，提出了基于浮动车区间平均速度的道路交通拥挤事件判别方法。该方法具有简单实用、可移植性强、检测效果较好等优点。
针对浮动车数据与固定检测器数据融合问题，在分析了交通信息融合的必要性基础上，阐述了交通信息融合的层次、方法和流程。在区间平均速度估计方面，提出了基于BP神经网络的信息融合方法，提高了区间平均速度估计的精度；在交通事件检测方面，利用加权平均法对基于浮动车数据和固定检测器数据的交通事件信息进行融合，提高了事件检测效果。
论文系统地研究了基于浮动车的交通信息采集与处理理论及方法，特别是在浮动车采样周期优化、浮动车样本数量优化、基于浮动车数据的道路交通参数估计、交通事件判别、浮动车数据与固定检测器数据融合等方面取得了新的研究成果，为浮动车采集方式的实际应用提供了理论依据和技术支持。
关键词：交通信息采集，信息处理，浮动车，交通参数估计，信息融合
英文摘要
ABSTRACT
Traffic information is the important foundation of Intelligent Transportation System. Now Traffic data is collected mainly by fixed detectors, which can not satisfy the demand of Intelligent Transportation System, because of its high cost, low coverage area, etc. Along with widely used of GPS, GIS and wireless communication system, floating-car system has became a new way to collect real-time traffic data. It has many advantages such as low investment, large coverage area, high precision, and so on.
This paper summarizes the domestic and foreign research foundation, studies the theories and methods about floating-car system, presents models for floating cars data collection, data processing, data fusion, and so on.
At first, the thesis describes framework of floating-car system, analyzes prior research on floating cars, and then confirms this paper’s detail research items.
In research on floating cars sampling frequency, this paper regards speed as stochastic signal, then analyzes its frequency spectrum using Fourier Transform, and decides the optimal sampling frequency by Shannon sampling theory. It provides a theoretical method for frequency optimizing.
Floating cars number is a key problem of floating-car system. Existing models based on road section or based on network have some disadvantages, which affect models’ accuracy distinctly. This paper presents a two-step model for the first time, which takes many important factors into account, such as calculating interval, floating cars speed, floating cars distribution, data precision requirement, etc. The two-step model is suitable for large scale road network.
For issue of floating cars data processing, the paper describes estimation models for average speed, link travel time and dynamic OD matrix. It puts forward a self-adaptation exponent smoothing method for average speed estimating when samples are not sufficient. A new calculating method for dynamic OD matrix is also presented.
Based on character of vehicle speed at incident section, a automatic incident detection method is proposed, which is simple, practical and effective.
Lastly, the paper discusses data fusion methods for floating cars data and fixed detectors data. After describing demand and architecture of data fusion, it brings forward a Back-propagation Neural Network Model for fusing average speed data from floating cars and fixed detectors, which can promote accuracy of speed data. A mathematics method for incident data fusion is also discussed in this paper.
The thesis investigates the theories and methods for floating-car system, obtains some new production about vehicle sampling frequency, sample size, traffic parameters estimate, data fusion method, etc. It could be used to conduct real floating-car system.
Key Words: traffic data collection, data processing, floating cars, traffic parameters estimation, data fusion
孙剑
入学时间：2003年
答辩时间：2006年
论文题目：ATIS环境下交通网络仿真关键技术研究
中文摘要
摘要
交通流仿真是再现交通流运行规律，对交通系统进行管理、控制和优化的重要实验手段和工具。它可以对真实世界中尚未得到实施的技术进行细致的分析，对已实施的技术提出优化建议，在不对现有交通系统产生任何干扰下进行多种系统方案的检验，引导更有效的系统实施。随着智能交通系统特别是先进的出行者信息系统在我国的逐步发展，由于受到交通信息这种新的决策因素的影响，出行者的出行环境、出行路径选择机制以及在途运行行为都会发生变化，因此交通网络仿真系统必须建立新的模型体系或对现有模型体系进行改造来应对这种变化。
论文总结了国内外先进的出行者信息系统的发展以及交通网络仿真的研究历程，分析了目前交通网络仿真存在的问题和不足，为论文研究指明了方向；从驾驶员运行级模型（加速度确定）、战术级模型（车道选择）以及战略级模型（路径选择）三个层次研究了提供信息条件下的出行者模拟模型与算法实现；基于仿真系统参数校正理论，研究了利用实验优化技术、启发式算法进行仿真模型参数校正的流程与方法。论文主要内容包括：
（1）总结了国内外先进的出行者信息系统的发展、应用以及交通仿真理论与技术，设计了ATIS环境下交通网络仿真系统的系统框架、模型体系结构、软件架构。在此基础上开发了同济交通网络仿真系统TESS。
（2）研究了提供信息条件下的微观交通流模拟器，包括：基础设施模型、交通需求模型、管理控制模型以及运行环境模型。针对拥挤条件下出行者复杂的驾驶行为，引入了多智能体概念，以及车辆之间的“通讯”和“协商”机制。重点研究了“主动型”换道模型、“协作型”换道模型、汇入点模型、换道计划模型，实现了控制车辆运动的核心算法－车辆加速度的决策流程。以上海市快速路两处交织区调查数据为基础，验证了模型的有效性。
（3）研究了先进的出行者信息系统环境下的出行者分类、出行者对信息的反应以及出行者的路径选择行为，实现了ATIS环境下多用户仿真算法。以杭州市中心区路网为研究对象，运用本文开发的仿真模型，研究了不同信息用户在不同的出行环境中的出行效益分析。
（4）研究了仿真系统的参数校正流程与方法。利用正交实验研究了TESS模型的参数校正；利用模拟退火算法对VISSIM模型进行了校正；集成模拟退火算法和遗传算法的优势，以合肥市大东门区域路网为对象开发了遗传模拟退火算法对VISSIM模型进行了校正研究。
关键词：先进的出行者信息系统；交通仿真；协作型换道模型；基于仿真的动态交通分配；出行行为；参数校正；正交实验；遗传模拟退火算法
英文摘要
ABSTRACT
Traffic flow simulation is an important experimental method and tool with which we can recuure the characteristics of traffic flow and manage, control and optimize transportation system. It can be used to analyze unacted technics in actual traffic environment, give optimized advice for the applied techniques and test the different traffic scenarios under uninterrupted traffic system. In China, As the development of Intelligent Transportation Systems (ITS) especially the Advanced Traveller Information Systems (ATIS), travellers’ trip envioroment, route choice and en-route behaviors have changed a lot compared with the traditional traffic system because of the influence of traffic information. Micro-traffic simulation system must build new model or change the existed model to fit the new information environment.
This dissertation summarized the status and research process in aspect of ATIS and traffic network simulation, point out the deficiency and problem of micro-traffic simulation among the past researchers, and designated the direction of studying in this paper. From the traveler operation model (acceleration determinded), tactical model (lane choice) and strategic model (route choice) to study the simulation model and algorithms under ATIS. Based on the simulation system parameter calibration theory, this paper analyzed the method using the orthogonal experiment and heuristic algorithm to calibrate the system parameters. The research can be summarized as follows:
(1)Considering the information influence under ATIS, A simulation experimental system applied for micro-traffic simulation overall design, simulation and validation which named Tongji traffic nEtwork Simulation System (TESS) was build.
(2)Established the micro-traffic flow simulator under the information environment, including: network model, traffic demand model, traffic management model and opertation environment model. Considering the complicated driver behaviors under congested situations, the agent based model is introduced, and the communication and cooperation behaviors between the vehicles are modeld, including the mandatory lane-changing model, the cooperative lane-changing, ramp-point model, lane-changing plan model and realized the acceleration and lane choice decision flow. Taking two weaving sections as examples in Shanghai expressway, the paper proved that the model is proper.
(3)Studied the travler classification, traveler reactions and route choice behaviors under the dynamic information conditions and realized the multi-user route choice algorithms based on traffic simualtion and Dynamic Traffic Assignment (DTA) theory. For the simulation model validation, the paper established the Hangzhou downtown TESS model to test the users benefit under the different route guidance strategic.
(4)Based on simulation system parameter calibration theory, it set up the micro-traffic simulation parameter calibration flow and method. In this paper, two weaving sections TESS model parameters were calibrated using orthogonal experiment, and Big-door area VISSIM model parameters were tested using Simulted Annealing (SA) and Genetic Simulted Annealing (GSA) algorithms respectively.
Keywords: Advanced Traveller Information Systems (ATIS); Micro-Traffic Simulation; Cooperative Lane-Changing; Simulation-Based Dynamic Traffic Assignment; Traveler Behavior; Parameter Calibration; Orthogonal Experiment; Genetic Simulted Annealing (GSA) Algorithm
伍速峰
入学时间：2003年
答辩时间：2006年
论文题目：交通信息效用解析研究
中文摘要
摘要
交通信息在智能交通运输系统中起着重要的作用，从智能交通运输系统的概念被提出以来，关于交通信息对出行者影响的研究就一直受到密切关注。国内外不少学者和研究机构都对此进行了研究，并提出了各自的研究方法。但由于缺乏严格的理论基础和适用的数学模型，目前该领域的研究进展缓慢。本论文在总结国内外交通信息效用研究理论与方法的基础上，系统地对不考虑交通信息有效性反馈（适用于非日常出行）和考虑交通信息有效性反馈（适用于日常出行）两种条件下交通信息的作用机理进行研究，尝试采用风险决策、贝叶斯推理、博弈论等理论和方法建立了相应的模型，并对影响交通信息效用的交通信息质量、交通系统稳定性、出行者的出行经验等因素进行了分析，提出了增加交通信息有效性的方法。
在不考虑交通信息有效性反馈条件下交通信息效用的研究中，论文首先从不确定性的角度分析了出行者出行时面临的两类不确定性：选择的不确定性（uncertainty of choice）和选择项执行的不确定性（uncertainty of execution），提出了交通信息在减少这两类不确定性可能起到的作用：使出行者能了解其以前未知的选择项，即能有更多的选择余地；使出行者对以前所了解的选择项有更加明确的认知，即选择更加明确。然后，基于不确定性选择理论和离散贝叶斯决策理论提出了信息变量离散时，出行经验与交通信息融合的计算方法，交通信息效用的作用机理和交通信息效用的数学表示方法。最后应用不确定性选择理论和连续贝叶斯推理，将这一研究结果推广到信息变量连续时的情形，并通过算例，分析了提供交通信息条件下的出行选择行为计算过程和交通信息效用的表示方法。研究表明本文提出的模型能定量地描述交通信息在减少出行选择不确定性的作用，合理地反映交通信息对于出行决策的影响，为研究提供交通信息条件下的出行奠定了理论基础。
在考虑交通信息有效性反馈条件下交通信息效用的研究中，论文首先分析了有、无交通信息两种条件下出行者日常出行的决策过程，并通过对比，提出了两种条件下出行者的认知更新过程和计算方法。然后在上述分析的基础上建立了解析日常出行中交通信息效用的仿真模型，为分析日常出行中的交通信息效用奠定了基础。最后通过该模型分析了交通信息质量、交通系统稳定性、出行者的出行经验等因素对交通信息效用影响的敏感程度。研究表明，交通信息准确性越高、交通系统稳定性越差、出行者经验越少则交通信息的效用越大；发布交通信息要有选择性，并不是交通信息越多交通信息的作用越大。
论文最后运用博弈论相关原理，提出了在日常出行和非日常出行的两种情况下考虑出行者之间选择相互影响的交通信息效用解析方法。研究表明当交通系统容量有限，且出行者数量又较多时，交通信息对于出行者作用的大小不仅与自身选择行为有关，还与其他出行者对交通信息的反应有关，如果出行者不能正确预测其他出行者对交通信息的反应，就会产生过度反应，反而会使交通系统更加不稳定。因此需要理性地对待交通信息，充分发挥其优势，减少产生的不良反应。
本论文从交通信息对减少出行选择不确定性影响的角度研究了交通信息的效用，提出了交通信息效用的数学表示方法，系统地建立了交通信息效用静态和动态模型，为解析交通信息的作用提供了一条新的途径，为研究提供交通信息条件下的选择行为奠定了理论基础，同时也为智能交通运输系统的评价分析以及交通信息的优化利用提供了理论和指导方法。
关键词：交通信息，信息效用，不确定性，贝叶斯决策，博弈论
英文摘要
ABSTRACT
The traffic information plays a vital role in the Intelligence Transportation System, and the effects of the traffic information have been paid much attention since the conception of Intelligence Transportation System was put forward. Many scholars and Research Institutes have been studying on this problem, and proposed respective methods. However, the research in this field is still in its initial stages, since lacking strict theories and suitable mathematical models. Based on the summarization of traffic information utility theories and methods at home and abroad, this thesis studies the traffic information utility in two kinds of conditions (one is for the non-daily travel without considering the feedback of traffic information validity, while the other is for the daily travel considering the feedback of traffic information validity), and uses the risk decision, Bayes formula and game theory to analyze the traffic information utility. Then, thesis analyzes the effecting factors, such as traffic information accuracy, traffic system stability and travelers’ experience, and puts forward the corresponding model.
Without considering the information validity feedback, this thesis analyzes the two kinds of uncertainty when travelers make a travel decision: uncertainty of choice and uncertainty of execution, and proposes the possible function of traffic information in reducing these two kinds of uncertainty: make the travelers be able to know more alternatives and characteristics of travel alternatives. Then based on the uncertainty choice theory and Bayes decision theory, the fusion method of travel experience and the traffic information, and the mathematics expression method of traffic information utility are put forward in the condition of discrete traffic information. At last, this research method was extended to the continuous traffic information, and the calculation procedure of travel behavior and description method of traffic information utility are analyzed via an example. The research indicates that the model proposed in this article could explain the function of the traffic information in reducing the uncertainty of travel choice and reflect the impacts of the traffic information on traveler’s behavior reasonably. Moreover, this model can provide a theory foundation for the travel with traffic information.
Considering the information validity feedback, the decision procedure of daily travel is analyzed in two kinds of conditions (with or without traffic information), the recognizance procedure and calculation method of travelers in these two kinds of conditions are put forward by comparing them. And then the simulation model for analyzing the traffic information utility in daily travel is established, which can provide a theory foundation for the traffic information utility in daily travel. At last, the sensitivity of traffic information utility affected by the quality of traffic information, the stability of traffic system and experience of travelers are analyzed in this model. The result shows the information utility will be more useful when the information is more accurate, the traffic system is less stable and the travelers have less experience. The traffic information should be released selectively since more information does not mean more useful.
Lastly this thesis proposed a method to calculate the traffic information utility under the non-daily and daily travel when considering the choice affects mutually. The research indicates the traffic information utility was determined not only by the choice behavior of the traveler, but also by the choice behaviors of other travelers, if the traveler can’t predict the others’ choice behaviors, the responded excessively will conduced and the traffic system will become less stable. Therefore we should be treated the information rationally, develop its advantage and reduce its disadvantage.
In conclusion, this thesis studies the utility of the traffic information from the view of reducing the uncertainty, and proposed a mathematics expression method for the utility. Then a static model and a dynamic model were developed. This study not only develops a new way for analyzing the effects of traffic information, but also provides a theory and method foundation for evaluating the Intelligence Transportation System.
Key Words: dynamic traffic information, information utility, uncertainty, Bayes decision-making, game theory
