高云峰

入学时间：2004级

答辩时间：2007年

论文题目：动态交叉口群协调控制基础问题研究

中文摘要

摘要

城市道路网络交叉口信号控制是缓解日益严重的城市交通拥挤的主要手段之一。由于城市居民出行行为和出行方式发生变化、信息因素在城市交通系统中的作用日益加强，以及与交通工程相关的学科新理论和新方法在交通系统中的应用，人们对城市交通系统的理论认识不断深化，对城市道路网络交通流协调控制理论和方法继续进行研究仍然具有重要的理论价值和现实意义。

论文对城市道路网络交通流协调控制理论和方法进行了深入研究。与已有的协调控制研究成果不同，论文以城市道路网络交通流的时间和空间变化特征为依据，把城市道路网络交通流协调控制问题系统地分成两个层面——战略层面和战术层面，这两个层面以不同的频率对城市道路网络交通流的时空变化进行跟踪和控制。战略层面根据城市道路网络交通流的结构变化特征对协调控制区域动态划分交叉口群，战术层面根据动态交叉口群交通流的时间变化特征对交叉口信号控制方案进行协调优化。论文把协调控制基本范围的动态划分和信号控制方案的协调优化作为城市道路网络交通流协调控制的两个基础问题。

首先，论文从城市道路网络交通流协调控制理论的国内外发展历程、典型系统结构、控制参数优化策略、协调控制基本理论问题、协调控制优化方法演变等方面对已有的城市道路网络交通流协调控制研究成果进行了总结，并对一些典型的城市交通控制系统进行了介绍。

其次，论文考虑了城市道路网络交叉口之间关联性的动态变化，提出了动态交叉口群的概念，使协调控制基本范围划分动态化。根据城市道路网络交通流动态OD分布特性提出了相邻交叉口的路段关联性和路径关联性概念，分析了动态交叉口群的划分依据，以相邻交叉口的路段关联性和路径关联性为基础建立了基于层次聚类方法的动态交叉口群划分算法。

再次，论文应用多智能体系统理论与方法建立城市道路网络交通流协调控制模型体系，提出了基于多智能体技术的分布式动态交叉口群协调控制系统（DDIGCCS）结构，定义了基于多智能体技术的分布式动态交叉口群交通流协调控制系统中各种智能体的结构和功能，并建立了实现各种功能的模型和方法。在动态交叉口群协调控制模型体系研究方面，论文对Robertson车队离散模型和水平排队模型进行了改进，并应用改进后的车队离散模型和水平排队模型描述路段上交通流的运动规律。论文提出了交叉口微观交通状态的概念，并以此为基础建立了车辆在交叉口进口的排队长度、控制延误、停车次数，以及出口流量的实时估计模型，用于实时评价交通流协调控制效益。针对路段上超长排队的现象，论文基于门控制策略建立了交叉口进口超长排队控制模型、应用动态规划方法建立了相位绿信号持续时间实时优化模型。针对相位差的协调优化，论文提出了主要车队和次要车队的概念，依据总控制延误最小的原则建立了交叉口相位差的局部优化模型；论文还提出了主协调路径的概念，对对应相位和协调相位进行了明确区分，建立了闭合路径上所有交叉口的对应相位差的闭合约束条件和协调相位差的虚拟约束条件，进而建立了交叉口参考相位差协调优化模型。论文还提出了信号控制参考方案的概念，建立了交叉口参考周期时长和绿信比协调优化模型。对于信号相位方案优化问题，论文提出了设计信号相位方案的一般原则，并基于HCM2000的间隙接受模型建立了设置左转相位的流量条件，按照总控制延误最小的原则对最优信号相位序列进行了研究。

最后，论文借助于计算机软件技术和系统仿真技术对基于多智能体技术的动态交叉口群协调控制策略、模型，以及算法进行了初步验证，初步证实了论文提出的基于多智能体技术的分布式动态交叉口群协调控制策略、模型和方法的有效性。

关键词：动态交叉口群、多智能体、交叉口微观交通状态、信号控制参考方案、主协调路径、对应相位、协调相位，DDIGCCS

英文摘要

Abstract

One of the main means for relieving traffic congestion and jam in urban road network is signal control. Because of changes in citizens’ trip behavior and trip mode, increasingly strengthened effect of information on municipal transportation system, as well as wide application of new theories and new methods in related disciplines to municipal transportation system, traffic researchers have a more deep insight into municipal transportation system, persistent research on theories and methods of coordinated control of traffic on urban road network is still to have important theory value and practical significance.

This dissertation researches theories and methods on coordinated control of traffic on urban road network. Different from existing researches, based on time and space characteristic changes of traffic on urban road network, the research divides the problem into two subproblems: strategic problem and tactical problem, which track and control the changes at different frequents. At the strategic level, an area is dynamically divided into some intersections groups according to structural changes of traffic. At the tactical level, signal control plans of intersections are coordinatedly optimized according time changes of traffic. The dissertation takes dynamic partition of basic region for coordinated control and coordinated optimization of signal control plans in basic regions as the two fundamental theoretical problems of coordinated control theory of traffic on urban road network.

First, the dissertation gives a systematical summary of existing coordinated control theories of traffic on urban road network. The summary is organized by internal and external growth course of signalization of urban road network, typical traffic control system structures, optimization strategies of control parameters, theories of coordinated optimization and traffic characteristics on signalized urban road network, evolution of optimization methods of signal control parameters, and introductions of some famous typical urban traffic control systems.

Second, the dissertation puts forward the concept of Dynamic Intersections Group (DIG), which takes the dynamic change of the relevance of adjacent intersections into consideration and makes partition of basic regions for coordinated control dynamically. The dissertation analyses the principles for partition of DIGs, and puts forward the concepts of link relevance and path relaevance for adjacent intersections according to the dynamic OD distribution of traffic on urban road network, establishes a mothod for partition of DIGs within an area according to the relevances of adjacent intersections based on Hierarchical Clustering Method.

Third, the dissertation applies related theories and methods of Multi-agent System (MAS) to establishing the system structure of coordinated control of traffic on urban road network, putting forward a Distributed Dynamic Intersections Group Coordinated Control System (DDIGCCS) based on MAS. The DDIGCCS defines some specific agents and their structures and roles in the system, and establishes corresponding models and methods. The dissertation makes improvements on Robertson Platoon Dispersion Model and horizontal vehicle queue model, and applies the improved models to the depiction of discipline of traffic flow on the links. The dissertation puts forward the concept of microscopic traffic state and establishes its model, and establishes the models of vehicle queue length, control delay, and vehicle stops on an approach which can realtimely evaluate the effectiveness of coordinated signal control plans on traffic. Aimed at avoiding extremely long vehicle queue lengths on a link, the dissertation establishes the Signal Control Parameter Regulation Model based on the gate control strategy. The dissertation establishes the Phase Duration Realtime Optimization Model based on dynamic programming. The dissertation puts forward the concepts of major platoon and minor platoon, establishes the Offset Local Optimization Model according to total control delay minimization goal. The dissertation puts forward the concepts of Signal Control Reference Plan, and establishes a Signal Control Reference Plan Optimization Model for optimizing the reference cyle length and splits at each intersection. The dissertation puts forward the concepts of Main Coordination Path, Corresponding Offset and Coordinated Offset, establishes the constraints for the corresponding offsets and coordination offsets on a closed path, and establishes the model for coordinated optimization of the reference offsets in a Dynamic Intersections Group. The dissertation gives the ordinary principles for signal phase plan optimization, establishes the traffic volume condition on protected left-turn phase, and gives optimization models of signal control plans according to total control delay minimization goal.

Last, the dissertation verifies the effectiveness and advantage of the strategies, models, and algorithoms of DDIGCCS by aid of computer software programming and system simulation techniques.

Key Words: dynamic intersections group, MAS, microscopic traffic state at intersection, reference signal control plan, main coordination path, corresponding offset, coordinated offset, DDIGCCS

马万经

入学时间：2004级

答辩时间：2007年

论文题目：公交专用道信号优先控制理论研究

中文摘要

摘要

交通拥挤在全世界范围内日益凸现为亟待缓解的重要问题，优先发展公共交通被认为是缓解交通问题，改善交通状况，提高交通（客流）运输效率，促进交通与城市可持续发展的最主要和最有效的途径之一。在我国及许多国家的大城市，公交（本文特指公共汽、电车交通）系统规模巨大，网络复杂，为数量庞大的出行者提供运输服务，在城市客流运输系统中占据非常重要的地位。在公交系统的总体服务水平和运行可靠性较差，机动化竞争越来越剧烈，公交吸引力较低且改善手段又相对匮乏和落后，而专用道等公交优先基础设施得到大力发展的情况下，公交优先控制理论的研究变得极具现实意义。

论文首先提出了公交优先控制系统的原型结构，将信号控制交叉口群，而不是单个交叉口，作为优先控制的对象；提出了离线优先策略与在线优先策略相结合，信号控制与调度策略相协调，协调控制与单点控制互补的优先控制思想，设计了系统运行的逻辑流程。这一原型设计使得优先控制系统不但可以应对公交网络简单、线路发车频率低的情形，还适用于道路网络密集，公交网络复杂且流量大，交通流总体运行可靠性不高，行人过街需求大且频繁的状况。论文随后分离线优先控制策略、在线优先控制策略、信号优先与调度协调策略和信号优先优化软件实现四个主要部分展开深入研究。

离线公交优先控制策略中，公交车流和社会车流共同参与优化，首先给出了控制的任务、目标和控制方案的生成逻辑，并主要以四个信号配时参数：信号周期、绿信比、相位差和相位相序的优化模型与计算方法为研究重点。提出了“基本周期”与“在线增量”的信号周期优化模型结构；结合仿真数据给出了模型的基本参数，分析并佐证了“人均延误最小”目标和“缩短周期利于信号优先”等相关研究结论之局限。在交叉口时空资源相互转化关系分析的基础上，将车道功能划分作为变量引入信号配时参数优化过程中，提出了以时空组合优化最优为目标的绿信比优化模型。在相位相序设计中，不再拘泥于单个交叉口和机动车流本身，而是将优化范围扩展到信号协调的相邻交叉口，优化对象扩展到行人相位和机动车相位，进而实现交叉口群内的最优化，并考虑了行人过街这一对公交系统服务水平有重要影响的因素。着重研究了考虑公交优先后，相邻交叉口左转相位协调优化的整数规划模型和行人相位与机动车相位的组合优化模型。建立了公交优先相位差优化模型，并进行了敏感性分析。在研究过程中，对所建立的模型都进行了相应的实证分析，模型的有效性及其相对于传统方法的优越性都得到了证明。

在线优先控制策略研究中，首先研究并确定了在线策略的主要任务和目标，给出了在线控制方案生成的逻辑流程。重点研究了有条件公交协调优先控制模型，该模型以协调控制交叉口群为控制对象，基于车辆通过交叉口群的延误预测，以公交车辆运行状态偏移值（含正、负两种情况）最小化为目标，以社会车流运行服务水平（排队和延误）和交通安全等为约束条件，进行交叉口群内各个交叉口优先策略的组合优化。模型包含了降低公交延误和增加公交延误两大类策略，对优先申请生成、优先时间分配、优先策略及其延误计算、交叉口群策略组合优化和多申请排序等关键组成部分进行了详细研究。协调优先控制模型具有为公交车辆提供有条件的优先，处理多申请，协调群内交叉口优先策略，平衡分配群内各交叉口的优先程度并保持社会交通信号协调的能力。

为应对由于协调模型中公交车辆在交叉口群延误预测不准的情形，在线优先策略还提出了协调条件下的单点感应优先逻辑模型。该模型的目标与协调策略的目标保持统一，根据车辆在交叉口的实时运行状态进行感应。为避免公交优先对社会车流造成较大影响，公交单点感应优先控制仅采用绿灯延长和红灯早断两种策略。各策略的控制逻辑和主要参数的确定方法都进行了详细的研究。

信号优先与调度在本质目标上的同一性和现实的关连性决定二者之间的协调对两种策略的实现具有重要意义。在对二者关系的初步解析的基础上，面向问题，分离线和在线两部分进行了公交信号优先与公交调度策略协调方法的初步探索。

以优先控制基本功能和核心模型实现为目标，设计了公交优先控制软件的数据结构和数据流图，并应用Java Eclipse编程实现了软件的基础版本V1.0。

论文设计了公交优先系统的原型结构，深入研究了其控制逻辑和优化模型，并进行了软件实现的探索；拓展和丰富了交通控制理论与方法，并为公交优先控制系统的开发奠定了理论基础。受研究时间和优先控制系统本身及其实施的复杂性所限，研究尚有诸多缺陷，特别是未能对研究成果进行应用研究，这是后续研究的重要内容。同时，非专用道条件下优先控制模型及非循环优化等都是后续研究的主要方向。

英文摘要

ABSTRACT

Traffic congestion is growing to a more and more important problem all over the world. Giving priority to bus movements has been taken as an efficient way to improve level of service of bus system and attact more passegers and reless traffic jam. Bus network is huge and complex and playing a very important roal in transportation system of China. Lots of people have no chice but take bus in daily life. However, the efficency, level of service and reliability of bus system is still very low. As the blossoming of motor traffic and automobiles, the unappealing bus system may loss it’s foregone and potential passegers dromaticly. Lots of exclusive bus lanes have been built to ameliorate bus system, and gving priority to transit has been adapted as a basical politics in many countries including China. Under this condition, the research of bus signal priority theries, an efficient tool to improve travel speed and reliability of bus systme is very important and significative.

A prototype of bus signal priority control system is proposed. It is grouped intersections not isolated intersection is adopted as control object. The core concept of the prototype is combination of offline bus priority strategies and online bus priority strategies, reinforcement of cordianted optimization of signal priority and isolated acuated signal priority and coordination of priority signal control strategies and bus operation strategies. This concept extend capability of the bus signal priority to adapt the condition of heavy bus volume, denseness road and bus network, low reliability traffic condition and huge demand of pedestrian crossing. Four parts are followed to realize this prototype, offline priority strategies, online priority strategies, coordination of signal control and bus operstion strategies and development of bus signal priority core sofeware.

In offline bus priority strategies, motor movements and bus movements are joined optimized. The control logic is gaven firstly and four models are developed respectively for optimization of cycle, split, phase sequence and offset. A structural model of optimal cycle length, including “basic cycle” and “online increment” is proposed. The limitations of foregone concolusions of “small cycle benefit to buses” and “minimize passager delay is an objective of bus priority” are proposed and analyzed. In the phase sequence model, coordination optimization of left turn phases of adjacent intersection and combination optimization of bus phases and pedestration phases is researched in-depth. In the split model, allocation of lane function is included as an optimization variable. In the offset model, the sensitivity analyse of bus travel time is done. Earch model was test and evaluated by simulation.

In online bus priority strategies, logic of complement earch other of corrdianted signal priority and isolated actuated priority is proposed. An in-depth study is put on a coordinated and conditional bus priority approach. Five models are included in it: bus delay prediction model, priority requests generation model, priority time allocation model, multi request sequence optimization model and corrdiantion optimization model of priority strategies of intersections in a group. A comprehensive simulation test shows that the approach can achive much benefit to buses and the adverse affect on motor movements is very low. In order to response to the variance of bus travel time between detection points to intersection stopline, an isolated actuated control priority strategy based on coordinated control plan is also developed.

The objective of bus signal priority strategies and bus operation strategies decied that the two strategies should be coordinated. Based on the analysis of the relationship between them, a coordination approach between bus signal priority control strategies and bus departure frequence is proposed and evaluated.

Aming at the realization of basic bus signal priority control function, a priority control software package is developed. The basic version is development by Jave Eclipse.

The dissertation proposed and transit signal priority control prototype, developed and evaluated the core models. Lay an academic fundation for development of transit signan priority control system and extend the traffic control theories.

袁长亮

入学时间：2004级

答辩时间：2007年

论文题目：城市路网过饱和交通信号控制策略研究

中文摘要

摘要

交通信号控制是调节路网交通需求，改善路网服务水平的重要手段，尤其在路网发生严重拥堵的出行高峰期更是如此。从理论研究的角度来看，信号控制策略体现了设计者的控制思想，是复杂系统建模过程中的必须环节，也是交通阻塞机理解析中的重要影响因素。

纵观对过饱和状态控制问题的研究，虽然起步较早，但限于人们对控制对象认识的水平，使得研究过程中采用的交通流模型、控制方法和控制问题描述等重要的环节有所局限。随着实际控制平台上拥挤状态控制研究的积累、元胞传输模型在城市路网交通控制问题中的突破性应用和路网可靠性问题研究的深入，使得对过饱和控制问题的认识程度逐步加深。从控制策略的发展历程上来看，交通控制与控制理论的结合也越来越紧密，从开环控制到闭环控制再到智能控制，一方面是道路用户对交通控制提出了更多更广的需求所致，另一方面是伴随着智能交通系统的深入研究和推广，通讯、数据采集和计算技术等基础技术的进步，使得研究人员的视野也更加宽广。本文在这一大的研究背景下，遵从系统建模→算法设计→结论推导的研究思路，采用仿真建模与控制建模一体化的方法，依托于详实的数据分析，对路网过饱和状态的信号控制策略进行了深入研究。

本文首先对以往涉及到的过饱和控制问题进行了详细的总结，归纳出过饱和状态控制中六个基础问题，即控制小区的划分问题、控制目标的类型问题、控制方法问题、优化计算方法问题、交通流模型问题和控制策略问题。

控制策略的研究与其它五个部分相互贯穿和渗透，在总结分析的基础上从道路断面、路段排队空间和路网拥挤分布等几个分析层面提取了描述过饱和状态特征的饱和度与路段密度两个指标。并进一步针对路段和交叉口设计中的正常渠化设计模式、短连线设计模式和连续车道设计模式这三种交通设计模式，以这两个指标对路段控制状态空间进行分析，得出路段过饱和控制状态特征和状态演变特征。在过饱和交通特性分析的最后部分详细阐述了协调层控制中的控制参数与控制结构的关系，以此为基点，揭示了过饱和状态下控制小区的划分本质是寻找路网中与过饱和路段相关的信号灯组的关系问题，并对过饱和控制问题进行了完整的描述。

在系统建模分析的基础上，把控制算法与元胞传输模型仿真通过数据库的交互联系在一起，实现了策略开发与测试的单机仿真实验平台。在控制算法设计与实现的过程中结合了控制理论中的反馈控制和分层递阶控制思想，围绕本文提出的概率性控制目标，构造了自上而下的三层结构控制算法，并通过了仿真平台的测试。

进一步通过大量的仿真数据分析得出本文研究的主要结论如下：

⑴ 通过建立元胞传输模型与信号控制一体化的仿真测试平台，获取了动态信号配时研究过程中的基础数据支撑，为复杂控制状态条件下控制策略的开发提供了方法支持。

⑵ 针对过饱和状态空间，通过状态频率特性分析和状态转移频率特性分析，揭示了过饱和状态空间特性，建立了概率性的控制目标函数。这一通过对控制状态空间特性分析，并进一步寻找结合工程背景的启发式优化方法的思路也可以应用在过饱和状态以外的路网交通信号控制中。

⑶ 控制系统软件数据结构设计中，应结合信号阶段和信号灯组的双重表达来设计，以支持同一信号阶段内或跨信号阶段信号灯组解除绿灯相同时长约束的算法要求，进而使动态控制算法具有良好的稳定性和灵活性。

⑷ 对路段控制状态特征进行大量的数据分析得出，在短连线交通设计模式中，过饱和状态特征呈现两种极端的情形。一种是呈现高密度高饱和度的严重阻塞状态，在整个状态空间中难以发生状态的转移；另一种是在除了低密度高饱和度状态之外的整个状态空间中发生不稳定的跳变。对于处理这种小范围区域排队空间严重受限路段的控制问题合理的控制策略或方法应是尽可能的设计稳定通行的绿波，使得短连线路段运行在低饱和度高密度的控制状态。

⑸ 从动态配时的角度来考虑过饱和状态时静态配时方案计算的问题，只要动态配时方案达到调整的极限稳定状态即可以用来作为多时段选择的备选方案，建立了过饱和条件下动态配时与静态配时之间的联系。

⑹ 上游截流的控制策略，对于线性路网，截流范围越大缓解阻塞的控制效果越明显；对于复合路径路网，截流范围的扩大并没有带来控制效果的直接改善，但交叉口不同饱和程度、不同流向车流间拥挤发生转移，在一定的程度上体现了截流控制的效果。

关键词：控制状态分析，过饱和控制策略，元胞传输模型，反馈控制，分层递阶控制，控制算法设计

英文摘要

ABSTRACT

Traffic signal has the ability of adjusting network demand and improving the network level of service, especially in rush hours. From the point of theoretical study, traffic control strategy represents the designer’s control idea. It is a critical step in complex system model construction and an important influence factor in traffic congestion mechanism resolving.

Summarizing the study of oversaturated traffic control, although it has a long researching history, the acquaintance limitation of network traffic flow retards its development. This limitation includes the traffic flow model, control methodology and control objectives. With the accumulation of field data study and the application of cell transmission model in urban network traffic control successfully and the development of network reliability study, the acquaintance degree of oversaturation enforces gradually. From the point of traffic control strategy, traffic control and automated control theory have combined tightly. The control method has experienced open-loop control, closed-loop control and intelligent control. This rapid development is motivated by two aspects. One is the developing demand of road users, and the other is the rapid development of ITS, communication technology, data collection technology and computation technology. This dissertation researches the oversaturated control strategy in above mentioned background. The research structure follows the logic of “system model construction→control algorithm design→conclusion derivation” with sufficient simulation data analysis.

In the first part, the dissertation gives a detailed literature review about oversaturated traffic control and summarizes six basic fundamental problems. They are control subsystem partition problem, control objective category problem, control methodology selection problem, parameter optimization problem and control strategy problem.

The study of control strategy combined with the other problems tightly. The researching process extracts two critical state indices, link density and degree of saturation, from the integrated point of road section, road link capacity and congestion distribution to construct the link state space. Then three typical traffic design modes－normal link design mode, short-link design mode, continuous link design mode－are selected out to analyzing the characters of the state space. Analysis results show the state characteristics and state evolution characteristics. In this character analysis part, another important relationship of coordinated control parameter and coordinated control structure is depicted in detail followed by an entire control problem describe. And the analysis results show that the nature of control subsystem partition problem is to find the oversaturation related signal groups in the network.

On the basis of system model construction, the dissertation develops a control & simulation platform. So, the control algorithm can interact with the CTM simulation by database and computer memory on-line. Feedback control and hierarchical control are basic control theory ideas used here. The dissertation gives a top-down algorithm design containing three levels to complete above derived probabilistic control objective, and control results are tested in the developed CTM simulation platform.

Large amount of CTM simulation data analysis shows six concluding remarks as follows:

⑴The control & simulation platform supports the oversaturated control strategy study, and this method can be extended to other complex control problem research.

⑵By control state analysis, this dissertation posts the inner characteristics of state space and constitute a probabilistic control objective. This methodology can also be extended to other control strategy development.

⑶ In the data structure design, signal stage structure and signal group structure should be combined organically. This management will support the timing demand of relieving green time uniform constraint in one stage. And it will also improve the stability and flexibility of control algorithm.

⑷By detailed data analysis, short-link design mode presents two extreme state. One is high-density-high-saturation control state with minute state transfer. The other is a very unstable control state with frequent state transfer. So, a proper control strategy is to stabilize the platoon green-wave in a high-density-low-saturation state.

⑸Clarify the relationship of dynamic signal timing plan and static signal timing plan in oversaturated control state. The limitation of dynamic signal control plan can be selected out as an off-line timing plan in the oversaturated region.

⑹The river-closure coordinated strategy produces different control results in linear network and complex-route network. In linear network, the larger the closure region is, the less the opportunity of overflow appears. In complex-route network, this character is not obvious. But turning flow congestion transfer reflects the hierarchical control effects in some degree.

Key Words:control state analysis, oversaturated control strategy, cell transmission model, feedback control, hierarchical control, control algorithm design

庄斌

入学时间：2004级

答辩时间：2007年

论文题目：基于多源交通信息的交通状态估计理论及其应用

中文摘要

摘 要

随着国家“十五”和“十一五”发展纲要中对信息化和智能化技术的重视，交通信息化和智能化也日益提上日程。其中，动态综合交通信息平台作为智能交通系统的核心内容，已经成为了目前整合交通信息资源、实现资源互通互联、提升系统效率、缓解交通拥挤的有效手段。

论文在以往有关动态综合交通信息平台理论研究的基础上，从分析用户主体、需求层次和功能需求的角度，深入探讨了面向ATIS的动态交通信息服务需求，从而提出本论文的主要研究方向，即围绕动态综合交通信息平台的数据分析处理模块的功能需求，利用实时多源交通信息来估计路段、路径和路网上的交通流运行状态，形成了一整套适合于城市道路交通状态估计的理论体系，并通过实际案例分析和微观仿真模拟，对文中给出的理论和方法进行了验证。

论文在对城市路网上的交通状态估计理论和方法上具有一定的创新性。论文从“点、线、面”的角度，采用数理统计建模、时间序列、系统分析等理论，对城市路网上路段、路径和路网的状态进行分别讨论，符合由简到繁、由浅入深、循序渐进的研究规律。在研究方法和理论创建上，注重了与实际工程应用的结合，文中提出的理论模型、算法建立在数据分析的基础上，保证了理论的真实可用性。本文中给出的交通状态估计理论和方法，理论模型简单，算法易于实现，比以往的研究成果更具有工程实用性和可实施性。

在分析城市道路路段交通状态估计理论时，本文从分析研究路段上的车流运行特性着手，利用固定检测装置采集到的交通流基本特征参数，构建了交通拥挤与消散状态的相对增量判别准则，来对交通流的状态相位进行判别。在此基础上，提出了路段的交通状态相对增量判别算法，实现对交通状态的自动识别过程。研究表明，本文给出的路段交通状态估计算法具有较高的理论意义和应用价值。

在分析城市道路路径交通状态估计理论时，在结合了路径上交通流的运行特性分析的基础上，本文利用自动车牌识别设备采集到的交通数据，建立起该种数据采集方式下的行程时间估计算法模型，并且考虑到实际中不可避免的问题，如设备故障或采集样本数量过少，给出了应对策略和算法方案。最后，对文中提出的行程时间估计算法进行了实例验证分析。

在分析城市道路路网交通状态估计理论时，本文首先分析了城市交通路网的拓扑结构形式和功能层次，提出适合分布式运算的路网划分原则。对路网上的交通流运行状态估计时，考虑了利用目前常用的GPS车辆定位信息进行路网状态估计的技术。重点分析了GPS车辆位置信息的时空特征，提出一种路网状态估计的面向分布式计算的搜索算法，并利用实验仿真和数据库编程的方法来实现整个路网状态估计和交叉口延误的计算公式。经过实例验证和仿真结果比较，本文给出的方法具有较高的实用价值和应用意义。

本论文的研究成果为建立我国动态综合交通信息平台模型体系提供了系统工程层面的理论支持，并丰富了原有的相关理论和技术，初步形成了相应的交通信息关键参数分析方法，为进一步开发有我国自主知识产权的动态综合交通信息系统提供了理论和方法指导。

关键词：交通信息，数据处理，状态估计，延误计算

英文摘要

ABSTRACT

With Chinese government attaching more attention to the national informationization and intelligentization policies in the 10th and 11th five years plans, the informationization and intelligentization of urban traffic are developing rapidly. The dynamic integrated traffic information platform, as one of the most important components of ITS, has been an important method for traffic information source integrating, resource sharing, system efficiency improving and traffic congestion reducing.

According to the basic theories on the dynamic integrated traffic information platform, we analyzed the dynamic traffic information service demands in ATIS with distinct aspects which include main body of user, different application demands and function demands, and then we bring forward the research object— to use real-time multi-traffic information resource to estimate traffic states on links, routes and networks. The total thesis will be unfolded obeying the principle of from easy to propagating and from “point” to “face”. In this thesis, we will not only present an integrated theory for the estimation traffic states of urban road networks, but also we will use case study and micro-simulation methods to validate the theories mentioned in this thesis.

As for the theory of link traffic state estimation, we analyzed the traffic stream characters on the links and set up a criterion and Automatic Algorithm for traffic stream congestion detection. It is indicated that the method provided in this thesis is very significant for the link traffic congestion detection and it also very useful in real world engineering projects.

As for the theory of route traffic state estimation, we firstly analyzed the traffic stream characters on the routes, and then we selected the license auto-recognized system as the input of our study. Using the data from such system, we built up an integrated statistics model to estimate the travel time on the route. For the application aims, we considered different scenarios for the facility failure and lacking of information. Finally, we provided a case study which was implemented in Shanghai.

As for the theory of network traffic state estimation, we study the basic network compartmentalization method and data fusion procedure with GIS technology. With the help of Micro-Simulation tools, we provide a new method to estimate traffic state on the road network which is based on GPS vehicle positioning data and geographic information system data. Finally, we validate the method by using the data from real world and draw the conclusions that the model present in this paper can provide finer levels of resolution and better accuracy in estimating travel time and speed of the network.

In this thesis, we considered an integrated method for traffic information key parameters analysis, which is the primary part in dynamic traffic information system. The solutions provided in this thesis are helpful for coming into being independence knowledge property right dynamic traffic information system.

Key words：Traffic information, data fusion, state estimation,control delay

储浩

入学时间：2004级

答辩时间：2007年

论文题目：城市道路交通信息采集基础理论与关键技术研究

中文摘要

摘要

城市道路交通是现代都市文明的核心基础架构之一，交通信息则是直接影响现代都市人日常生活最重要的公众信息类别。交通信息化水平已经成为衡量一个城市现代文明发达程度、城市综合竞争实力最重要的指标之一。交通信息采集是城市道路交通信息化与智能化的基础，开展城市道路交通信息采集基础理论与关键技术研究具有实际价值和现实意义。

本文基于我国城市道路交通系统特点，立足城市道路交通系统中实际交通信息采集问题，从实验交通工程学的角度，结合交通工程学和信息科学理论知识研究交通信息采集基础理论和关键技术，获得的主要结论如下：

（1）系统提出了研究交通信息采集的技术路线，即“需求→理论→技术→应用”，围绕“为什么采集→怎么采集←→采集什么”开展交通信息采集理论和技术研究，对后续交通信息采集系统的各项研究具有指导作用；

（2）交通信息采集是对交通系统在时间和空间上的一个采样过程，交通信息采集数据具有时间和空间属性。对于时间采样，本文通过功率谱分析认为：不同数据采样间隔，获得交通信息不同，从而分析出的交通系统特征也不同；对于空间采样，本文通过采样定理分析认为：可以利用交通信息采集空间属性对交通系统特征的影响来指导检测器布设；

（3）通过实际数据分析和研究表明：在正常交通状态或者事件发生情况下，交通信息采集空间（位置）不同，获得的交通流特征也不同，不存在一种完全准确和通用的交通模型描述整个快速路的交通特征，需要根据不同位置选择合适的交通流模型；通过小波理论分析认为在交通信息采集的时间粒度在一临界值之后同一地点的交通信号存在较强的自相似性，而小于该临界值的信号自相似性不明显，因此在研究交通系统各种特征时还需考虑采集系统的时间粒度；

（4）采集数据的空间属性、时间属性和数据质量直接制约了交通信息处理的结果。其中交通信息采集密度和交通信息采集频率集中反映了采集数据的空间属性、时间属性对交通信息处理结果和算法精度的影响；

（5）多元交通信息采集方式的相互搭配可以实现功能性互补，保证交通信息采集的可靠性，本文提出的多元交通信息采集方式优化选择技术，可以指导工程中进行合理选择、组合多元固定交通信息采集方式，增加交通信息源的可靠性；

（6）基于相似性理论及交通信息处理算法与交通信息采集的互动关系（行程时间估计为例）来进行检测器的优化布设，可以达到对已布设的检测点进行重要度等级划分、对未布设的交通信息采集系统确定合理检测器布设密度的目的，使交通信息采集系统的各种效能合理化；

（7）交通移动采集技术能够和现有城市交通中固定采集技术很好的融合，并且两种交通采集技术在时空上能够实现互补。智能公交车辆可以作为交通移动采集技术的一种，采集的各种实时交通数据经过处理后可以用来分析公交车行驶道路的交通状况；

（8）以WSN为例，形成了信息新技术应用于交通信息采集的技术路线，探讨了WSN技术作为智能公交系统的信息采集和传输手段，所构建的系统和现有技术、系统相比具有很多优势，是一次有益的技术应用探索。

本文通过运用先进的信息技术和方法，初步建立和形成了交通信号频谱分析、交通信息采集系统时空特性分析、交通信息采集系统时空优化、交通信号的小波分析等交通信息采集基础理论和方法；并在交通信息采集系统时空特性与交通系统特征和算法模型的互动关系研究、基于交通系统特征的采样定理和频谱分析理论等方面实现了交通信息系统理论研究的拓展和深入，部分研究成果已具体应用到上海快速路交通信息采集系统的设计与建设过程中，获得了很好的效果。

综上所述，本文以实验交通工程学理论为指导，面向中国城市道路交通复杂大系统，坚持理论与实践相结合，研究方法和结论具有理论意义和实际价值。

关键词：交通信息采集，实验交通工程学，采样定理，交通信号频谱分析，多元交通信息采集，检测器优化布设，公交探测车，WSN

英文摘要

ABSTRACT

City road traffic system is the core and basal parts of city. Traffic information is one of the most important public information sorts which affecting the peoples’ daily lives. Traffic information level have been become the most important target for evaluating the city civilization and complex competition. Traffic data collection is the basement of city road traffic information and intelligent. There are the actual values and practical meanings for research base theory and key technology.

In paper the real characteristics of city road traffic system in China have been considered based on experiment traffic engineering theory. The traffic engineering knowledge and information theory have been combined for research. The main conclusions could be shown as follow:

(1) The technical route, ‘Requirement→Theory→Technology→Application’, for research traffic information collection have been shaped systemically in paper. Research method, ‘Why collecting→How collecting←→collecting What’, can direct the further research on traffic information collection system.

(2) Traffic information collection is a sampling progress for traffic system in time and space. There are time and space attributes for traffic data. For time sampling, according to the power spectral density (PSD) analyzing, different traffic information could be obtained based on different sampling interval data and then the traffic system characteristics are different analyzed by these data. For space sampling, according to the sampling theory, the traffic data space attributes could be used for directing detectors location.

(3) According to actual data, in normal traffic conditions and accident conditions, analyzing the data from different collecting location can achieve the different traffic characteristics. There is not an absolutely exactly and universal traffic model for describing the whole expressway traffic characteristics. So we must select the appropriate traffic model considering different space. According to the wavelet analyzing the traffic signal has self-comparability phenomena at the same space, when the time granularity of traffic information collection system is more than certain time and self-comparability phenomena could not been discovered when the time granularity is less than certain time. So the time granularity is another important factor for researching traffic system characteristics.

(4) Space and time attributions and quality of traffic data have affected the results of traffic information progressing directly and the traffic information collection density and frequency have reflected these affection.

(5) Traffic multi-styles collection can assure the reliability of traffic information collection system through detectors’ function inter-compensation. In paper the selecting method of traffic multi-styles collection styles can direct the practical project.

(6) The detector optimizing location methods, based on comparability theory and relationship between traffic arithmetic and collection system, can direct the detectors’ location. It also can plot the importance of existing detectors and make the reasonable traffic data collection density for constructing the new traffic information collection system. All of this may achieve the reasonable efficiency of traffic information system.

(7) Traffic moving collection technology can combine with the traffic fixed collection technology for function inter-compensation. Intelligent bus can be looked as one type of traffic moving collection technology. The real data from these buses can be used for analyzing the bus road traffic conditions.

(8) The application route of new information technology using in traffic system has been research in paper taken wireless sensor network (WSN) as example. The intelligent bus system constructing by WSN have many traits comparing with existing system. The research is a profitable technology application exploration.

In paper there are some innovations in information technology and traffic method, for example research methods of traffic information collection, traffic signal frequency and spectral analyzing methods, time-space attribution analyzing and time-space optimizing of traffic data collection system and wavelet analyzing etc. Some methods, the relationship between time-space attribution of traffic data collection system and traffic system characteristic and arithmetic model, sampling theory and frequency and spectral analyzing based on traffic signal characteristic, make the traffic information system research further. Parts of research results in paper have already been used in practical projects construction in Shanghai city and the effects are good enough.

So based on the directing of engineering traffic theory and facing the big and complex traffic system of China there are theory meaning and practical value for the methods and results in paper.

Key Words:Traffic information collection, engineering traffic theory, sampling theory, traffic signal frequency and spectral analyzing, traffic multi-styles collection, detector optimizing location, bus probe, WSN。

俞洁

入学时间：2004级

答辩时间：2007年

论文题目：城市公共交通枢纽布局选址理论与方法研究

中文摘要

摘 要

公共交通枢纽作为锚固城市公共交通网络体系的基础以及各种客运交通方式衔接的纽带，是公共交通系统的重要基础设施。公共交通枢纽布局与选址研究是进行公共交通枢纽科学规划和建设的切入点，功能完善、级配合理、分布有序的公共交通枢纽是城市一体化综合客运系统得以正常运行的前提与基础。在我国及许多国家的大城市，公交（本文特指公共汽、电车交通）系统规模巨大，网络复杂，为数量庞大的出行者提供运输服务，在城市客流运输系统中占据非常重要的地位。论文主要针对现有理论与方法的空白点和薄弱点，在大量查阅相关文献和研究成果的基础上，综合利用系统工程理论、交通工程学、网络优化理论及计算机软件工程理论等对城市公共交通枢纽布局选址理论与方法展开系统研究，以期推动和完善城市公共交通枢纽的规划设计理论。

首先，论文从机理解析层面，依据公共交通发展不同时期（近期与远期）的特点，分别构筑了已知线网与未知线网条件下的公共交通枢纽网络布局模式。已知线网条件下公共交通枢纽网络布局模式可概括为“枢纽服务分区、线路干支分离”；未知线网条件下公共交通枢纽网络布局模式可概括为“枢纽分级、线路分类、服务分区”。

其次，论文从需求分析层面，分别提出了公交线网已知与公交线网未知条件下公共交通客流需求分布的研究思路与客流需求分析方法。在公交线网已知的条件下，本文创新性地尝试了根据多组公交站点上下客实际观测数据,运用改进的BP神经网络对公交线网OD矩阵进行推算，并结合实际案例验证了模型与算法的有效性；在公交线网未知的条件下，本文建议沿用国内外已普遍使用的较为成熟的“四阶段”法。

再次，论文从网络优化层面分别提出了公交线网已知与公交线网未知条件下的公共交通枢纽布局选址的研究思路、建立了以系统运输效率最大化为优化目标的公共交通枢纽布局选址模型并设计了模型的求解算法。在公交线网已知的条件下，论文提出了“基于站点上下客人数”的一阶段布局选址优化方法，该方法实现了“枢纽服务分区、线路干支分离”的数学表述；在公交线网未知的条件下，论文提出了“宏观布局、微观选址”的二阶段布局选址优化方法，该方法实现了“公交枢纽分级、公交线路分类、公交服务分区”的数学表述。此外，论文还结合了实际案例验证了上述模型与算法的有效性。

随后，论文从系统评价层面建立了公共交通枢纽布局选址方案综合评价的指标体系、确定了各评价指标的具体量化方法并将“基于改进层次分析法的模糊综合评价模型”应用到综合评价分析方法中。此外，论文还结合实际案例验证了该评价分析方法的有效性。

最后，论文从战略决策层面开发了城市公共交通枢纽布局选址辅助决策系统，包括系统结构设计、功能设计与优化流程设计，并基于对中观交通仿真系统TranSmart的二次开发将公共交通枢纽布局选址系统加以实现。

论文的创新点包括：从已知线网与未知线网两个层面设计了公共交通枢纽网络布局模式，深入研究了已知线网条件下基于站点上下客人数的公交客流OD推算模型、已知线网条件下公交枢纽布局选址模型及未知线网条件下公交枢纽布局选址模型（上述两类布局选址模型均克服了传统“流量决定型”模型的缺陷，通过引入服务子区在模型中实现了交通与土地利用互动的数学表述），并进行了公交枢纽布局选址辅助决策系统软件实现的探索。为了能够更加真实地反映现实世界的交通情况，本文构筑的枢纽布局选址模型还有待在后续的研究工作中进行延伸和拓展。此外，开发出更加高效的优化算法也是后续研究的重要内容。

关键词：公共交通枢纽，布局选址方法，公共交通网络，整数规划