蔡云

入学时间：2006级

答辩时间：2009年

论文题目：嵌入式单点交通感应控制系统设计研究

中文摘要

摘要

目前中国交通控制系统的发展正处于标准化逐渐完善、技术积累逐渐成熟和中国特色日益明显的交汇期。单点交通控制系统作为城市交通控制的基本单元尤其受到关注。协调控制功能、交叉口群控制功能、网络控制功能的实现均是在其基础上的扩展和集成。

在单点交通控制领域，主要存在控制理论及系统开发两方面的研究。在交通控制理论的研究方面，相关学术机构及高校往往具有优势，但在硬件开发和系统集成上无法与市场接轨，往往不能使复杂的模型算法得到有效利用。而硬件设备生产厂家虽然具有较强的开发实力，但由于对交通控制科学理论方面认识的缺乏，无法充分理解交通控制策略及模型算法。此外由于交通控制系统本身是一个缺乏标准化的开放的复杂的系统，各家生产单位在研发系统的过程中又往往“各自为政”，导致系统的运行效果、兼容性以及可扩展性都不尽如人意。因此本文以国家自然科学基金重点项目——“城市道路信号控制交叉口群交通阻塞机理”为依托，寻找控制算法与硬件开发的结合点，研究以当前发展迅速且技术日趋成熟的嵌入式技术为实现平台，单点控制理论为基础的交通控制系统设计方法，旨在为填补国内单点交通控制系统设计理论的缺乏与不足提供一定的借鉴作用。

本文以交通控制系统工程学为研究视角，运用系统设计方法，在理论和应用两方面展开以下研究内容：

国内外交通控制系统及其相关技术梳理归纳——论文在查阅大量国内外文献的基础上，围绕本文研究对象——嵌入式单点交通感应控制系统，将前人的研究成果分成四方面进行总结。首先对国外成熟交通控制系统的系统组成、功能特征进行分析，进而对交通信号控制器国内外两方面的发展轨迹进行梳理。然后针对目前控制器先进的实现技术——嵌入式系统，从结构、处理器以及操作系统的发展脉络进行了归纳。最后详细总结了具有不同控制规则和算法多种感应控制策略的原理，并指出单点交通控制系统的发展趋势和研究的切入点，即以嵌入式系统为实现平台的单点交通感应控制系统的设计研究。

单点感应控制理论分析与系统需求分析——基于交通控制系统工程学的系统设计方法决定了系统的设计必须从系统拟实现的功能着手。论文描述了单点感应控制策略的上层参数计算获取方法（包括最小绿灯时间、最大绿灯时间和通过时间），分析了标准相位结构的特点和组合方案。在此基础上论文制定了系统需求分析的架构内容，并对系统的功能性需求，包括功能需求、数据需求以及非功能性需求，包括运行需求、接口需求进行了全面的阐述。

嵌入式单点交通感应控制系统框架体系与方案研究——在系统需求分析的基础上，从交通控制系统结构入手，对系统的模块进行划分，并建立了与控制主板以外设备的映射，设计了以嵌入式处理器为核心的硬件结构标准化的系统框架。对系统的数据流进行分层描述，并给出了系统的处理器、操作系统以及网络通信协议的选择方案。

考虑安全的单点感应控制算法研究——与以往的单点感应控制策略不同，本文在实现感应控制基本功能的前提下，提出了以解决“两难区”问题为重点的控制算法。在双环结构的基础上，将相位切换点划分成两类。针对第一类切换点，提出了将绿灯延长时间根据不同控制目标划分为两个阶段——绿灯延长阶段一及绿灯延长阶段二。这对第二类切换点，提出了相位组的切换算法。根据控制算法的要求，设计了预测时间窗和观测时间窗方法，并建立了上游检测器的距离模型。

嵌入式控制主板的硬件结构以及感应控制软件设计研究——在系统原型的基础上，以嵌入式微处理器为核心扩展功能模块，设计了系统的硬件组成结构，指出了微处理器I/O及引脚输出的分配。并针对各个功能模块进行了分析阐述。根据单点感应控制算法以及嵌入式软件实现要求，分析了控制软件程序的分析和建模过程，并给出三大模块的程序流程。

本文创新点包括：以交通控制系统工程的崭新视角剖析了城市嵌入式单点交通感应控制系统的需求分析、框架组成、算法模型以及开发实现过程；建立了一种考虑安全的单点感应控制策略。设计预测时间窗与观测时间窗来动态预测车辆进出“两难区”的时间点，并对相位组切换算法进行了研究；建立了一种适用于单点感应交通控制系统，面向本文提出算法的检测器布设方法和距离模型。

关键词：嵌入式，单点，感应控制，系统设计

英文摘要

ABSTRACT

At present, the development of traffic control system in China is in the process of being standard, mature and characteristic. As the basic unit, isolated traffic control system is paid more attention. The coordinated control, interection group control and network control are all the expansion and integration on it.

In the field of isolated traffic control, the research can be divided into two aspects: control theory and system development. In the aspect of traffic control theory, the colleges have the advantages except the system integration. So the complicated algorithm can not produce the whole effect. In the contrary, the enterprises have the strong ability of hardware exploitation, but can not understand the algorithm and model. As the system is an opened and complicated one, each company has its own rule of producing. As the result, the running effect, compatibility and expansibility of the system are all less than expectation. Supported by the National Natural Science Foundation of China (Grant No. 50578123), this research tries to find out the combination of the control algorithm and hardware integration and use the embedded technology to design the isolated traffic control system, which is expected to make complete of the deficiency of the theory of the system design in China.

The dissertation conducts the research through the view of traffic control system engineering. The main content includes:

The Literature Review of Traffic Control System and Its Interrelated Technology—Centered on embedded isolated actuated traffic control system, the research gives a review of the profit has been achieved in the past. At first, the dissertation analyses the system components, function characteristics of the mature control system. Then the development of traffic controller in domestic and abroad is introduced. After that, the embedded technology, including the system structure, processor and operation system is summarized. The chapter also reviewed the strategies and algorithm of the isolated actuated control. At last, the tendency of the development of system is indicated. The whole review extracts the breakthrough point of the research.

Isolated Actuated Control Theory and System Demand Analysis—The design approach based on the traffic control system engineering begins from the system function. The paper describes the model for the actuated upper parameters, including the minimum green, maximum green and the passage time, and analyzed the dual-ring structure and its combination. On the basis, the content of system demand analysis is proposed. The content contains the system function, data, the demand for system running and the interface, which are all expounded.

The Prototype and Plan of Embedded Isolated Actuated Control System—According to the system analysis, the dissertation analyzes the structure of system, defines the function modules and indicated the mapping of them. The system structure is hardware-standardized in the core of the embedded processor. The data graph is divided into different level and described in the detail. The plan of the system is introduced as the core processor, operation system and the network protocol.

An Isolated Actuated Algorithm Considering Safety—Being different with the strategies in the past, the actuated algorithm proposed a control approach to resolve the dilemma zone. According to the dual-ring structure, the phase decision-making is classified as two stages. The dissertation proposed that the green extension can be cut up to two parts—Green-Extension1 and Green-Extension2. The algorithm for phase group change is designed for Green-Extension2. And facing the Green-Extension1, the more important one, the dissertation makes the prediction and observing window and builds a distance model for upper detector zone.

The Hardware Structure of Embedded Control System and Actuated Control Software Design—On the basis of system prototype, the function modules, which are expanded centered on the micro-processor.，are introduced in detailed view. The actuated control software is designed according to the algorithm described above. The dissertation analyzed the process of software modeling and explained the flow of the three main procedures.

The innovative points in the dissertation includes: 1)Analyzed the demand, structure, algorithm and development of the embedded isolated traffic control system through a new view of traffic control system engineering; 2)Builded a isolated actuated strategy and algorithm considering safety, designed the time window to predict the moment in which the vehicles fall into the dilemma zone and proposed the decision-making way for phase group; 3)designed the method of detectors placement and builded the distance model for upper one.

Key Words: embedded system, isolated intersection, actuated control, system design

江南

入学时间：2006级

答辩时间：2009年

论文题目：快速路入口匝道控制模式及放行方式适应性研究

中文摘要

摘要

现在许多大城市如北京、上海已建成自己的快速路系统，许多其他城市也正在建设或筹划建设快速路的过程中。但是已建成的快速路系统在运行过程中经常存在匝道车辆无法进入主线或由于匝道车辆的汇入导致主线拥堵、匝道与主线汇合区安全事故频发等现象。导致这些现象的原因主要是入口匝道是否应该实行控制、应该采取何种放行方式和是否可以对主线进行控制等问题没有得到解决。而解决这些问题是保证匝道控制在实施后能达到预期效果的重要前提。因此,本文在自然科学基金重点项目的支持下，对解答这些问题进行了一定的探索。

针对是否应该采取入口匝道控制的问题，本文对入口匝道有、无控制两种模式的适应性进行了分析。利用仿真对比研究了快速路系统在两种控制模式下的运行状况及系统运行效益指标（主要包括快速路车辆平均出行时间、主线车辆延误、匝道车辆延误等），得出了入口匝道有、无信号控制两种控制模式各自适应的交通条件，如主线流量水平，匝道流量等。研究结果为在什么情况下应该采取入口匝道控制提供了依据。

针对采取何种放行方式的问题，本文对三种放行方式（单车道单车放行，单车道双车放行和双车道放行）分别在三种控制模式（入口匝道无控制、入口匝道采取ALINEA控制和对入口匝道与主线均控制）下对系统运行效益指标的影响进行了研究，得出了三种放行方式在不同控制模式下的适应性，为放行方式的选择提供了参考依据。

针对是否可以对主线进行控制的问题，本文基于ALINEA入口匝道控制算法，设计了一种新的对主线和匝道均进行信号控制的控制策略。该策略弥补了大多数匝道控制算法由于完全优先主线而导致主线富余通行能力无谓浪费的缺陷。通过对比分析新的控制策略与仅对入口匝道实施ALINEA控制策略下的快速路系统运行效益指标，发现新的控制策略虽然导致主线车辆的延误有少量增加，但却降低了所有快速路车辆的平均出行时间并显著提高了匝道车辆的运行效益，从而验证了所提控制策略的可行性与先进性。

本研究主要在以下几方面有所创新：

1.通过分析不同入口匝道控制模式适应性，回答了什么条件下应该实施入口匝道控制的问题，解决了快速路建设过程中盲目实施入口匝道控制的问题；

2.回答了何种情况下应该选择何种放行方式的问题，解决了入口匝道车道设计和实施控制后放行方式选择的问题；

3.提出了一种新的同时控制主线和匝道的控制算法，证明了控制主线的可行性，并探讨了其适用性。

关键词：匝道控制，ALINEA，适应性，仿真实验

英文摘要

ABSTRACT

Now, many big cities including Beijing, Shanghai inChinahas built freeways and many other cities are going to build. However, phenomenon including ramp vehicles can not merge in mainline, congestion happens often in bottleneck area and traffic accident in merging area can be observed often. The reason of these problems is that there is no clear answer for whether one on-ramp should be controlled, which kind of flow control method should be chosen and whether mainline vehicles can be controlled. Solving these problems is an important precondition for guaranteeing ramp control’s effectiveness. So, supported by the Key Project of Natural Science Fund, this research went a step forward solving these problems.

For the question that whether ramp metering should be applied, this research analyzed adaptabilities of two control modes, on-ramp with and without control. Using micro-simulation, this thesis analyzed and compared freeway system’s performance and evaluation indexes including freeway vehicles’ average travel time, mainline vehicles’ delay and on-ramp vehicles’ delay. As a result, traffic conditions that two control modes (with and without ramp control) adapt were given. The result of this part’s research provides basis for deciding when to apply ramp control for an on-ramp.

Then, for the question of choosing which flow control method, this thesis researched the influence of three flow control methods, one lane one vehicle per green, one lane two vehicles per green and two lanes one vehicle per green for each lane on freeway system performance under three control modes, no ramp control, with ramp control and control both on-ramp and mainline. Then, three flow control methods’ adaptabilities under different control modes were given, which lays a foundation for choosing best flow control method.

Finally, for the question of whether mainline can be controlled, basing on ramp metering algorithm ALINEA, this research proposed a new control strategy of controlling both the on-ramp and mainline. This strategy, compared with most of former ramp metering strategy, takes good use of excess capacity of mainline. By comparing system performance when this algorithm is applied and when only ALINEA is applied, it is found that although the new strategy increases mainline vehicle’s delay slightly, it reduces average travel time of all vehicles in the freeway system (including vehicles on ramp and on mainline), which validated this new strategy’s feasibility and reasonability

To sum, this research is innovative in following three aspects:

1.Through analyzing the adaptability of different ramp control modes, answered the question that under what condition should ramp control is applied, which avoid blindly decision during implementation.

2.Answered the problem that when should which flow control method was chosen which provide guidance for deciding the number of on-ramp and way of flow control.

3.Proposed a new strategy which control both the on-ramp and mainline, which proved the feasibility of controlling mainline. The adaptability of this new

潘振兴

入学时间：2006级

答辩时间：2009年

论文题目：基于手机信息的交通小区划分与OD预测问题研究

中文摘要

摘要

长期以来，社会经济的发展与交通的发展一直存在着互相促进又互相制约的关系。如何让交通拥挤等问题不再困扰社会和经济发展是业界人士孜孜以求的目标。加强城市交通管理对策及其相关技术的研究，对充分挖掘现有交通系统的潜力和提高交通管理水平具有重要的意义。同时人均手机保有量的激增在带给我们通信便利的同时还产生了大量数据，因此若能从中挖掘出交通管理所需的信息，将极大地促进交通信息化的发展。

本文基于上述背景，特别针对实时交通状态分析关键技术——交通小区划分与OD预测展开研究。

论文首先从动态交通小区划分与OD预测信息需求分析入手，在对移动通信运营商调研和分析移动通信技术现状的基础上，通过对手机数据的格式、采集频率、蜂窝小区覆盖范围等特性的定性分析，得出将手机信息应用于动态交通小区的划分及动态OD预测是可行的。

在此基础上，本文利用Agent技术，以JAVA为开发工具，建立了包括虚拟路网，车辆和基站等对象的实验平台。平台中的各种参数采用安德伍德指数模型等经典的交通流模型和现实手机运营商的运作模式来进行标定。通过该平台模拟现实世界的交通及手机运营规则，从而获得了大量模拟手机数据。这一方法的准确性在理论上亦得到验证。

随后，通过数据挖掘，从实验所得的手机数据中得出一定时间段内各基站覆盖区的OD量，并基于人工神经网络的方法，利用MATLAB7这一工具，估计出了各基站覆盖区内的交通状态，同时以单个基站覆盖区为最小单位，将某一时刻地理位置上相邻的、且交通状态一致的基站覆盖区合并成面向交通管理的动态交通小区。

最后，在研究了基站覆盖区内动态OD的产生机理，明确了动态OD与基站覆盖区内的路段断面上的交通量存在着一定的联系后，参考断面交通量的预测方法，利用多步短时预测法对基站覆盖区内的动态OD进行了预测。

本文的主要研究成果是在不改变真实手机数据表结构的前提下，结合交通运行规则，利用Agent技术研究了产生手机模拟数据的方法，构建了基于Agent的模拟手机数据产生平台；结合交通需求与移动通信的数据特性，建立了基于基站覆盖区交通状态的动态交通小区划分方法；克服以往的不足并借鉴断面交通量预测方法，提出了简便易行的动态OD预测方法。

关键词：动态交通小区，动态OD，交通预测，Agent，神经网络

英文摘要

ABSTRACT

The development of social economy and that of transportation have always been mutually promoting and restricting. How to avoid traffic congestion becoming obstruct of social and economic development is transportation professinals’ goal. Advancing the research of urban traffic management policy and related technology is meaningful for taking good use of present transportation systems and improving transportation management. At the same time, increasement of mobile phones’ number not only makes our life more convenient but also brings about a great number of data, so if from which the information needed by transportation management can be mined out, the development of transportation informationization will be promoted greatly.

Based on this background, this paper researched traffic zone’s delimitation and OD prediction which are key technologies for analyzing real-time traffic state.

Firstly, this research analyzed the demand for dynamically dividing traffic zone and predicting OD information. Based on investigating mobile operator and analyzing the present state of mobile communication technology, through qualitativly analyzing the format of mobile information, collecting frequency and cellular zone’s cover area, the feasibility of applying mobile information into dynamic delimitation of traffic zone and OD prediction was proved.

Then, using Agent and JAVA, virtual road network, vehicles and base stations were established. Various references of this experiment network were calibrated by classical traffic flow model such as Underwood Index model and operation mode of mobile operator. This experiement network simulated real world’s transportation and mobile operation rules, and then generated a large number of simulated mobile information. The correctness of this method was proved theoretically.

Secondly, each station’s cover area’s OD was mined from mobile information obtained from experiment. Based on artificial neural network and using Matlab 7, the traffic state of each station area was estimated. Then, those areas which near each other and have similar traffic state at the same time were combined as one traffic zone oriented to traffic management.

Finally, after analyzing the generating mechanism of each station’s cover area’s dynamic OD and realizing the relationship between dynamic OD and traffic volume of roads in the area, based on the method for predicting road’s traffic volume, each area’s OD was predicted by using the multi-steps short time predicting method.

The main research results of this thesis including: under the precondition of not changing real mobile data form structure, combining traffic rules and using Agent, proposed a method for generating simulated mobile information and constructed a simulated network for generating simulated mobile data; combining the characteristic of traffic demand and mobile communication data, and basing on station area’s traffic state, a method to dynamically divide traffic zone was proposed; in addition, a simple method for dynamically predicting OD was given which overcomes former methods’ disadvantages and uses volume predicting method for reference.

Key Words:dynamic traffic zone, dynamic OD, traffic prediction, Agent, neural network

赵靖

入学时间：2006级

答辩时间：2009年

论文题目：考虑上下游影响的信号控制交叉口通行能力研究

中文摘要

摘要

道路通行能力反映道路系统对交通的处理能力，在交通学科研究供给和需求这对矛盾中，通行能力扮演了客观反映道路供给能力的角色。因此，关于道路通行能力的研究一直以来是交通领域研究的中心内容之一。深入研究信号控制交叉口通行能力是进行城市道路科学规划、设计和管理的重要切入点，明确城市道路交叉口所能提供的服务能力是城市道路合理建设以及正常运行的前提和基础。

论文主要针对现有理论与方法在考察断面和影响范围的局限性，在对相关研究成果进行综述的基础上，确定以考虑上下游道路交通条件影响为核心思想，即研究的焦点不仅仅停留在停车线上，而是将研究视野拓展到停车线的上下游，对信号控制交叉口通行能力展开研究。研究中综合利用交通工程理论、概率论及数理统计和计算机仿真等方法对问题进行解析和验证，以期推动和完善通行能力理论体系。

论文首先从解析机动车流通过交叉口的实际运行过程出发，从中寻找可能造成通行效率折减的因素，综合考虑我国城市道路特征、研究的迫切性、以往研究的完善程度等因素，将本文研究重点集中于短车道、出入口和交织段这三个影响因素进行深入探讨。

在对短车道影响研究中，从定性分析和定量计算两方面研究了短车道排队阻塞对信号交叉口通行能力的影响。论文建立了考虑短车道潜在排队阻塞情况下的交叉口通行能力计算模型，并通过仿真对模型进行了检验。在此基础上，进行了短车道长度、信号相位相序、信号周期等相关因素敏感性分析。

在对进出交通影响研究中，分析了城市道路进出节点交通流构成及其对主线通行能力的影响，提出了考虑进出交通影响的信号控制交叉口通行能力计算方法。在此基础上，对车道组车道数、进出交通节点与交叉口相对位置、进出交通节点与交叉口间距等因素的敏感性进行了分析。

在对交织段影响研究中，利用数据拟合确定了交织段通行能力的基本模型，并结合车流在交叉口到达驶离特性，建立了交织段影响下的交叉口通行能力计算模型，并通过仿真对模型进行了检验。在此基础上，进行了交织段长度、进口道长度、信号周期等相关因素的敏感性分析。

论文主要从微观层面对考虑上下游影响的信号控制交叉口通行能力计算方法展开研究。论文创新点包括：在学术思想方面，将研究信号控制交叉口通行能力的关注点从停车线拓展到交叉口的上游和下游，基于交通流的“流入”“流出”关系，分析通行能力的影响因素；在细部分析方面，主要有3点突破：1、增加了短车道影响模型的适用范围，2、提出了考虑交叉口上下游存在进出交通节点影响的信号控制交叉口通行能力定量计算方法，3、提出了考虑交叉口上游存在交织段影响的信号控制交叉口通行能力定量计算方法。

关键词：交通工程，通行能力，理论解析，信号控制交叉口，交通流