【作者】 张萌萌；

【导师】 贾磊；

【作者基本信息】 山东大学 ， 检测技术与自动化装置， 2011， 博士

【摘要】 随着社会经济的发展和城市化进程的加快,城市人口数量以及机动车保有量呈现逐年上升的趋势。城市道路交通拥挤和交通阻塞已经成为世界大中城市普遍存在的问题,并且由此带来了一系列的社会问题,比如交通事故剧增、环境污染、能源紧缺等。大力发展城市交通智能控制系统是解决现代社会交通需求与供给矛盾的重要途径之一。实现城市交通系统的智能控制不仅有利于提高交通运输效率,增加道路交通运营安全,而且还关系到土地资源与能源的合理利用、城市环境的改善,乃至国民经济的持续发展和社会效益的提高。基于智能计算方法的城市交通信号控制是城市交通智能控制的重要内容之一,对提高城市道路网络通行能力,减少车辆行车延误具有重要意义。目前,我国大中城市的交叉口多为信号控制交叉口,并且大部分信号控制机采用定时信号控制策略,路网部分交叉口布设了视频检测器或者感应线圈用以检测交通流量。本论文立足我国城市交通基础设施建设实际,尝试构建一个基于交通模式识别的交通信号控制系统：如果控制区域内所有交叉口均布设车流检测设备,则根据车流检测设备实时采集的交通流信息,进行交叉口交通模式识别,进而调用与交通模式相适应的信号控制方案；如果控制区域内部分交叉口未布设车流检测设备,则根据车流的时间分布规律对交叉口交通模式进行识别,将控制时域划分为若干时间段,针对不同的时间段调用相应的信号控制方案。该交通信号控制系统的方案库是预先给出的,信号配时优化算法将干线协调控制与单点定时鲁棒控制相结合,在保证信号控制系统效率的前提下,有效提高了信号控制方案对于流量波动的稳定性。对于布设车流检测设备的区域,该系统采用方案选择式信号控制策略,能够充分利用现有硬件设备,提高信号控制方案的效率以及稳定性；对于未布设车流检测设备的区域,该系统采用定时信号控制策略,对交通信号机的要求低,实施和维护费用低,能够以最少的经济投入,提高交叉口通行能力,最大限度地提高信号控制路网车辆运行效率,减少车辆延误及停车次数,进而降低车辆燃油消耗以及尾气排放。论文针对构建该系统的关键技术问题展开研究,主要包括了以下六个方面：(1)在总结和评述国内外学者对于交通流模型研究成果的基础上,基于元胞自动机原理,提出了一个更具普适性的信号控制路网交通流模型。通过改进开放性边界条件,利用一维元胞自动机模型模拟协调控制主干路交通流状况。该模型采用差分方程的形式描述车辆动态行为,解除了信号灯等间距布设的限制,每个交叉口的信号灯可以根据交通流变化自由选择绿信比,相邻交叉口采用绿波控制方式调整相位差。论文利用Matlab软件对模型进行仿真,分析了驶入主干路流量以及干、支路的转弯流量对于主干路平均速度、平均密度和平均流量的影响。在仿真结果的基础上,提出了相应的控制措施改善主干路交通状况,为城市交通信号配时方案库的建立提供了理论依据。(2)结合我国城市道路信号控制机以及车流检测器的布设现状,提出了一个基于交通模式识别的城市交通信号控制系统框架,分析了该系统的工作原理以及功能模块的构成,并对构建该系统所需的智能计算方法(粗糙集理论、模糊神经网络、遗传算法、元胞传输理论)的概念、原理以及适用性进行了阐述和分析。(3)基于粗糙集理论、模糊逻辑以及人工神经网络等智能计算方法,构建了用于交叉口交通模式识别的粗模糊神经网络模型。该模型由两个阶段构成：第一阶段,基于粗糙集理论的交通参数约简,得到能够表征交叉口交通特征的最小属性集合；第二阶段,利用第一阶段得到的最小属性集合,建立基于模糊神经网络的交叉口交通模式识别模型。该模型为交通参数数据采集、分析以及处理提供了必要的理论依据,为实现交通模式识别提供了技术支持,是建立交通信号控制系统的前提与基础。(4)针对单点交叉口定时信号控制无法适应交通流变化的缺陷,提出了单点交叉口定时信号鲁棒控制的思想。在信号控制方案效率优化的基础上,加入了改善控制方案鲁棒性的子目标。论文选择车辆平均延误标准差最小这一指标衡量信号控制方案的稳定性。并通过试算实验,得到了鲁棒控制目标与效率控制目标的权重系数,建立了多目标规划模型,对交叉口的周期以及绿信比两个参数进行优化。(5)建立了基于元胞传输模型的干线协调控制模型,以优化路网相邻交叉口的相位差。该模型利用元胞传输理论模拟信号控制主干线交通流,并以此为平台分析车辆延误、停车次数、车辆通过量等评价指标。以主干线车辆总延误最小,停车总次数最少以及单位周期内通过车辆数最多为目标函数,以相位差约束为约束条件建立了协调控制优化模型。并采用遗传算法对模型进行求解。求解结果表明,基于元胞传输模型的城市主干线协调控制优化模型有效降低了车辆延误,减少了车辆停车次数,提高了主干线通行能力。(6)设计交通信号控制系统算法,构建城市交通信号控制方案库,建立基于交通模式识别的交通信号控制系统。将单点交通信号鲁棒控制模型与干线协调控制模型有机结合为一个整体,用以优化交叉口交通信号配时参数(周期、绿信比以及相位差)。该模型充分考虑了城市主干线的交通流特性、上游交叉口排队长度对下游交叉口信号配时的影响以及信号控制对交通流的调节作用,为区域信号控制优化提供了思路和借鉴。最后,将此模型推广到区域范畴,并结合道路网络实际,进行了计算机仿真分析。仿真结果表明,该控制系统在不同流量状态下均能够有效地控制车辆延误、减少车辆的停车次数以及提高车辆通过量,控制方案的稳定性较好。最后,对论文的创新点以及不足进行了总结分析,提出了未来的研究重点和方向。更多还原

【Abstract】 With the development of economy and advancement of urbanization, urban population and vehicle number is increasing year by year. Traffic congestion and jam has become a common problem to be faced by large and medium-sized cities in the world. And a series of social problems brought by traffic congestion also trouble them, such as traffic accidents, environment pollution, energy crisis and so on. Developing urban intelligent traffic control system is one of the important ways to solve contradictions between traffic demand and supply. Implementing intelligent control of urban traffic system is not only beneficial to improve transportation efficiency and enhance road traffic safety, but also relates to making full use of land source and energy, improving urban environment, developing national economy and social benefit.Urban traffic signal control based on intelligent computation is one of the important contents of urban traffic intelligent control, which is of significance to improve traffic capacity of urban road network and reduce delay of vehicles. Recently, most intersections in larger and major Chinese cities are signalized, and signal controllers mostly adopt fixed time signal control strategy. Otherwise, video detectors and induction detectors are placed on the mass of intersecitons on arterial roads to detect traffic flow volume. Based on the construction situations of urban transportation infrastructure facilities, the paper tries to establish a signal control system based on traffic pattern recognition. If traffic flow detectors are equipped for all of the intersections of control area, suitable signal timing plans can be called according to traffic pattern recognition based on traffic flow information detected by detectors. If detectors are not installed in part of intersections of control area, control time domain will be divided into a series of sections according to traffic pattern recognition based on temporal distribution of traffic flow, and corresponding signal timing plans will be invoked at different time. Signal timing plan library is prepared beforehand, and optimal algorithm combines arterial coordinated control with robust optimization of signal timing at single-point intersections. For the control area without detectors, the system employs fixed time signal control strategy with low implement and maintain cost of traffic controllers. Besides, the system has ability of enhancing capacity of intersections, improving efficiency of vehicles on signalized network, reducing delay and stopping rate, lowering energy consumption and exhaust emission. For the control area with detectors, the system utilizes plan-selection signal control system, which can make full use of hardware equipments and has ability of improving efficiency and stability of control plans.The main contents of this thesis are listed as follows:(1) Based on summarizing and reviewing research fruits on traffic flow model of urban signalized road network, a traffic flow model of signalized road network with wide applications is proposed based on cellular automaton theory. A one-dimensional cellular automaton model with improved open boundary conditions is used to simulate the traffic flow on arterial roads with coordinated control system. The model employs difference equations to describe dynamic behaviors of vehicles. The restriction on regularly spaced distribution of traffic signal lamps can be eliminated. Furthermore, the split on every intersection can be chosen according to traffic flow fluctuations. The offset between adjacent intersections can be adjusted by green wave control. Matlab is employed to simulate this model to analyze the impacts on mean velocity, density and volume of arterial traffic flow by the flow volume on the artery and the turning flow volumes from branches. Based on the results of simulation, a series of proposals for improving the arterial traffic situations are put forward, which is a prerequisite for constructing urban traffic signal timing library.(2) In consideration of the placement situations of urban signal controllers and detectors, a framework of economical and effective urban traffic signal control system is proposed. The paper analyzes basic principle, structure and functional modules of the system, and then explains concepts, principle and applied scope of intelligent computation methods such as rough set theory, fuzzy neural network, genetic algorithm, and cell transmission model, which are used to construct the system. (3) Based on rough set theory and fuzzy neural network, a rough fuzzy neural network model is proposed to realize traffic pattern recognition of intersections. The model is comprised of two stages. At the first stage, traffic parameters are reduced based on rough set theory to obtain the least reduction of attribute set which can describe traffic characteristics. At the second stage, traffic pattern recognition model is built based on fuzzy neural network using reduction parameters above. The model provides necessary theoretical basis for data collection, analysis and processing of traffic parameters, and it provides technical support for traffic pattern recognition, and it is prerequisite to establish traffic signal control system.(4) The idea of robust control for intersections is proposed to remedy the disadvantage of fixed-time signal control that could not be suitable for large fluctuations of traffic flow. The sub-objective function is added to traditional optimal objective function. The robust objective function which strengths stability of signal control is to minimize standard deviation of vehicle delay. Based on simulation and analysis of intersections under various traffic conditions, the study establishes the relationship between sub-target weights and flow fluctuating ranges. Then the paper builds a multi-objective optimization model to optimize cycle length and splits of single-point intersections.(5) Coordinated control system of arterial roads is constructed to optimize offset between adjacent intersections. The model simulates traffic flow on urban signalized arterial road by cell transmission model, and constructs mathematic models of delay, stopping rate and traffic volume based on the platform. The paper proposes a optimization model to optimize the offset between adjacent intersections of coordinated control system. Its objective function is to minimize total delay and stopping rate and to maximize traffic capacity of arterial road. And its constraint condition is offset constraint. Genetic algorithm is executed by Matlab to solve the model. And the experiment results show that the model effectively reduces the delay and stopping rate of vehicles running on arterial road and largely improves traffic capacity of artery.(6) Combing single-point signal robust control model with coordinated control model of arterial roads, traffic signal control library is built to optimize signal timing of intersections, including cycle length, split and offset. The model provides thoughts for area signal control, considering traffic flow characteristics and the effect of queue length of upstream intersections on signal timing of downstream intersections and adjustment effect of signal control to traffic flow. Finally, the model is applied to area signal control. And combined with real road network, computerized simulation is carried out. The results show that the model not only effectively reduces average delay and stopping rate of vehicles running on arterial road and largely improves traffic capacity of arterial road, but also reduces the sensitivity of signal control for flow volatility.Finally, the paper summarizes innovations and shortcomings, and proposes future research priorities and orientations.更多还原