【作者】 唐殿奎；

【导师】 赵建玉；

【作者基本信息】 济南大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 当前,随着我国经济的快速发展,城市化进程的加快、车辆保有数量迅速增加。随之而来的交通问题成为严重制约我国城市发展和居民生活质量的重要瓶颈。面对急剧增加的汽车数量和有限的交通资源,如何缓解大范围的交通拥堵,提高路网的通行能力,减少交通事件的发生数量、避免宝贵资源的浪费成为城市交通路网管理迫切需要解决的问题。区域交通控制作为智能交通系统研究的关键技术,因其可以显著的缓解以上交通问题,具有著的社会意义和经济价值,必将是未来智能交通研究领域的热点和关键。本文以区域路网中的交叉口为研究对象,以实现交通流的有效疏导为目的,在广泛阅读相关资料的基础上,在以下几方面的做了深入的研究工作:(1)总结了当前交通控制领域的研究现状,其中包括单路口交通控制研究与发展,干线交通控制研究与发展,区域交通控制研究与发展。特别就智能计算在交通控制中的研究发展进行了介绍,就常用的智能计算方法:模糊控制、进化计算、神经网络分别阐述了其在交通领域的应用研究现状,同时作为后续章节的必要准备,概述了交通控制过程中交通控制的重要参数和概念。(2)以典型区域单路口为控制对象,构造了一个单路口离散化交通模型。该模型以车辆的放行流量和车辆到达率为输入常量,以排队长度为系统的状态变量,通过程序设计实现了交通模型的建模,采用爬山法优化了绿灯时间,减少了排队长度,从而验证了该模型模拟仿真交叉口通行状况的有效性,为后续研究提供了合适的参数尺度。(3)针对固定相位单路口交通流的有效控制问题,提出包含相位转换模糊控制器和延时模糊控制器的两级模糊控制方法,其中相位转换模糊控制器实现相位转换的判断,延时模糊控制器判定延时的绿灯时间,文中就两级模糊控制器的实现给出了设计过程,通过仿真与感应控制比较得出,此方法达到了缩短绿灯延时时间,减少延误,使通行更为畅通的目的,但适应效果受车流饱和情况的影响。(4)针对多路口区域交通控制具有复杂、随机、离散性强、受干扰因素多的特点,提出一种基于多智能体技术搭建的松散耦合关系的多智能体区域控制结构,并分别就其路口级控制多智能体结构设计、区域控制多智能体结构设计、中心控制多智能体结构设计给出了详细的描述。在具体的仿真过程中,以一个区域级交通智能体控制的路口单元为研究范围,重点研究了两路口在出现交通拥堵状况时的协调解决过程,在区域级智能体应对路口级智能体协调请求时,采用遗传算法优化控制参数,以实现对区域交通流的整体优化。结果表明,同等条件下,与传统的定时控制比较,各项指标性能有较大的提高,此方法有效可行。更多还原

【Abstract】 At present, with the rapid economic development, accelerating the urbanization process, the number of vehicles maintains rapid growth in traffic. Therefore traffic problems as the important bottleneck seriously constrain urban development and quality of life for residents. The face of rapid increases in vehicle traffic volume and limited resources, how to relieve a wide range of traffic congestion, improve road network capacity, reduce the number of traffic incidents to avoid the waste of valuable resources into Urban Road Network Management urgently needs to be resolved problem. Regional Traffic Control as Intelligent Transportation System key technology, as it can significantly alleviate these traffic problems, with the social significance and economic value, will be the next hot field of intelligent transportation and key.In this paper, in order to achieve effective to ease traffic flow for the purpose of information in a wide range of reading, the intersection of the regional road network in this study, based on the following aspects of the work done in-depth study:(1)Summarize the current status of research in the field of traffic control, including the intersection traffic control research and development, Route Traffic Control research and development, regional traffic control research and development. Particularly on Intelligent Computing in research and development of traffic control were introduced, the commonly used methods of computational intelligence: fuzzy control, evolutionary computation, neural networks are described in the application of traffic sector Research. At the same time as the necessary preparations for the follow-up section, an overview is given about traffic control important parameters and concepts in the process of traffic control.(2)To a typical single-intersection region as the control object, construct a single intersection discrete traffic model. In this model, the release of vehicle traffic and vehicles enter the constant arrival rate to queue length for the system state variables, through program design and implementation of traffic modeling, using small step test method optimized green time, reduce the queuing length, thereby validate the model simulation of the effectiveness of intersection conditions.(3)for a fixed phase isolated intersection traffic flow control problem, including phase change proposed fuzzy controller and fuzzy controller for two-stage fuzzy control method, in which the phase transformation of fuzzy controller to judge the phase transition, delay fuzzy controller determine the delay of the green time, the text on the two fuzzy controller to achieve given the design process, the simulation compared with the sensor control obtained, this method achieved the green light to shorten the delay time, reduce delays, make access more the purpose of smooth, but the adaptation effect of saturation by the traffic situation.(4)Considering multi-junction area traffic control complex, stochastic, discrete strong interference characteristics of many factors, the multi-agent control structure of the region is proposed based on multi-agent technology built of loosely coupled relationship and their control structure design of Intersection-level multi-agent, control structure design of regional multi-agent, structure design of the center control multi-agent are given a detailed description. In particular the simulation process to a regional multi-agent controller on the intersection traffic unit as the research scope, focus on the two junctions in the traffic jam situation of appears to coordinate the solution process. When the regional level agent faces intersection level agent coordination request, the genetic algorithm is used an important tool to optimize control parameters to achieve the overall regional traffic flow optimization. The results show that under the same conditions, compared with the traditional timing control, the index performance is greatly improved and this method is effective and feasible.更多还原