【作者】 赵建玉；

【导师】 贾磊；

【作者基本信息】 山东大学 ， 控制理论与控制工程， 2006， 博士

【摘要】 本文以城市交通为研究对象，包含两个部分的研究内容：一是智能计算在城市交通中的应用研究，内容涉及基于智能计算的城市交通流控制与短期城市交通流预测模型；二是交通流建模研究，内容涉及基于气体动力学的交通流模型研究。 交通流运动具有随机性、离散性及较强的非线性。模糊控制、神经网络等现代人工智能技术，可以使人们在不表达交通系统精确模型的情况下，通过对问题的归纳与并行处理，为交通控制和交通管理提供依据。与此同时，由于计算机科学，尤其是计算机模拟技术的发展，可以使人们克服复杂数学推导的困难，从而使传统的数学与物理方法建立可靠交通流模型的研究与应用成为可能。 本论文在广泛查阅各类文献的基础上，分别从上述两方面开展工作，符合当前交通流理论的研究方向，是交通流理论当前研究的热点与难点。论文的主要内容与创新点包括： 1、以城市中心地带主干道典型的交叉路口为研究对象，提出了一种基于模糊逻辑的城市交叉口信号灯控制器，该控制器的设计思想基于误差闭环控制理论，在交通流中引入误差变化率概念，从而更为真实的反应了交警的人工智能活动。提出的控制策略具有很强的鲁棒性与抗系统时变性特点，可有效的提高交叉口相关区域内的道路使用率。 2、将基于误差闭环控制的城市道路交叉口模糊控制器以神经网络的方法实现。设计研究的模糊神经网络控制器把神经网络的学习和计算功能带到模糊系统中，也可把模糊系统的思维规则和推理嵌入到神经网络中，引入模拟退火算法训练网络，使模糊控制器能够自行调整隶属度函数，弥补了模糊控制的不足，同时提高了网络的性能。 3、进行了城市主干道多交叉口神经模糊控制器的研究与仿真，提出了一种基于模糊神经网络的新型城市道路多路口协调控制系统。该控制系统提取了“绿波带”的知识方法，并用模糊神经网络的方法实现，可以在设定的时段内自动对交通流各类信息进行汇总与统计，实时输出优化的城市干线交通控制协调控制配时方案。仿真研究表明，该控制系统可以充分利用交叉口空闲时空，使交通流以理想的饱和度值整体通过交叉路口，说明了模糊神经网络方法是解决城市交通大更多还原

【Abstract】 Focusing on urban traffic, the paper includes two parts of research contents. The first is application research of intelligent computing on urban traffic, which deals with the control of urban traffic flow and the forecasting model of urban short-term traffic flow; The second is the research of traffic flow modeling, which deals with the traffic modeling based on gas -kinetic.Traffic flow has characteristic of randomicity, dispersion and stronger nonlinearity. Thanks to modern artificial intelligence technology such as fuzzy control, neural network, etc, now we can deal with traffic system via concluding and parallelly processing without expressing its exact model. At the same time , as a result of computer science, especially the development of computer simulation, the difficulty of complex maths derivation is overcame, makes it is possible to establish credible traffic flow model with traditional maths and physics methods.Referring to correlative literatures widely, two aspects of work above mentioned is carried out in the paper, which accords with current research direction of traffic flow theory and is the hotspot and the difficult spot in the field. The main contents and innovation points include:1, Aiming at urban arterial roads typical intersection, an intersection traffic signal controller based on fuzzy logic is presented. The design thought of this controller is based on error closed-loop control theory, the concept of error variety rate being introduced here. Then the artificial intelligence activities of the traffic police is reflected more factually. The presented control strategy has very strong robustness and anti-time-variety, can effectively improve the using rate of roads around the intersection.2, The fuzzy logic controller of urban intersection based on error close-loop control is carried out by the method of artificial neural networks. A Fuzzy Neural Network (FNN) controller is designed and researched in the paper, bringing the learning and calculating function into fuzzy system, as well as embedding the thinking and reasoning method of fuzzy system into neural network. The simulated annealing更多还原