基于多智能体MAS的智能交通控制系统的研究

发布时间：2018-02-21 19:20

本文关键词： 多智能体智能交通控制智能体通讯模型仿真　出处：《长沙理工大学》2007年硕士论文　论文类型：学位论文

【摘要】： 交通问题是全球普遍关注的重要问题，传统的基于精确数学模型的交通控制方法不仅复杂而且在实际应用中往往达不到预期的控制效果。而多智能体(Multi-Agent)技术是人工智能领域中的一个研究热点，单个Agent能够自主地、主动地对对象进行控制，并且Agent之间可以进行协作从而使整个系统达到更大的智能性。本文在论述Agent理论和多智能主体系统(Multi-Agent System，MAS)技术的基础上，针对目前智能交通控制系统的现状及交通系统存在的非线性、不确定性、时变性和不完全性等问题，将Agent技术应用于交通控制系统，提出了一种基于多智能体的智能交通控制系统结构。该系统在每个交叉口均设置一个交叉口Agent，，由中央Agent监督与管理，并探讨了各个交通元素Agent的功能以及它们的协调合作关系。控制系统采用部分全局的方法规划整个控制过程。交叉口Agent内部控制策略采用广义的知识模型实现，采用预测、协调、感应调整三段式的控制流程，用动态规划的学习方法制定初始控制方案，采用博弈论实现与相邻交叉口Agent及与中央Agent的协调，采用点对点直接通讯方式完成与相邻交叉口Agent及与中央Agent的通讯。同时运用Delph7.0对基于多智能主体的智能交通控制系统进行了初步仿真。仿真结果表明：基于多智能主体的智能交通控制系统拥有一种全新的智能交通控制结构，它能够适应交通系统的复杂性和随机性，较好地克服传统交通控制系统所存在的缺陷。
[Abstract]:The traffic problem is an important problem all over the world, the traditional traffic control methods based on precise mathematics models is not only complex but also in practical applications are often not up to the expected control effect. The multi-agent system (Multi-Agent) technology is a hot research topic in the field of artificial intelligence, a single Agent can independently control active objects, and can carry out collaboration to make the whole system to achieve greater intelligence Agent.
This paper discusses the Agent theory and multi agent system (Multi-Agent, System, MAS) on the basis of technology, aiming at the nonlinear, existing traffic system and intelligent traffic control system uncertainty, time-varying and uncompleteness, applies Agent technology to traffic control system, put forward a kind of intelligent traffic control the system structure based on Multi-Agent system. The system sets an intersection Agent at each intersection, by the central Agent supervision and management, all traffic elements Agent functions and their coordination and cooperation is discussed. Method of control system using partial global planning to control the whole process. The intersection of the Agent internal control strategy is adopted the realization of the generalized knowledge model, prediction, coordination, induction adjustment control three stage process, making the initial control scheme by learning dynamic programming method, using game On the realization and adjacent intersection of Agent and coordination with the central Agent, using point-to-point direct communication mode and adjacent intersection with central Agent and Agent communication. At the same time Delph7.0 a preliminary simulation of the intelligent traffic control system based on multi-agent system application. The simulation results show that the intelligent traffic control system of multi agent have a new intelligent traffic control based on the structure, it can adapt to the complexity of traffic system and randomness, overcome the drawbacks of traditional traffic control system.

【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2007
【分类号】：U495

【参考文献】