摘要
提出了一种基于M u lti-A gen t的区域交通协调控制系统。系统针对路网中各交叉口交通流相互影响的特点,构造了一种基于分布权值函数的分布式Q学习算法,采用此算法实现了M u lti-A gen t的学习以及协调机制。通过各A gen t间的协调控制来协调相邻交叉口处的控制信号,从而消除路网中的交通拥塞。最后利用微观交通仿真软件Param ics对控制算法进行了仿真研究,仿真结果表明了控制算法的有效性。
In this paper an urban traffic coordination control system based on Multi-Agent is put forward. As for the interactive intersection, an algorithm of distributed Q-learnlng based on distributed weight function is brought forward. The learning and coordinating function of Multi- Agent is realized by the distributed Q-learning. The paper uses the coordination control of each Agent to coordinate the signal of adjacent intersections for eliminating the congestion of traffic network, Finally, the control algorithm is simulated by Paramics, which is a microscopic traffic simulation software. Experimental results indicate that the control algorithm is effective and reliable.
出处
《交通与计算机》
2006年第2期94-98,共5页
Computer and Communications
基金
北京市教委科技发展计划基金资助项目(批准号:KM200410005001)
北京市自然科学基金资助项目(批准号:4042006)
北京工业大学青年科研基金项目(批准号:97002011200501)
