As current urban area traffic condition is quite complex,which is difficult to be described accurately with traditional mathematic model,a distributed traffic coordinated control model based on fuzzy Q-Learning algorithm is proposed in this paper. The model regards the traffic control system at every junction as an individual junction Agent. The Agent uses fuzzy Q-Learning algorithm to determine the current phase time based on the local phase traffic flow information,next phase traffic flow information and downstream road traffic flow information which is predicted. At last,utilize VISSIM4. 2 simulation platform to simulate a simple traffic network and verify the algo-rithm. The results of simulation show that the model provided greatly improves the whole efficiency of road network traffic control over fixed time traffic control model.% 针对当前城市区域交通状况复杂,难以用传统数学模型对其进行精确描述的特点,提出了一种基于模糊Q学习的分布式交通协调控制模型.该模型将每一个路口的交通控制系统看作一个独立的路口Agent,每一个路口Agent根据预测的当前相位和下一相位的交通流信息以及下游路段的交通流信息采用模糊Q学习算法决策出当前相位的绿灯时间,最后利用VISSIM4.2交通仿真平台进行了简单网络仿真实验,验证了该算法的可行性.仿真结果表明,该模型运用于交通控制中相比于定时控制能有效提高路网控制效率.
展开▼
