GameTraffic:基于交通调度历史数据挖掘的路口最优调度及道路改造预测

GameTraffic:Optimal-traffic-scheduling and road-reconstruction based on the mining of historical traffic-scheduling data

最大化车流量和最小化平均等待时间是交通路口调度的目标.交通调度中各路口与其它路口间存在博弈关系,相邻路口间为使其自身利益最大化而存在策略间的相互协调.我们基于博弈论对交通系统进行建模,基于博弈均衡的增强学习算法对交通调度历史数据进行挖掘分析,学习得到交通路口的最优调度策略,并进行道路改造预测.展示了交通路口最优调度及道路改造预测系统GameTraffic,旨在为智能交通管理及决策提供一种科学的依据.

The target of traffic intersection scheduling is to maximize the flow rates and minimize the average waiting time of all concerned vehicles.Game relationships exist among intersections in the traffic scheduling,and there is mutual coordination among strategies for the maximal profits of neighboring intersections.We model the traffic system based on the game theory,and then learn the optimal scheduling strategies by mining the historical traffic-scheduling data based on the reinforcement learning algorithm.In this paper,we demonstrate the system for traffic-scheduling and road-reconstruction,called GameTraffic,in order to provide a scientific basis for intelligent traffic management and corresponding decision support.