Research On Adaptive Traffic Signal Control With Network-wide Coordination

Urban traffic congestion is becoming an increasingly serious problem.It is an important way to alleviate the congestion by controlling the traffic in urban road network through traffic signal control.The practical application of traffic signal control system mostly use off-line control method,such control method is not flexible enough.Adaptive traffic signal control can develop more efficient traffic signal control scheme according to the real-time traffic flow variation in the road network,and guide the vehicles in the urban road network more accurately.It can reduce the traffic congestion and reduce vehicles’ delay time.Due to the lack of effective network-wide coordination between the traffic signal controllers,recent method of adaptive traffic signal control cannot achieve ideal global optimization control effect.Therefore,in order to reach the global optimization of adaptive traffic signal control in urban road network,we propose an adaptive traffic signal control method with network-wide coordination.The adaptive traffic signal control method with network-wide coordination predicts the vehicles’ global delay time of each road link through cooperates between traffic signal controllers in the whole road network.So that the traffic signal controller of each intersection can perceive vehicles’ global state,obtain a global optimization control objectives,and achieve the global control optimization in the entire road network.Firstly,a vehicles’ global delay time prediction model based on network-wide coordination is established.The vehicles’ local delay time of each road link is predicted by real-time monitoring data.The prediction of vehicles’ global delay time of each road link is completed by iteration update between the traffic signal controllers at each intersection.Then,we proposes a heuristic adaptive traffic signal control method with network-wide coordination and a reinforcement learning adaptive traffic signal control method with network-wide coordination.Using the predicted vehicles’ global delay time to calculate the heuristic control optimization target and the immediate return value of reinforcement learning,to achieve the network-wide traffic signal optimization control.The adaptive traffic signal control method with network-wide coordination is evaluated on the real data set of the traffic flow in the Yunyan district of Guiyang city.The results of the experiment show that the improved traffic signal control method can reduce the total vehicles’ delay time and the average queue length of the vehicles in the road network.This method can effectively alleviate the traffic congestion in the urban road network.Especially in the case of high traffic pressure,the adaptive traffic signal control method with network-wide coordination can achieve a better global control effect.