Over-saturated Intersection Traffic Signal Optimization Based On Dynamic Programming

With the development of economics and motorization, urban road traffic congestion is becoming more and more serious. The traffic flow in rush hours is often saturated or oversaturated. The traffic jams result in the increase of travel delay, reduce the travel time reliability and speed, and severely restrict the urban development. Therefore, it is urgent to ease traffic congestion with the study of saturated and oversaturated traffic flow characteristics, and use traffic management and control measures to improve traffic capacity.The urban traffic state changes dynamically. Video detection and license plate recognition technology can provide the detailed information of real-time traffic path, delay and travel time for traffic control. Traffic signal control optimization takes both the traffic state of previous cycle and the traffic flow of current time into consideration. Based on each stage of the optimal traffic control decision-making for the optimal traffic state, the optimal traffic states provide the basis for traffic control decision in next phase. Therefore, the signal control system needs to be discretized into a multi-step decision-making process in order to adapt to the changing traffic conditions. Based on the analysis of oversaturated traffic flow characteristics and dynamic programming theory, this paper focuses on signal optimization about single intersection and coordinated intersections under saturated traffic condition.Firstly, this paper reviews the development in the oversaturated dynamic traffic control and identifies the existing problems. What’s more, this paper analyzes the oversaturated traffic flow characteristics and the causes and control strategies of oversaturated traffic state. The geometrical distribution of traffic flow discrete model is modified, which is combined with steering ratio and traffic flow dissipation rate to forecast traffic flow rate. The essential of over-saturated traffic flow is high traffic demand and low supply. Therefore, in order to control queue length and balance the demand and supply, it is necessary to increase the traffic supply and control the traffic demand to enlarge the capacity. Secondly, based on the dynamic programming, this paper proposes signal optimization model for oversaturated single intersection to achieve the control objectives: the minimum delay for unsaturated flow and the maximum capacity for saturated and over-saturated flow. Through iteration operations, it decides to retain or change the current phase, and adjusts future signal control with the feedback of the real time control effects. Thirdly, for coordinated control under over-saturated traffic state, the traffic arrival rate is predicted with the steering ratio, traffic flow dissipation rate and traffic flow discretization. This paper proposes the coordinated control plan based on traditional coordinated control parameters and dynamic programming, and proposes the optimization control plan based on traffic arrival rate and dynamic programming. With the program it predicts queuing and delay through traffic arrival rate and the dissipation rate, and adjusts timing plan of next stage with the control effects of real-time feedback.Based on the actual data, the case calculation verifies the advantage of the dynamic programming optimization process of signal control model. Compared with TRANSYT(Traffic Network Study Tool) system optimization timing, it realizes the dynamic programing model and calculates the optimization results with MATLAB. Results show that dynamic programming traffic control planning under over-saturated state effectively reduces the degree of saturation and delays.