Research On Connected Automated Vehicle Speed Control At A Signalized Intersection

Traffic stream will be interupted by red lights at signalized intersections so that will cause “stop-and-go” traffic,and thus signal intersections rise the interests to researchers.Connected automated vehicle is an inexorable trend of the future traffic.Connected automated vehicle speed control method is one of the effecive way to solve traffic oscillation.Therefore,this study proposes centralized and decentralized based connected automated vehicle speed control systems at a signalized intersection,and investigates the theory upper bound of these two control systems in pure connected automated vehicles environment.Then,this study focuses on the research under mixed traffic,which contains car following behaviors of human driven vehicles and speed control methods of connected automated vehicles.First,this study designs a centralized based connected automated vehicle speed control system at a signalized intersection under pure connected automated vehicles environment.The joint objective is traffic mobility,fuel consumption and traffic safety.This study deletes the assumptions with fixed arrival and departure states and thus is more applicable in reality.Additionally,time discretization and piece-wise trajecotry approaximation are used to modify the complex non-linear original problem.Divided rectangle algorithm with smoothed shooting heuristic based initial solutions and parallel computing approach are applied to solve the modified model.The results show that the proposed system can fast the soltuion efficiency wi th an effective and stable optimal solution.Second,this study designs a decentralized based connected automated vehicle speed control system at a signalized intersection under pure connected automated vehicles environment.The same as the centralized control system,the objective of the decentralized control system is a joint objective of traffic mobility,fuel consumption and traffic safety.Time discretization,piece-wise trajectory approaximation,Divided rectangle algorithm with smoothed shooting heuristic based initial solutions and parallel computing approach are also applied in the decentralized control system.Different with the centralized control system,the decentralized control system considers each vehicle as an intelligent agent instead of us ing a central controller.Thus,the decentralized control system separates the original problem into several sub-problems.The results show that the decentralized control system yields much better performance on computational efficiency without a significa nt loss on sytem performance.Thirdly,this study invsetigates car following behaviors of human driven vehicles at a signalized intersection under mixed traffic.Considering obvious traffic oscillations caused at signalized intersections,this study analys es the dynamics of traffic oscillation with high resolution vehicle trajectory data.A time gap function is proposed to revise traditional car following models according to the empirical analysis results.Models calibration and validation are applied on traditional and revised models.The results show that the revised mdoels yield better performance on prediction accuracy than the traditional models.Then,a more suitable revised car following model is chosen to describe car following behaviors of human driven vehicles,and thus provides a more accurate recognization technology in connected automated vehicle speed control system at a signalized intersection under mixed traffic.Finally,based on the centralized and decentralized control systems,this study considers the effects of human driven vehicles and designs connected automated vehicle speed control system at a signalized intersection under mixed traffic.The results of connected automated vehicles market penetration rates analysis demonstrate the effectiveness and advancement of the centralized and decentralized control systems under different mixed traffic conditions.Based on connected automated vehicle technologies,this study proposes centralized and decentralied based connected automated vehicle speed control systems at a signalized intesection under pure connected automated vehicle.The results show that the decentralized system yield more computation efficiency without a significan loss in system optimality compared with the centralized system.Then,this study investigates car following behaviors of human driven vehicles and extends the connected automated vehicle speed control models at a signalized intersection under mixed traffic.The analyses of connedted automated vehicle market penetration show that,at a low level of connedted automated vehicle market penetration rate,the decentralized control system is more applicable considering the cost of systems and the computation efficiency;with the connedted automated vehicle market penetration rate increases,the centralized system is suggested to be used if the system optimality is the priority,or the decentralized system will be used if considering less cost of system.In summary,the proposed systems and models can provide theoretical support and implementation guidance for eliminating traffic congestion and extra energy waste caused by traffic oscillation at signalized intersections in the future.