Study On Optimization Of Urban Traffic Signal Herarchical Control Based On Fuel Cost

Vehicles have greatly improved our life and society since they were invented; however, they have also consumed massive oil resources and resulted in serious traffic problems and environment pollution. Thus how to save energy has been an important issue of urban traffic. Against this background, this dissertation deals with urban traffic control from the point of view of fuel cost by using hierarchical control, agent based control, fuzzy control, genetic algorithm and traffic simulation software. All in all, this dissertation provides a possible way to study energy saving in urban traffic.First of all, this dissertation summarizes current status of energy saving of urban traffic and countermeasures of energy saving of urban traffic. And then a brief review on urban traffic control system is given, which focuses on some classical systems. Then the contents and structure of this dissertation is put forward. This dissertation studies signal control of urban traffic from three layers:single intersection control layer, area coordination layer and central dispatching layer.Secondly, as for single intersection control layer, three single intersection signal control algorithms:pretimed control algorithm, actuated control algorithm and adaptive control algorithm are introduced in this dissertation. These three control algorithms have three corresponding control agents:pretimed control agent, actuated control agent and adaptive control agent. Then the effects of the three control agents under five traffic flow levels are evaluated, which also lays the foundations for area coordinating and central dispatching.Thirdly, as for area coordination layer, fuzzy control is introduced to optimize cycle of pretimed control agent and maximum phase green time of actuated control agent and adaptive control agent according to traffic status from the point of view of fuel cost. Through simulation results, the effects of area coordination are proved.At last, as for central dispatching layer this dissertation mainly studies control agents' off-line dispatching learning optimization algorithm and on-line dispatching learning optimization algorithm. Genetic algorithm is applied in off-line dispatching learning optimization algorithm by using historical traffic flow data. And on-line dispatching learning optimization algorithm selects control agents through setting threshold of fuel cost for special intersection in special period. Then control agent dispatching scheme was updated also. Performance of off-line dispatching learning optimization algorithm and on-line dispatching learning optimization algorithm are described respectively by simulation results.