Regional Perimeter Traffic Signal Control Based On Explicit Model Predictive Control Method

With the increasing of motor vehicle ownership in China,urban traffic congestion has gradually spread from nodes and arteries throughout the entire network.The topology of regional road networks is complicated in space,and the traffic conditions on them are extremely variable in time.As a result,a control technique that can effectively alleviate traffic congestion on regional level road networks in real time has steadily become a hot research topic in the field of intelligent transportation.Aiming at the real-time and effective control of urban network congestion,this thesis proposes a hierarchical perimeter signal control strategy for urban road network combining macroscopic and microscopic.At the macroscopic level,the regional road network composed of intersections of a certain scale is taken as the research object,and the protected region is guaranteed to be in the desired state of maximum traffic flow by limiting the traffic flow from the external region into the protected region at the boundary of the two regions.At the microscopic level,by timing the traffic signals at the boundary intersections with the objective of minimising the overall vehicle waiting time,a mathematical optimization model of the signal timing at multiple intersections is established and the proportion of released vehicles at the boundary of the two regions derived from the macroscopic control is concretely realised at the microscopic level.The main research contents of this thesis are as follows.First,macroscopic traffic flow dynamics are modeled.An urban road network is divided into two regions: the protected and external ones.By using the macroscopic fundamental diagram as a tool for evaluating the traffic performance in these regions,the flow conservation equations are established to describe the dynamics of the system.Based on conservation equations,a state equation with the number of vehicles transferred between regions as the state variable is further developed to establish a macroscopic perimeter control multi-parameter quadratic optimization model.The objective is to maintain the number of vehicles in the protected region in the expected state.Second,the macroscopic perimeter control strategy is designed.To deal with the low efficiency of solving perimeter control problems,an explicit model prediction control method is first introduced by combining offline and online to establish the explicit model prediction perimeter control framework in both offline and online phases.At the offline stage,the explicit model predictive control method is used to divide the system state space and obtain the corresponding explicit control law,so that the optimal control parameters（the vehicle release ratio from the external region to the protected one）can be calculated in real time.At the online stage,the state deviations caused by the time-varying characteristics are compensated by a state error compensator.The compensated system state is then partitioned and positioned such that the control parameters can be obtained quickly by dynamic iteration.In this way,the high computational complexity of the perimeter control sovled by traditional model predictive control can be overcomed.Also,the traffic signal control in realistic situations conducted by macroscopic perimeter control can be realized.Finally,the microscopic signal control simulation is established.With the objective of minimizing the vehicle waiting timein the intersections at the boundary between the two regions and the green time as the decision variable,a microscopic signal control optimization model is developed.This optimization model is solved by a genetic algorithm through the Tra CI（Traffic Control Interface）secondary development interface provided by SUMO（Simulation of Urban Mobility）microscopic simulation software.The signal timing scheme at the boundary intersections is updated simultaneously.The effectiveness of the proposed perimeter control-guided microscopic signal control strategy and the optimization method are verified by comparing with the traditional fixed timing strategy.