Traffic Signal Optimal Control Approach Based On Explicit Model Predictive Control

Urban traffic congestion has become more and more severe,owing to the rapid growth of vehicles.Traffic signal control system is a complex large system with the features of multiple input and output,time varying and so on.The model predictive control approach has many advantages in dealing with the control problem of this system and has been applied to control urban traffic network.However,as the traditional model predictive control needs to be repeated online optimization,the computational complexity of traffic signal control optimization problem increases rapidly with the increase of the scale of the problem,excessive computational costs limits its use in scenarios where realtime requirements are relatively high.Therefore,the main work of this paper as follows:Considering the shortcomings of the model predictive control,in this paper,the Explicit Model Predictive Control(EMPC)is firstly introduced to design the transportation signal controller,which is based on the store-and-forward model for the traffic flow process.In order to avoid the repeated online rolling optimization process of the traditional MPC,the EMPC approach is divided into two parts: offline calculation and online search,to improve the real-time operability of the forecast control in the traffic system.The simulation experiments show that the signal control method based on the EMPC approach is superior to the fixed time signal control.At the same time,compared with the traditional MPC,the EMPC approach is a breakthrough method to reduce the on-line computational complexity,and to increase the applicability of the EMPC in reallife traffic networks.With the expansion of the traffic network,the number of state partitions also increases rapidly.The computational complexity of online search has become the biggest challenge of applying EMPC in traffic systems.Therefore,based on the theory of the reachable region,a new approach called reachable partitions algorithm is designed to address the point location problem of EMPC for traffic signal control in this paper.The proposed algorithm limits the online search range of EMPC to the area where the next state point can be transferred,which greatly reduces the complexity of online computation and improves the control effect of EMPC for traffic system.Through the simulation experiments,the reachable partitions algorithm is compared with the direct search method,and the validity of the reachable partition algorithm is verified.