Variable Speed Limit Control Method For Highway Construction Area Based On Reinforcement Learning

In the engineering practice,the reconstruction and expansion project of Expressway usually adopts the construction mode of uninterrupted traffic flow,which will affect the traffic safety and traffic flow operation in the construction area.The traditional speed limit control method is difficult to achieve the pre adaptive change of traffic flow in the construction area,and can not prevent congestion in advance.It has a certain lag,which affects the actual traffic efficiency of the road in the construction area.Therefore,this paper proposes a variable speed limit control method based on reinforcement learning to eliminate or alleviate traffic congestion in the construction area.Based on the definition,layout and construction form of expressway construction area,this paper studies the vehicle operation characteristics of expressway construction area.According to the collected traffic flow data,the data-driven method is applied to analyze the speed change characteristics,flow and density relationship changes in the construction area;according to the analysis results,the formation mechanism of the bottleneck section in the expressway construction area is explored to provide the basis for the establishment of the traffic flow model and the variable speed limit control method in the construction area.Based on the cellular transport modeling theory of traffic flow,considering the traffic flow phenomenon of "capacity mutation" and the influence mechanism of variable speed limit control,the basic diagram of traffic flow is improved.The traffic flow model of expressway construction area based on cell transmission model is established,and the parameters of the model are calibrated and tested to provide analysis and evaluation basis for the establishment of a new variable speed limit control method in construction area.Based on the analysis of variable speed control mechanism and reinforcement learning method,combined with the traffic flow characteristics of expressway construction area,a new variable speed control method is established by using Q-learning algorithm.Taking the set of variable speed limit values as the action space,the set of upstream traffic demand flow,variable speed limit control section and traffic density of bottleneck section as the state space,the number of vehicles passing through the construction area within the speed limit period as the reward function,and the ε-Grey strategy as the action selection strategy,The variable speed limit control method of expressway construction area based on reinforcement learning is formed,and the corresponding implementation process is designed.The new method not only considers the upstream traffic demand of the construction area,but also integrates the key density values of the construction speed limit control section and the bottleneck section.Combined with the Q-learning iterative mechanism,the speed limit optimization value can effectively improve the traffic efficiency of vehicles.Taking the reconstruction and expansion of Beijing Shanghai Expressway as the application background,MATLAB software is applied to develop the corresponding simulation program,which describes the traffic flow cell transmission model established in the construction area and the variable speed control method based on Q-Learning.Simulation experiments are carried out on the traditional variable speed control method based on feedback control and the variable speed control method proposed in this paper.The simulation results show that under the same traffic demand,the new variable speed limit control method can improve the traffic efficiency of the construction area by 24.5% and 9.9% respectively,and reduce the congestion time by 42.1% and 11.5% compared with other methods,which has obvious optimization effect.This result proves that the variable speed limit control method proposed in this paper is effective in improving the traffic operation and traffic efficiency in the construction area.