Research On Agglomeration Control Model Of Connected Autonomous Vehicles Based On Multi-Agent System

With the development of networked and intelligent vehicles,more and more Connected and Autonomous Vehicles(CAV)will rush into the road traffic,forming a mixed traffic flow with traditional human-pilot vehicles(HPV).It is a hot research topic to design a reasonable control model of CAV for improving the traffic efficiency of mixed traffic flow.The main problem is the lack of unified CAV control theory and guiding ideology,single model and strategy and not close relationship with the future practical application.Based on “The National Key Research and Development Program of China(Grant No.2018YFB1600500)”,this paper takes the formation and maintenance of CAV platoon in mixed traffic flow as the starting point.By introducing MultiAgent System(MAS)theory and agglomeration idea,the framework of CAV agglomeration control in mixed traffic environment is constructed.Based on the control framework,this paper introduces designs the Agglomeration Control Model of Connected and Autonomous vehicle based on Multi-Agent System(ACMCAV-MAS)and the Control Model of Agglomeration Passing through Intersection based on Multi-Agent System(CMAPI-MAS).The main research work is as follows:(1)The framework of the CAV agglomeration control based on MAS is constructed.Specifically,through analyzing the characteristics of mixed traffic environment,various elements in the traffic flow are analyzed.The upper management Agent and the bottom vehicle Agent are designed for mixed traffic agglomeration.On this basis,the CAV agglomeration control framework based on MAS is constructed.The agglomeration control framework can obtain and process the traffic information in the mixed traffic environment in time,and convey the matching control information to the vehicle Agents in different driving states based on the communication system and agglomeration algorithm.(2)The model of CAV agglomeration control based on MAS is designed.The control model is oriented to mixed road segment scenarios,which mainly includes two agglomeration strategies and corresponding vehicle Agent horizontal and vertical control algorithm.The two agglomeration strategies are called Platoon Level Agglomeration(PLA)and Lane Level Agglomeration(LLA).Through simulation experiments,it is found that both PLA and LLA strategies can improve the traffic efficiency within a certain density and CAV-Agent permeability range,and the best benefit can be achieved under the CAV-Agent permeability of 60%.Meanwhile,the maximum proportion of risk reduction of the two agglomeration strategies can reach more than 20%.(3)The control model of agglomeration passing through intersection based on MAS is designed.The control model solves the constraints of vehicle Agent in the intersection traffic based on the improved genetic algorithm,and then optimizes the signal timing and speed trajectory of the mixed traffic flow in the intersection.The simulation results show that the control model can reduce the average delay and fuel consumption of vehicle Agent by 20.6% and 22% at a singlepoint intersection,and the average travel time of vehicle Agent within the region can be reduced by 34.8% after the control model is introduced into the urban area.In conclusion,ACMCAV-MAS and CMAPI-MAS agglomeration control models are designed based on the constructed agglomeration framework in this paper,which has important theoretical significance and value for enhancing the control ability of mixed traffic flow.