Modelling And Simulation Of Cooperative Lane-Changing At Intersection Based On Indirect Reciprocity In Internet Of Vehicles Environment

Mandatory lane changing before the intersection is a kind of traffic reduction behavior.The problem stems from the contradiction between the temporal and spatially urgent need to change lanes by lane-changing vehicles and the uncooperative willingness of the lagging vehicles for their own traffic efficiency.Therefore,how to promote the cooperative lane change of drivers is the key to solving this problem.In view of this,this thesis conducts the following research work for the cooperative lane change at the intersection:Firstly,the data is pre-processed in three aspects: data cleaning,data correction and data completion.An intersection lane change trajectory identification method based on lane change rules and clustering is proposed to accurately extract and identify the intersection near-end lane change trajectories that have the most significant impact on intersection traffic conditions.On this basis,key micro characteristics such as intersection lane change time,lane change spatial location and lane change speed are analysed.Secondly,the basic theory of indirect reciprocity and its application areas and scenarios are sorted out.The multi-scenario spatio-temporal portrait and game model of intersection lane change are established.By analysing the characteristics of four typical spatio-temporal scenarios of intersection near-end lane change,a chicken game model of the intersection lane change problem is established and the payoff matrix of its singleround game is analysed.Drawing on the solution of the chicken game,an evolutionary game model of intersection cooperative lane change considering indirect reciprocity in the connected vehicle environment is established and the payoff matrix of the multi-round game is analysed.Furthermore,a cooperative lane-changing strategy at intersections based on indirect reciprocity is proposed.Starting from the image sub-mechanism,it designs the addition and subtraction rules and improved OR strategies that reflect the core of indirect reciprocity.A cooperative lane-changing strategy and reinforcement learning algorithm at intersections based on indirect reciprocity are proposed.The algorithm improves the Bellman equation in the traditional Q-learning algorithm,and reflects the bidirectional iteration of the learning matrix of the lane-changing vehicle and the lagging vehicle in the reinforcement learning process.Based on the intersection cooperative lane change strategy and the reinforcement learning algorithm,a model of intersection cooperative lane change and spatio-temporal evolution is developed,including lane-changing carfollowing modelling,cooperative or competitive lane change modelling,turning modelling and signal control modelling.Finally,a Net Logo multi-agents simulation platform is built and intersection cooperative lane change simulation experiments are designed,with covariates designed and evaluation indicators selected based on the previous study.Based on the validation of the simulation model,control experiments and pre-experiments,99 sets of simulation experiments are carried out for different combinations of variables.The results of the simulation experiments are analysed in terms of driver benefits,intersection passing efficiency and a combination of both.The results show that the cooperative lane change mechanism has a positive effect on increasing the probability of driver cooperation,improving lane change efficiency and transit efficiency,and improving the stability of transit times and delays,while there are also certain inapplicable situations.