Abstract
To improve the safety of ramp vehicles changing lane and to shorten the merging distance, this paper explores the dynamic game interaction properties of vehicles merging and the consistency of vehicles' decision-making behaviors at the macro-microscopic levels. Using the exiD dataset and evolutionary game theory, the merging behavior of ramp vehicles is modeled to explore the effects of different driving states on the evolutionary convergence of strategies. Based on the game cost theory, the lane choice behavior of mainline vehicles is modeled. Validated by SUMO software, the results show that the model in this paper can significantly improve the safety of vehicle merging and reduce the merging distance. The mainline vehicles are more inclined to change lane and cut out in advance when facing the ramp vehicles under the influence of the change of advantage in the subsequent game.