Abstract: To solve the problems of insufficient identification conditions of large vehicle barrier in expressway confluence area under automatic driving environment, less research on traffic impact and more qualitative analysis, a car following model with headway and a lane changing model that considers the safety distance were established to form a cellular automaton model, which can indicate the traffic characteristics of the confluence area. The effects of macro parameters (flow of main road and on-ramp and proportion of large vehicles on main road) and microscopic parameters (headway and length of barrier) on the formation of large vehicle barrier were analyzed quantitatively. At the macro level, different combinations of main road and ramp flow were obtained when the large vehicle barrier appears, and it is found that the critical value of the formation of large vehicle barrier is the proportion of large vehicles on the main road of 0.4. The microscopic criterion of large vehicle barrier is that three or more large vehicles run in a queue and the headway between vehicles is no more than 1.5 s. Additionally, the effect of large vehicle barrier was mainly manifested by the decrease of on-ramp flow, the increase of on-ramp delay and the number of traffic conflicts. Results show that it can provide a theoretical basis for the formulation of large vehicles management and control strategy under automatic driving environment.