To seek the best traffic management and control method to alleviate traffic congestion under sudden traffic accidents on highways, the paper proposed a method for toll station entrance control aimed at accident bottlenecks. Firstly, an modified METANET model was used to predict traffic density and flow of highway network. Taking the storage space of ramp and toll plaza, the traffic capacity of road section as constraints, and aiming to minimize the travel time on the main line and the waiting time at toll station, an optimal control rate model for toll station entrance was established. The particle swarm optimization algorithm (PSOA) was used to solve the optimal control rate at the toll station entrance, which was substituted into the established toll station channel control model to obtain the control scheme for each toll station channel. Finally, the VISSIM simulation software was applied for secondary development, and case simulations were conducted to verify the proposed method can reduce vehicle delays and congestion duration time. After accident cleaning, the vehicles could quickly dissipate, which reached a ideal effect.