A traffic control strategy for freeway merging zones cooperating safety and efficiency in the intelligent connected environment of mixed vehicles

The manner and intensity of vehicle interactions in a mixed-vehicle traffic flow differ from those in a typical traffic flow. This difference leads to greater potential conflicts and decreased efficiency in freeway merging zones, which involve a large amount of vehicle crossing behaviour. To avoid the deterioration of traffic status, cooperative control of safety and efficiency for mixed-vehicle traffic flow using connected and automated vehicles (CAVs) in freeway merging zones is proposed. First, a multi-objective nonlinear mixed-integer program model for cooperative safety and efficiency is presented at the vehicle level to optimize CAV’s behavioural decisions using historical predicted data. Second, a Transformer neural network is adopted to forecast the traffic state under different control weights, accounting for the dynamic characteristics of the traffic system. An adaptive weighting model is constructed to choose the optimal solution from the Pareto frontier derived from the multi-objective problem. To ensure the feasibility of vehicle-level decisions and to facilitate system-level optimization, CAVs are capable of sharing and coordinating their behaviour decisions through iterations. A typical scenario involving a two-lane freeway merging area is analysed, and the results show that the cooperative control strategy can effectively optimize the traffic state. Even at 20% CAV penetration rates, this strategy reduces total parking delays by 48.7% and time-integrated time-to-collision (TIT) by 72.2%.