Vehicle-cloud cooperative hierarchical optimization control for mixed traffic in the bottleneck area of the tunnel entrance

Physical spatial factors and environmental conditions, such as lane reduction and variations between internal and external velocity limitations, can lead to traffic congestion at the tunnel entrance. Moreover, the varying levels of intelligence between Connected and Autonomous Vehicle (CAV) and Human-driven Vehicle (HV), combined with the complex interaction between the cyber and physical dynamic of mixed traffic, may further exacerbate congestion in this area. Therefore, designing a control strategy from the Cyber-Physical System (CPS) perspective is essential to alleviating traffic congestion and enhancing traffic efficiency at the bottleneck area of the tunnel entrance. Leveraging the characteristics of CPS, we propose a vehicle-cloud cooperative framework that serves as the foundation for the vehicle-cloud cooperative hierarchical optimization control (VCCHOC) strategy. The framework facilitates seamless coordination between vehicles and cloud infrastructure, enabling hierarchical optimization. Specifically, the upper-level control employs a lane-level Mixed-Integer Linear Programming (MILP) model, which balances the differences in penetration rate (PR) and vehicle densities across lanes. At the lower level, a fixed-time longitudinal control algorithm is designed for CAV, ensuring stable spacing and enabling real-time updates of acceleration and velocity. Additionally, Model Predictive Control (MPC) tracks reference trajectory in real-time during lane-changing. Simulation results demonstrate that the proposed VCCHOC significantly enhances traffic efficiency, reducing traffic delays and fuel consumption as CAV PR increases.