A novel hierarchical perimeter control method for road networks considering boundary congestion in a mixed CAV and HV traffic environment

Under dynamic traffic demand conditions, two issues must be addressed when perimeter control is implemented for congested areas of a road network. The first is to avoid intersection spillback at the boundaries and expansion of the congestion, and the second is to improve the output traffic efficiency of the congested areas to quickly relieve traffic congestion. To address these two issues and solve the traffic congestion problem, in this paper, we adopt a dynamic buffer area to store boundary queuing vehicles and use connected and autonomous vehicle (CAV) technology to improve the traffic flow transmission efficiency of a congested area. First, considering the macro- and micro-level relationships between the macroscopic fundamental diagram (MFD) regions and buffer areas (link- and node-based), a traffic flow transmission model of the network embedded in dynamic buffer areas is built. Second, based on road network state sensing, a dynamic adjustment method of buffer volume is presented to optimize MFD region boundary flows. Third, a hierarchical control method based on model prediction (HCMMP) is proposed for the scenario of a single kernel network. The HCMMP's upper level adopts model predictive control (MPC) to adjust the traffic flow into the dynamic buffer area, and the lower level uses real-time CAV information to optimize the signal timing in the dynamic buffer area. Finally, a complex cellular multiregional road network is selected as a scenario case, and the proposed HCMMP is analysed and compared with no control (NC), proportional integral (PI) control, MPC (both without a dynamic buffer area), PI control with a dynamic buffer area (PIBA) and hierarchical control method based on the PI (HCMPI). The results show that the proposed HCMMP can improve the traffic efficiency, outperforming the other control methods.