Abstract This study proposes a hierarchical control framework to maximize the throughput of a road network driven by travel demand with uncertainties. In the upper level, a perimeter controller regulates the traffic influx into the core road network. The upper level uses a reinforcement learning algorithm that learns and responds to the traffic dynamics in the core road network without the need for an underlying system model and macroscopic fundamental diagram. The lower level is a local signal control system that regulates the spatial distribution of traffic flow within the core network. The results show that the hierarchical control framework can improve road network throughput by coordinating control actions conducted at the two levels. The improvement in system-wide performance is validated by a range of performance metrics and macroscopic flow-accumulation patterns realized under different control settings. The study contributes to the management of urban road networks with advanced computing technologies.
Highlights A hierarchical control framework maximizing stochastic network throughput. A reinforcement learning algorithm which learns and responds to traffic dynamics. Two case studies demonstrating the efficiency and practicality. Improvement validated by a range of performance metrics.
Hierarchical control for stochastic network traffic with reinforcement learning
Transportation Research Part B: Methodological ; 167 ; 196-216
2022-12-01
21 pages
Article (Journal)
Electronic Resource
English