Multi-level Model-based Control for Large Scale Urban Traffic Networks

In this paper, a multi-level model-based control strategy is applied to control large scale urban traffic networks. Structure of the strategy contains two levels, i.e., a coordinator in the upper level and model predictive controllers in the lower level. The coordinator aims to keep the distribution of vehicles in the network homogeneous, and the two-level structure makes the computational burden acceptable. The simulation shows the performance of multilevel model-based control is close to that of centralized model predictive control and better than decentralized control. Meanwhile, the computational complexity is greatly reduced comparing with centralized control.