PREDICTIVE CONTROL FOR URBAN TRAFFIC NETWORKS: INITIAL EVALUATION

The paper proposes a control method for urban traffic networks (UTN) through a modified approach of a traffic-responsive urban control (TUC) strategy. TUC uses a store-and-forward traffic flow model and the linear-quadratic theory to efficiently compute network-wide control signals, at the cost of sub-optimality for not considering constrains explicitly. In order to attain an improved controller a model predictive control (MPC) approach is designed. Simulated results of a sample network are shown for fixed time control, TUC, and TUC-MPC strategies. The results are also compared through a mean queue criterion. Finally, conclusions are drawn and future work is outlined.