An Approach of Model Predictive Control for Urban Transportation Network

This paper focuses on the traffic signal control of urban transportation network, and presents an on-line optimal strategy by means of Model Predictive Control (MPC). To achieve this, firstly, the transportation network is modeled into a linear form with time-variant parameters. Then, for evaluating the system performance, this paper compares the transportation system with thermodynamic system, and introduces the entropy notion to measure the system disorder. Furthermore, to guarantee the robust stability of controller, the dissipativity theory is applied to address necessary conditions. By combining all these efforts into the framework of MPC, a traffic signal control strategy is presented to minimize the system disorder in finite horizons of time with respect to the constraints on both state and control. Finally, a network including four intersections is taken as an example to illustrate the results.