A signal split optimization approach based on model predictive control for large-scale urban traffic networks

It is generally recognized that Model Predictive Control (MPC) has many advantages in signal control of urban traffic networks. However, the computational complexity grows exponentially with the increase in network scale and predictive time horizon. In order to overcome this drawback, a signal split optimization approach is proposed in this paper, in which a large-scale traffic network was decomposed into a set of subnetworks. Based on the store-and-forward modeling paradigm, the optimization framework of each subnetwork is developed firstly. Then, Lagrange multipliers are employed to deal with interconnecting constraints among subnetworks, and the dual optimization problem corresponding to the whole network is constructed. Moreover, the dual optimization problem is optimized under a two-level optimization structure by interaction prediction approach. In the end, simulation experiments are given to illustrate the effectiveness of the proposed approach.