Multi-objective optimal predictive control of signals in urban traffic network

Traffic congestion in urban network has been a serious problem for decades. In this paper, a novel dynamic multi-objective optimization method for designing predictive controls of network signals is proposed. The popular cell transmission model (CTM) is used for traffic prediction. Two network models are considered, i.e., simple network which captures basic macroscopic traffic characteristics and advanced network that further considers vehicle turning and different traveling routes between origins and destinations. A network signal predictive control algorithm is developed for online multi-objective optimization. A variety of objectives are considered such as system throughput, vehicle delay, intersection crossing volume, and spillbacks. The genetic algorithm (GA) is applied to solve the optimization problem. Three example networks with different complexities are studied. It is observed that the optimal traffic performance can be achieved by the dynamic control in different situations. The influence of the objective selection on short-term and long-term network benefits is studied. With the help of parallel computing, the proposed method can be implemented in real time and is promising to improve the performance of real traffic network.