Parallel management for traffic signal control

With the rapid growth of the number of urban vehicles, it will be not advisable to alleviate traffic congestion by changing the traffic facilities only. And the traditional control strategies for single intersection or regional multiple intersections have been confirmed to have some effect in the past few decades, but still need to be improved. Based on ACP (Artificial societies, Computational experiments, Parallel execution) idea, we firstly proposed the concept of “event agent” in this paper, which refers to the ratings that traffic states give corresponding timing plans. Based on event agent, we used computational methods to establish a Parallel transportation Management Systems (PtMS), which was a self-completing system. In the system plenty of artificial events were generated, and some of them can not only simulate the actual traffic events, but also be substitutes for the actual events. Then through the parallel execution between actual and artificial events, the system recommends the most suitable timing plans to the current traffic state. Different from traditional control strategies, event agent based PtMS takes results as an orientation according to the idea of data-driven, which is more adaptive to the characteristics of transportation systems. For ensuring the validity and accuracy of experiments, our related data are all based on the famous traffic micro-simulation software Paramics. Furthermore, we compared our method with the classic Webster method, and experiments achieved good results.