Multi-objective traffic signal control model for traffic management

Abstract

Traffic signals are one of the main traffic management tools used to control traffic flow on the roads and should reflect traffic managers' intentions in different tasks. This paper showed a multi-objective optimization model, and its algorithm was aimed at the intricate structure of traffic control. First, the indexes of queuing lengths, delay times, and stop times were chosen as the evaluation indexes of optimization model. Second, the weight of the optimization indexes was confirmed with a fuzzy analytic process (FAP) according to the traffic managers' strategy. Finally, the multi-objective optimization model was solved with the Q-learning algorithm, and thus the signal control scheme of the intersection was produced in real-time while considering the traffic management strategy. The results of the simulation showed that the method could effectively improve traffic efficiency at the intersection, and at the same time, the intentions of the traffic managers could be fully embodied.