A neural network-based stochastic approximation approach to special events traffic signal timing control

A sudden traffic surge immediately after special events (e.g., conventions, hockey games, concerts, etc.) can create substantial traffic congestion in the area where the events are held. It is desired to implement a short-term traffic signal timing adjustment for the high volume traffic movements associated with special events so that progression is as efficient as possible. This paper presents a case study of special events traffic signal timing control for the City of Duluth Entertainment and Convention Center (DECC). Our optimization approach is based on neural networks (NNs) with the weight estimation via the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm. Using the traffic data collected, the NN-based SPSA optimization method is applied to make signal timing adjustments. A tolerance index is chosen as our measure-of-effectiveness (MOE). The NN weights are determined by use of the SPSA parallel estimation algorithm that minimizes the MOE criterion at the selected intersections following DECC events. The performance evaluations, based on different MOEs, using the existing signal timing and the one generated by the SPSA algorithm are investigated. The results show the potential of the proposed optimization method.