An urban traffic control system integrating neural networks

Abstract

The paper is divided into four parts. First, we briefly review the main generations of UTC systems. Then, the second part is dedicated to a more detailed presentation of the fully adaptive real time UTC system proposed; its architecture and conventional modelling techniques are described. The remaining part concentrates on the very first step of our work, i.e., the modelling and prediction of traffic flow on a single signalized link by neural networks. Finally, some numerical results from simulated data conclude the paper.