On the use of neural network techniques for traffic flow modelling
LEDOUX ; BOILLOT ; SELLAM ; GALLINARI
Type de document
COMMUNICATION AVEC ACTES INTERNATIONAL (ACTI)
Langue
anglais
Auteur
LEDOUX ; BOILLOT ; SELLAM ; GALLINARI
Résumé / Abstract
This paper presents the results of our investigations on the potentials of neural networks to model traffic flow. Neural network architectures have drawn considerable attention in recent years because they are non linear data driven models, ie they are able to infer function from a set of numerical data. As far as traffic modelling is concerned, they exhibit interesting features particularly suited for this application. Neural networks are introduced here to model the queue length dynamics, on a signalized urban link.