Estimation de la demande dynamique

Abstract

Traffic demand evaluation is a really important feature of traffic modelling. So, it's important to know the methods allowing to estimate origin-destination matrices. There are two categories of methods : the ones which estimate static O/D matrices and the ones which estimate dynamic O/D matrices. The methods of static estimation are presented in this master thesis. Then, two methods of dynamic estimation are deepened and compared : on one hand the LSQR method and on the other hand the Kalman filtering method, both are based on least square formulation. We will show that the LSQR method is better than the Kalman filtering method (as well as in flops than at numerical point of view).

Date of Award	2004
Original language	French
Supervisor	Philippe TOINT (Supervisor), Eric Cornelis (Co-Supervisor) & Annick Sartenaer (Jury)