Algunas herramientas estadísticas y matemáticas para la modelización del tráfico

Résumé

In this thesis, the following original statistical and mathematical models are presented: - Two static traffic assignment model with heterogeneous users that permits obtaining the link and path flows from the flow on the origin-destination pairs. These models consider different users classes by their punctuality and overtaking desire, respectively. - A Bayesian origin-destination matrix estimation model based on hierarchical optimization. The estimates are obtained by means of counted links. - The minimum number of sensors to be installed on links for total link observability is derived. - A continuous dynamic network loading problem considering overtaking that gives link travel times and flows all over the network at any time of the period under study. - Some graphical methods to analyze trajectory plots with and without overtaking that lead to an evaluation of the system state (speed, acceleration, etc.) by means of the trajectories physical characteristics (slope, curvature, etc.). All these models have been tested in fictitious and real traffic networks (the Spanish cities of Cuenca and Ciudad Real), with the aim of analyzing its characteristics and performance. Furthermore, a literature review about existing traffic problems and the most widely used models to solve them is done.