Linear statistical decision models for air traffic flow simulation: a comparative study
TABTI ; EL FAOUZI ; FONDACCI ; THOMAS
Type de document
ARTICLE DE PERIODIQUE
Langue
anglais
Auteur
TABTI ; EL FAOUZI ; FONDACCI ; THOMAS
Résumé / Abstract
Research on Air Traffic Flow Management (ATFM) has provided various models. Generally, these models are used for the strategic or even tactical control of flows. In this paper, we will focus on short term ATFM models. We first give an approach of macroscopic traffic flow modeling for the short term control of flights and then show how statistical decision techniques will contribute to the conception of such models. Basically, the statistical approach is used to estimate dynamically two exogenous parameters of the traffic flow model developed in this paper: the flow orientation of an air traffic flow through a volume of controlled airspace (sector) and its travel time within the volume. Two linear discriminant analysis methods designed for categorical variables are used to estimate the decision rules of the previous output's parameters. Comparisons of the simulation results are made through a set of 35 sectors by varying, in each case, the set of descriptor variables. Mots-clés : Trafic aérien ; Gestion trafic ; Simulation numérique ; Système aide décision ; Analyse statistique ; Prédiction linéaire ; Analyse discriminante ; Base donnée ; Evaluation performance ; Critère décision ; Critère pertinence ; Validation croisée ; Modèle prévision ; Analyse multivariable.
