Analyse de réseaux et intégration d'une approche déductive aux systèmes d'information géographique : l'exemple du transport routier des matières dangereuses.

Abstract

The main topic of this thesis is transportation geography and more specifically the problem of safe path finding for the transportation of dangerous goods. Most of the path finding algorithms use impedance as a unique variable for optimisation. This approach is limited, especially in the domain of dangerous goods, where quantitative and qualitative variables must be considered when selecting a path. This research shows that decision theory and expert systems are among the best tools to make conventional path finding algorithms more flexible and able to deal with quantitative and qualitative data. A path finding model based on the coupling of the Dijkstra algorithm and an expert system has been developed in this thesis. The integration of this model to a geographic information system creates a spatial decision support system for the transportation of dangerous goods. Formalisation of the safe path finding model and development of the decision support system are demonstrated by an example of the Ottawa-Hull area.