Modelisation totalement desagregee et orientee-objet appliquee aux transports urbains

Transportation planning models are mostly based on the "Transportation Planning Process" and aggregate models. In counterpart, this thesis presents an object-oriented and totally disaggregate approach aimed to improve the modelling tools and procedures in transportation planning. Its main contributions are the definition of a transportation object-oriented vocabulary, the architecture of new methods and the creation of related software.; The upcoming of new technologies in geographic information systems, the large origin-destination surveys and many new data querying tools are questioning the usage of aggregate models, based on zone data aggregation. In order to consider all available information within the organization information system, this thesis defines interactive planning methods based on data integration and conservation into modelling ("informodels").; The first step is to define transportation object-oriented concepts. Four meta-classes are explained: dynamic, kinetic, static and systemic objects. The resulting objects are used in several experiments on transportation supply and demand. The goal is to improve and multiply the usage of totally disaggregate models.; The experiments made on transportation supply objects (networks) demonstrate that the evolutive and adaptive object-oriented approach can conduct to new data integration without disturbing existing systems. In transit modelling, the approach benefits from the totally disaggregate modelling in information processing. The MADTOO software (MADITUC Transportation Object-Oriented) is the result from experiments conducted on transit and road networks objects.; The transportation demand objects are also used for experiments. In the Montreal context, large origin-destination surveys are achieved to obtain several data on population mobility, socio-demographics and supply use. The object-oriented approach completes the totally disaggregate approach by giving properties and methods to Household, Person and Trip objects in order to conduct deep survey data analysis. The processing of individual objects and collection of objects are experimented in the MADAME software (MADITUC data survey attributes and methods).; Finally, the thesis describes the reusability of methods in the construction of more complex models typically used in the totally disaggregate approach, like individual trip assignment on networks and incremental modal split models.; Since the information systems will tremendously grow in transportation organisation in the next years, planners should rely on an object-oriented modelling, combined with the totally disaggregate approach, to develop better ways to obtain a maximal use of available information.