Concepts et méthodes pour le développement d'environnements de simulation intelligents. Application au bâtiment

Numerical simulation is a powerful design and audit tool. But in the building industry, it is not as widely spread as in other industries (electronics, car industry, space industry...). Meanwhile, in building related research centers, various simulation codes have been developed. These codes are generally powerful on the numerical level and can cope with a large range of building related issues. Nevertheless, these codes lack some features in order to make their use simple and productive (user guidance, coherency check, share of data...). The aim of this work is to design an Intelligent Simulation Environment (ISE) taking advantage of both numerical performances of existing simulation codes and symbolic performances of knowledge based systems. The symbolic part of the ISE will act as a semantic server taking in charge operations on a high level of abstraction (sharing of data, reasoning strategies...) and controling the numerical part of the ISE which will act as a number cruncher. After a short presentation of the modelling/simulation/result analysis process and of the possible Human/Computer interactions, the components of the ISE are described using an object oriented formal specification method. A first application to the building field is then presented. This application is implemented within a knowledge modelling software platform developed in the CSTB and uses the Integrated Data Model resulting from the IDM task of the CEC project COMBINE. Key-words : Modelling Simulation Building Knowledge Based Systems Human/Computer dialogue. Object Oriented Design.