System Engineering for dependency analysis - a Bayesian approach: application to obsolescence study.

Throughout its life cycle, systems undergo several modifications in their architecture. These changes target at remaining competitive and responding quickly to new customer requirements. However, any entity change (i.e. component, function or functionality) can produce unexpected consequences, propagated throughout the whole system architecture. It is then necessary to model, predict and control them. System engineering tools and techniques allow dealing with complex systems design. That is why we have developed a novel methodology to analyze changes using a system engineering methodology called ARCADIA, developed by Thales and its associated software Capella. The obtained models allow mapping various kinds of dependencies within a system architecture. The method, presented in this paper, shows how these models are used to integrate change propagation and transform them into Bayesian networks. A set of experiments allows then to obtain insightful pieces of knowledge about the changes propagation. An illustrative case is developed with a focus of particular changes caused by obsolescence of component, function or functionality. [ABSTRACT FROM AUTHOR]