Stratégies d'optimisation par modélisation explicite des différents acteurs influençant le processus de conception

摘要

The commercial success or failure of engineered systems has always been significantly affected by their interactions with competing designs, end users, and regulatory bodies. Designs which deliver too little performance, have too high a cost, or are deemed unsafe or harmful will inevitably be overcome by competing designs which better meet the needs of customers and society as a whole. Recent efforts to address these issues have led to techniques such as design for customers or design for market systems. In this dissertation, we seek to utilize a game theory framework in order to directly incorporate the effect of these interactions into a design optimization problem which seeks to maximize designer profitability. This approach allows designers to consider the effects of uncertainty both from traditional design variabilities as well as uncertain future market conditions and the effect of customers and competitors acting as dynamic decision makers. Additionally, we develop techniques for modeling and understanding the nature of these complex interactions from observed data by utilizing causal models. Finally, we examine the complex effects of safety on design by examining the history of federal regulation on the transportation industry. These efforts lead to several key findings; first, by considering the effect of interactions designers may choose vastly different design concepts than would otherwise be considered. This is demonstrated through several case studies with applications to the design of commercial transport aircraft. Secondly, we develop a novel method for selecting causal models which allows designers to gauge the level of confidence in their understanding of stakeholder interactions, including uncertainty in the impact of potential design changes. Finally, we demonstrate through our review of regulations and other safety improvements that the demand for safety improvement is not simply related to ratio of dollars spent to lives saved; instead the level of personal responsibility and the nature and scale of potential safety concerns are found to have causal influence on the demand for increased safety in the form of new regulations.