Rationalizing approaches to multi-objective optimization in systems architecture design

摘要

The decision process is considered very important in many systems engineering methodologies, and multi-objective optimization is known to be an important tool for decision-making support. However, there is still a lack of a theoretical ground to guide its smart use. In a context of conception of naval systems, we present the outline of a process designed to output a set of alternatives for its architecture, optimizing several objectives and subject to constraints. The key point here is to outline the importance of the development of methods to choose smartly our approach to multi-objective optimization, in order to get the desired quality of results along several criteria, under several cost constraints. These choices include the choice of the algorithm and its parameters, but also the choice of the width of the architecture space. We show how complexity measures may be used to this end. We also show how we can take advantage of design with architecture frameworks in our approach.