A weak product on partially ordered sets to define a compromise between multiobjective optimization problems

Benoît Guédas (1) , Xavier Gandibleux (2) , Philippe Dépincé (1)

1 IRCCyN - Institut de Recherche en Communications et en Cybernétique de Nantes
2 LINA - Laboratoire d'Informatique de Nantes Atlantique

Multidisciplinary design optimization (MDO) deals with complex engineering problems which are decomposed into several subproblems called disciplines. The disciplines are often hierarchically organized. Moreover, each discipline may have many conflicting objectives to achieve at the same time. Thus, at each level of the hierarchy, trade-offs have to be found between multiobjective optimization problems. Multiobjective multidisciplinary methods designed to solve hierarchical multiobjective optimization problems such as EM-MOGA, MORDACE or COSMOS are searching for the whole Pareto set which corresponds to the multiobjective optimization problem that bring all the objectives of the problem together. But the solutions are just in a subset of the whole Pareto set. In this paper, we propose a definition of compromise between disciplines which are multiobjective optimization problems.

Mots clés: Multidisciplinary Design Optimization; Multiobjective Optimization; Compromise; Ordered Sets

Domaines: Recherche opérationnelle [math.OC]

Benoît Guédas, Xavier Gandibleux, Philippe Dépincé. A weak product on partially ordered sets to define a compromise between multiobjective optimization problems. 20th International Conference on Multiple Criteria Decision Making, Jun 2009, Chengdu, China. ⟨hal-00449888⟩