Optimisation multi-objectifs à base de métamodèle pour des applications en mise en forme des métaux

Abstract

To apply multi-objective optimization algorithms to highly time expensive metal forming applications, the coupling of the NSGA-II genetic algorithm proposed by Deb with metamodels based on the Meshless Finite Difference Method (MFDM) proposed by Liszka and Orkisz is investigated. The importance of iteratively improving the metamodel during the optimization iterations is highlighted, as well as the capability to accurately determine the Pareto optimal fronts of the studied problems within less than hundred calculations.