Technique de réduction adaptative de l'espace de recherche pour les méthodes d'optimisation par recherche locale
Résumé
When applied to practical problems, tuning parameters of optimization methods can be a critical problem for engineer users. In this article we suggest a novel approach to consider constrained optimization problems. Its principal caracteristics are flexibility, adaptativity and user friendship. Indeed constraints are totally separated from the cost function. It allows to be directly coupled with any local search algorithm in order to explore the solution space which is reduced in an adaptive manner. Its implementation is compared to other existing methods with discrete and continuous classical problems. Results show that the method approaches the best results thus offering a good trade-off between flexibity and quality for the solutions.