Optimization of Multi-Objective Stochastic Systems Beyond Two Objectives Using Direct Search Methods

Abstract

In this work, existing techniques and methodologies are evaluated, combined, and sometimes modified, to form a new algorithm for the optimization of multi-objective problems with or without a closed form representation and with or without stochastic responses. The resulting, flexible algorithm combines direct search methods and singleobjective formulations, extending the BiMADS approach to more than two objectives. Surrogates and n-dimensional visualizations can also be available within the context of the algorithm to allow for a complete, yet efficient approximation for two or more objectives.