Development of genetic algorithms based weighted method for multi-objective optimization

Abstract

This study presents a method to determine weights of objectives in multi-objective optimization without decision maker's preference This method is based on Scalarization and normalization of convex objective functions as well as the principle of proportional selection of Genetic Algorithms procedure. The proposed method is shown with a numerical example and several fuzzy solution approaches are used to get a solution by using obtained weights. Also the results of problems that are obtained from literature are presented. The method developed in aggregating the two or more conflicting objective functions show the possibility of solving multi-objective problems without: (i) the articulation of preferences among the criteria and (ii) arbitrary choice of weights; which usually required sensitivity analysis to be carried out.

Keywords: weighted method, multi-objective optimization, genetic algorithms, Pareto method, aggregated objective functions