Multi-objective Optimization Problems Solving based on Evolutionary Algorithm

Art. 04 – Vol. 25 – Nr. 2 – 2015

Iulia Cristina Rădulescu
iulia.radulescu1702@yahoo.com

POLITEHNICA University of Bucharest

Abstract: Most optimization problems that arise in practice have multiple objectives because they suppose to optimize simultaneously several objective functions. In this article we analyze several approaches for solving multi-objective optimization problems, focusing on the approaches based on evolutionary algorithms. We present a classification of optimization techniques in enumerative, deterministic and stochastic. From the stochastic optimization techniques are detailed Evolutionary Algorithms – EA. These algorithms are successfully used for solving multi-objective optimization. We present the basic concepts used in evolutionary algorithms and an overview of several evolutionary algorithms for solving multi-objective optimization problems. Finally, we solve four multi-objective optimization problems using a well-known evolutionary algorithm called the Non-Dominated Sorting Genetic Algorithms – NSGA-II. The Matlab program, that implements the algorithm NSGA-II, computes the Pareto frontier of the four multi-objective optimization problems considered.

Keywords: Multi-objective optimization, Evolutionary Algorithms, Non-Dominated Sorting Genetic Algorithms, Pareto frontier.

View full article