The Nondominated Sorting Genetic Algorithm: NSGA

Abstract

Unlike many search methods that develop a single “current best” solution and then try to improve it, a GA maintains a set of possible solutions called the population. At the intuitive level this would suggest that a suitably designed GA might be able to capture the members of the Pareto optimal set of solutions, if Pareto optimality were somehow used as the basis for measuring fitness. In this chapter we describe several approaches to endow GAs with the ability to capture and preserve the Pareto solutions in multiobjective optimization. We then describe the Nondominated Sorting Genetic Algorithm (NSGA), a multiobjective GA designed by Srinivas and Deb (1995) that seeks out Pareto solutions efficiently. A numerical problem, a bi-criteria robust design of an electronic filter, is then solved to illustrate its efficacy.