An improved NSGA-II to solve multi-objective optimization problem

Abstract

NSGA-II(nondominated sorting genetic algorithm II) is a popular multi-objective evolution algorithm (MOEA), which applies binary tournament selection, elitist preserving strategy, nondominated sorting and crowding distance mechanism to obtain a good quality and uniform spread nondominated solution set. In this paper, an improved version of NSGA-II (INSGA-II) is proposed aiming to increase the diversity and enhance the local search ability. The INSGA-II has two populations: interior population and external population. The external population is used to store the nondominated solution found in the search process, while the interior population takes part in generation evolution. When the interior population tends to converge, it is updated by the individuals in the external population and generated randomly. A local search based on the amount of domination is applied to enhance the local search ability. In order to demonstrate the effectiveness of the proposed INSGA-II, comparisons with NSGA-II is carried out by ten functions, and the results show the quality and spread of INSGA-II are better than NSGA-II.