Quantcast

An improved gradient-based NSGA-II algorithm by a new chaotic map model

Research paper by Tan Liu, Xianwen Gao; Qingyun Yuan

Indexed on: 14 Aug '16Published on: 03 Aug '16Published in: Soft Computing



Abstract

Abstract Gradient-based non-dominated sorting genetic algorithm II (G-NSGA-II) is successful for solving multi-objective optimization problems. However, the effectiveness of gradient-based hybrid operator is influenced by the distribution of individuals in the population. In order to solve the problem, based on the framework of G-NSGA-II, we propose an improved gradient-based NSGA-II algorithm by introducing a new chaotic map model named IG-NSGA-II. In this algorithm, a new hybrid chaotic map model is first established to initialize population for keeping the diversity of the initial population. Then, the substitution operation of chaotic population candidate is introduced to maintain the diversity and uniformity of the Pareto optimal solution set. Finally, the proposed algorithm is tested on several standard test problems and compared with other algorithms. The experimental results indicate that the proposed algorithm leads to better performance results in terms of the convergence to Pareto front or the diversity of the obtained non-dominated solutions.AbstractGradient-based non-dominated sorting genetic algorithm II (G-NSGA-II) is successful for solving multi-objective optimization problems. However, the effectiveness of gradient-based hybrid operator is influenced by the distribution of individuals in the population. In order to solve the problem, based on the framework of G-NSGA-II, we propose an improved gradient-based NSGA-II algorithm by introducing a new chaotic map model named IG-NSGA-II. In this algorithm, a new hybrid chaotic map model is first established to initialize population for keeping the diversity of the initial population. Then, the substitution operation of chaotic population candidate is introduced to maintain the diversity and uniformity of the Pareto optimal solution set. Finally, the proposed algorithm is tested on several standard test problems and compared with other algorithms. The experimental results indicate that the proposed algorithm leads to better performance results in terms of the convergence to Pareto front or the diversity of the obtained non-dominated solutions.