A new dominance relation based on convergence indicators and niching for many-objective optimization

Abstract

Maintaining a good balance between convergence and diversity is crucial in many-objective optimization, while most existing dominance relations can not achieve a good balance between them. In this paper, we propose a new dominance relation to better balance the convergence and diversity. In the proposed dominance relation, a convergence indicator and a niching technique based adaptive parameter are adopted to ensure the convergence and diversity of the nondominated solution set. Based on the proposed dominance relation, a new many-objective evolutionary algorithm is proposed. In the algorithm, a new distribution estimation method is proposed to obtain better solutions for mating selection. Experimental results indicate that the proposed dominance relation outperforms existing dominance relations in balancing the convergence and diversity and the proposed algorithms has a competitive performance against several state-of-art many-objective evolutionary algorithms.