Multi-Objective Evolutionary Algorithm based on a Novel Angle Dominance Relation
The paper studies how to achieve the balance between the convergence and diversity of multi-objective evolutionary algorithms.Based on the theories of evolutionary populations'dominance relation and diversity,and drawing on the advantages of grid dominance relation and angle dominance relation,this paper presents a novel dominance relation to describe the distribution of evolutionary populations in the objective space,aiming at achieving a better balance between the convergence and diversity.A multi-objective evolutionary algorithm based on a novel angle dominance is developed.By using the novel angle dominance relation,it can improve the selection pressure and maintain the diversity of evolutionary populations.In the framework of NSGAⅡ algorithm,the convergence,diversity and convergence rate of the newly proposed dominance relation are compared with those of the existing two dominance relations.The results show that the performance of the IGD,GD and Spacing obtained by the novel dominance relation on the DTLZ and WFG test suites has been improved,with high-quality solution sets obtained more quickly.It is concluded that the dominance relation proposed in this paper can balance convergence and diversity more effectively.
万方数据
维普
