On balancing neighborhood and global replacement strategies in MOEA/D

Abstract

In recent years, the multiobjective evolutionary algorithm based on decomposition (MOEA/D) has shown superior performance in solving multiobjective optimization problems (MOPs). In MOEA/D, the adaptive replacement strategy (ARS) plays a key role in balancing convergence and diversity. However, existing ARSs do not effectively balance convergence and diversity. To overcome this disadvantage, we propose a mechanism for adapting neighborhood and global replacement. This mechanism determines whether a neighborhood or global replacement strategy should be employed in the search process. Furthermore, we design an offspring generation strategy to generate high-quality solutions. We call this new algorithm framework MOEA/D-ARS. The experimental results suggest that the proposed algorithm performs better than certain state-of-the-art MOEAs.