Multi-objective particle swarm optimization with R2 indicator and adaptive method

Abstract Multi-objective particle swarm optimization algorithms encounter significant challenges when tackling many-objective problems. This is mainly because of the imbalance between convergence and diversity that occurs increasing selection pressure. In this paper, a novel adaptive MOPSO (ANMPSO) algorithm based on R2 contribution method developed to improve performance MOPSO. First, new global best solutions mechanism with introduced select leaders better convergence. Second, obtain uniform distribution particles, an used guide flight particles. Third, re-initialization strategy proposed prevent particles from trapping into local optima. Empirical studies large number (64 in total) problem instances have demonstrated ANMPSO performs well terms inverted generational distance hyper-volume metrics. Experimental practical application also revealed could effectively solve problems real world.