Abstract
Particle Swarm Optimizer (PSO) has gained wide applications in different fields. However, it loses its efficiency when facing an optimization problem in a noisy environment, since the inaccuracy of each particle's own "best" might mislead the entire swarm. Staying together is often of great selective advantage for social animals in nature. Social animals frequently make consensus decisions, and the decisions made by a majority of informed group members should be beneficial as they intend to avoid extreme outcomes or risky decisions. Inspired by this social behavior, a new particle swarm optimizer based on group decision-making (PSOGD) is developed for noisy optimization problems. Its significant feature is the elimination of resampling that is commonly used for noise optimization problems. The proposed algorithm is compared experimentally on 20 large-scale benchmark functions with various noise. The results demonstrate its superiority over other existing PSO variants.