The Novel Characterizing Method of Collective Behavior Pattern in PSO

Although swarm intelligence algorithms have attracted extensive attention, there is little research on the behavior of collective dynamics. Inspired by the moving patterns of fish school, we propose a visualization method of collective behavior patterns based on the velocity field, discover various collective behavior patterns, and propose a discriminate index named swarm trend factor. In addition, this paper proposes a novel swarm states division method, on the basis of swarm trend factor. In the experiments, we demonstrate that swarm trend factor can reflect the performance of PSO. And, we also compare the difference between swarm trend factor and another swarm state division method.