A Family Particle Swarm Optimization Based on the Animal Collective Behavior

Abstract

To study the organizational structure of particles in particle swarm optimization (PSO), we have proposed the family PSO (FPSO) previously. To further study the internal structure of FPSO, this paper introduced the animal collective behavior into the FPSO. It made the interaction ruling among particles was not based on random selection but topological distance. Each family interacted on average with a fixed number of neighbors, rather than with all neighbors within a fixed metric distance. Simulations for four benchmark functions demonstrated that the interaction ruling based on topological distance among particles was more reasonable than that on random selection.

Particle Swarm Optimization Based on Punctuated-equilibrium Model

“…Therefore, avoiding the local optima and maintaining a population's diverse have become the goals of many researches. Recently, multi-swarm and multi-role techniques [3][4][5][6][7] have been proposed to achieve these goals. The whole population was divided into a large number sub-swarms and the sub-swarms were regrouped frequently by using various regrouping schedules in Ref.…”

“…The concept of family was introduced in the PSO and analyzed the convergence of algorithm theoretically in Ref. [4,5]. Three roles of leader, rambler, and follower were assigned for particles at each generation based on their fitness in Ref.…”