Family Particle Swarm Optimization

To overcome the premature convergence of particle swarm optimization (PSO), we introduce a sociological conception, called family, into the PSO. Family is a common activity form of life. Each family in a population usually competes for the resource with other family and enhances collaboration among family members. We introduced this sociological conception into PSO and proposed the family PSO(F-PSO), in which the particle swarm consists of different families and each family consists of different members. Simulations for nine benchmark functions demonstrated that F-PSO could speed up the optimization and learning processes. The experiments also showed that an individual had smaller fluctuations during finding the global best fitness by F-PSO than by the original PSO. Results indicate the effective impact of this conception on PSO.