Dynamic multi-swarm optimization based on clonal selection and particle swarm

Based on the promising fusion of the clonal selection and particle swarm principles, a dynamic multi-swarm optimization algorithm is proposed. In the approach, the whole swarm is divided into dynamic subpopulations, which are considered as the evolving antibodies. These subpopulations are further optimized by using the particle swarm method to increase the necessary antibody diversity. Moreover, they can exchange useful optimization information among themselves during the iteration procedure. The cloning, hypermutation, selection and receptor editing operators are also employed in the proposed hybrid scheme. Simulations demonstrate that the optimization algorithm can overcome the premature and slow convergence drawbacks of the standard particle swarm and clonal selection methods, and it is very effective in dealing with the challenging nonlinear function optimization problems.