Numerical optimization using organizational particle swarm algorithm

The classical particle swarm optimization (PSO) has its own disadvantages, such as low convergence speed and prematurity. All of these make solutions have probability to convergence to local optimizations. In order to overcome the disadvantages of PSO, an organizational particle swarm algorithm (OPSA) is presented in this paper. In OPSA, the initial organization is a set of particles. By competition and cooperation between organizations in every generation, particles can adapt the environment better, and the algorithm can converge to global optimizations. In experiments, OPSA is tested on 6 unconstrained benchmark problems, and the experiment results are compared with PSO_TVIW, MPSO_TVAC, HPSO_TVAC and FER The results indicate that OPSA performs much better than other algorithms both in quality of solutions and in computational complexity. Finally, the relationship between parameters and success ratio are analyzed.