Adaptive fractional order particle swarm optimization using swarm activity feedback and mutation operator

The basic particle swarm optimizer with fractional-order (FOPSO) is easy to fall into premature convergence, because its overall performance depends on the fractional order α. To solve the problem, a new adaptive fractional-order PSO algorithm, SFOPSO is proposed, by cooperating mutation operators into swarm activity feedback with S-model. During the iteration of this new algorithm, the fractional-order α of particles is adjusted adaptively according to the swarm activity with S-model and the activity value of single particles. At the same time, to enhance the ability of the swarm to escape out of local optimum during the process of exploitation or exploration, the hybrid model was designed by using mutation operators. The convergence of the proposed algorithm SFOPSO is analyzed theoretically and the experimental results show that the proposed algorithm is practicable and effective in improving convergence accuracy and convergence speed.