The initial study on the potential of super-sized swarm in PSO

Abstract

In this initial study it is addressed the issue of population size for the PSO algorithm. For many years now it is understood that population size of several dozens is sufficient for the vast majority of optimization tasks. With strict limitation of cost function evaluations (CFEs) the setting is typically limited to adjusting the number of iterations of the algorithm. In this study it is investigated the possibility of using population of thousands of particles and its effect on the performance of the algorithm in limited CFEs. It is also proposed an alternative setting of acceleration constants in order to improve the performance of the PSO with super-sized population. The performance of the proposed method is tested on IEEE CEC 2013 benchmark set and compared with original PSO design and state of art methods. © Springer International Publishing Switzerland 2015.