Improved algorithms based on the simple particle swarm optimization

Abstract

As one of the representative algorithms in swarm intelligence, particle swarm optimization has been applied to many fields because of its several merits, such as simple concept, easy realizing and fast convergence rate in the early evolutionary. However, it still has some disadvantages such as easy falling into the local extremum, slow convergence velocity and low convergence precision in the late evolutionary. Two new algorithms based on the simple particle swarm optimization are proposed to try to improve the precision of the algorithm in a certain error range of the length of time. The algorithms have been simulated and compared with the particle swarm optimization and the simple particle swarm optimization. The simulations show that the algorithms have a higher convergence precision for some functions or a particular issue. © 2013 Springer-Verlag Berlin Heidelberg.