Novel particle swarm optimization algorithm
Yong Zhou · Control theory & applications · 2008
Existing particle swarm optimization has disadvantages of local convergence and being sensitive to adjustable parameters. A novel particle swarm optimization algorithm is proposed to avoid the above disadvantages in this paper. Firstly, the formula for updating particles is simplified by analyzing the cognition rule of individuals to their environment, the update of a particle location is only related to its own velocity and the optimal particle location in its neighborhood. Secondly, strategies of mutation for superior particle velocities with a small probability and the random evaluation for inferior particle velocities are presented based on partition of particle velocities. Finally, the significant performance in quality of the optimal solutions, convergence speed and robustness of algorithm proposed in this paper are validated by optimizing four benchmark functions.