Multi-Ring Dispersed Particle Swarm Optimization
Carmelo J. A. Bastos-Filho, Marcel P. Caraciolo, Pericles B. C. de Miranda, Danilo Ferreira de Carvalho · 2008
Particle swarm optimization (PSO) has been widely used to solve unconstrained optimization problems. However, problems in hyper dimensional spaces require the development of enhanced issues. For this, some variations of the original PSO form have been proposed, mainly concerning on the velocity update equation and sophisticated communication topologies of the swarm. In this paper, we propose a PSO topology based on multiples rings. In this approach, the acceleration coefficients and the communication inside each ring are adjusted based on a grade, which is calculated by evaluating a fitness comparison. The diversity of the system is provided by the ring rotations. The neighborhood of each particle is defined according its grade as defined in the dispersed PSO. A comparison with star, ring and simple multi-ring topologies was performed. Our simulation results showed that the proposed topology always achieve good results when compared to other approaches.