A Novel Swarm Model With Quasi-oppositional Particle

Abstract

This paper proposes an enhanced version of the opposition-based PSO (OCLPSO) that we call the quasi-oppositional comprehensive learning particle swarm optimizers (QCLPSO). OCLPSO employs opposition based learning (OBL) for population initialization and also for exemplar selecting. Instead of opposition numbers, QCLPSO uses quasi opposite particles, which is generated from the interval between the median and the opposite position of the particle. Mathematical proof shows that quasi-opposite particles have a higher chance to be closer to the optimum than opposite particles in problems without apriori information. Experiments were conducted on benchmark functions and comparisons between the original CLPSO, OCLPSO and the QCLPSO are presented. The results are very promising, as the new algorithm outperforms CLPSO and OCLPSO in terms of convergence speed and global search ability.