PSO with adaptive clearing operator to promote diversity

Abstract

In this paper an adaptive Clearing operator is applied to the Particle Swarm Optimization Metha-heuristic (PSO) in order to encourage diversity in the population. The goal is to determine whether the use of the Clearing operator improves levels of exploration PSO algorithm increasing diversity of the population without adding too much computational cost to the search process. Thus, an adaptive way is proposed to control the allowable distance between solutions and disturb the solutions that are very close to another with better quality. This idea allows to study different alternatives as to disturb the selected solutions, presenting at work two different approaches based on the current position and in the best historical position. In addition a study is detailed to determine when to be applied Clearing operator for maximum contribution to the search process implemented by PSO. The proposal was tested in a set of 20 multi-modal functions, where the results showed that the operator clearing improves exploration of meta-heuristic PSO, yielding signi cantly superior results in quality that Original PSO model.