Particle swarm optimisation with time varying cognitive avoidance component

Abstract

Interactive cooperation of local best or global best solutions encourages particles to move towards them, hoping that better solution may present in the neighbouring positions around local best or global best. This encouragement does not guarantee that movements taken by the particles will always be suitable. Sometimes, it may mislead particles in the wrong direction towards the worst solution. Prior knowledge of worst solutions may predict such misguidance and avoid such moves. The worst solution cannot be known in prior and can be known only by experiencing it. This paper introduces a cognitive avoidance scheme to the particle swarm optimisation method. A very similar kind of mechanism is used to incorporate worst solutions into strategic movement of particles as utilised during incorporation of best solutions. Time varying approach is also extrapolated to the cognitive avoidance scheme to deal with negative effects. The proposed approach is tested with 25 benchmark functions of CEC 2005 special session on real parameter optimisation as well as with four other very popular benchmark functions.