Non-uniform distributions of quantum particles in multi-swarm optimization for dynamic tasks

Abstract

This paper presents research considering mixed multi-swarm optimization approach applied to dynamic environments. One of the versions of this approach, called mQSO is a subject of our special interest. The mQSO algorithm works with a set of particles divided into sub-swarms where every sub-swarm consists of two types of particles: classic and quantum ones. The research is focused on studying properties of the latter type. Two new distributions of new locations for the quantum particles are proposed: static and adaptive one. Both of them are based on an α-stable symmetric distribution. In opposite to already published methods of distribution of new locations the proposed methods allow the locations to be distributed over the entire search space. Obtained results show high efficiency of the mQSO approach equipped with the proposed two new methods. © 2008 Springer-Verlag Berlin Heidelberg.