A selective teaching-learning based niching technique with local diversification strategy

Abstract

Real world problems present instances where more than one optimal solution can be obtained for a system under consideration so as to switch between them without considerably affecting efficiency. In such instances the idea of niching provides a solution. In this paper we propose a swarm-based niching technique that enhances diversity by Teaching and Learning strategy that adapts to the local neighbourhood by controlled exploitation and the knowledge learned helps to preserve population diversity. Our algorithm, imitates the local-explorative swarm behaviour to hover around local sites in groups, exploiting the peaks with high degree of accuracy, is called TLB-lDS (Teaching-Learning Based Optimization with Local Diversification Strategy), without using any niching parameter. TLB-lDS algorithm is compared against sophisticated niching algorithms tested on a set of standard numerical benchmarks. © 2012 Springer-Verlag.