Firefly algorithm and pattern search hybridized for global optimization

Abstract

Firefly optimization algorithm is one of the latest swarm intelligence based optimization algorithm. A new hybrid optimization algorithm, which combines pattern search with firefly algorithm, namely FAPS, is proposed for numerical global optimization. There are two alternative phases of the proposed algorithm: the global exploration phase realized by firefly algorithm and the exploitation phase completed by pattern search. The performance of the proposed FAPS algorithm was tested on a comprehensive set of benchmark functions. The numerical experiments demonstrate that the new algorithm has high viability, accuracy and stability and the performance of firefly algorithm is much improved by introducing a pattern search method. © 2013 Springer-Verlag.