An adaptive hybrid optimizer based on particle swarm and differential evolution for global optimization
- 169 Downloads
This paper presents extensive experiments on a hybrid optimization algorithm (DEPSO) we recently developed by combining the advantages of two powerful population-based metaheuristics—differential evolution (DE) and particle swarm optimization (PSO). The hybrid optimizer achieves on-the-fly adaptation of evolution methods for individuals in a statistical learning way. Two primary parameters for the novel algorithm including its learning period and population size are empirically analyzed. The dynamics of the hybrid optimizer is revealed by tracking and analyzing the relative success ratio of PSO versus DE in the optimization of several typical problems. The comparison between the proposed DEPSO and its competitors involved in our previous research is enriched by using multiple rotated functions. Benchmark tests involving scalability test validate that the DEPSO is competent for the global optimization of numerical functions due to its high optimization quality and wide applicability.
Keywordsglobal optimization statistical learning differential evolution particle swarm optimization hybridization adaptation rotated function
Unable to display preview. Download preview PDF.
- 4.Kennedy J, Mendes R. Population structure and particle swarm performance. In: Proceedings of the World Congress on Computational Intelligence, Honolulu, HI, USA, 2002. 1671–1676Google Scholar
- 11.Liang J J, Suganthan P N. Dynamic multi-swarm particle swarm optimizer. In: Proceedings of the IEEE Swarm Intelligence Symposium, Pasadena, California, USA, 2005. 124–129Google Scholar
- 12.Zhang W J, Xie X F. DEPSO: hybrid particle swarm with differential evolution operator. In: Proceedings of the IEEE International Conference on System, Man, Cybernetics, Washington, DC, USA, 2003. 3816–3821Google Scholar
- 17.Sutton A M, Whitley D, Lunacek M, Howe A. PSO and multi-funnel landscapes: how cooperation might limit exploration. In: Proceedings of the Annual Conference on Genetic and Evolutionary Computation, Seattle, Washington, USA, 2006. 75–82Google Scholar
- 19.Liang J J, Suganthan P N, Deb K. Novel composition functions for numerical global optimization. In: Proceedings of the IEEE Swarm Intelligence Symposium, Pasadena, California, USA, 2005. 68–75Google Scholar