A variant of differential evolution based on permutation regulation mechanism

Abstract

Differential evolution (DE) is a stochastic, population based search method, which has emerged as a powerful tool for solving optimization problems. This paper presents a novel algorithm based on traditional DE and permutation regulation mechanism to enhance the performance of DE. As a kind of enhanced learning strategy, the permutation regulation mechanism, which makes efforts in the evolving, is constructed by rearranging the selected three father vectors. In order to verify the performance of the proposed algorithm, two experiments on some well-known benchmark functions are conducted. Performance compared with other three DE variants confirms that the new algorithm outperforms better in terms of solution accuracy. © 2010 Springer-Verlag Berlin Heidelberg.