期刊名称：International Journal of Computer Science Issues
印刷版ISSN：1694-0784
电子版ISSN：1694-0814
出版年度：2013
卷号：10
期号：2
出版社：IJCSI Press
摘要：Differential Evolution (DE) is a global numerical optimization algorithm which is robust, easy to use, and lends itself very well to parallel computation. The original DE adopts one-to-one tournament selection to select the individuals surviving in the next generation, which maybe lead to the individuals with lower object function values abandoned. In the evaluation strategy, (u+T)-selection has been proved to be an effective selection. In this selection method, we compare all the individuals mixed with the parents and the offspring, and the individuals with lower fitness must survive in the next generation. Inspired by this, we proposed an improve method to improve the speed of convergence of DE in the paper. We combine this selection with five DE mutation strategies, and testing them with the 13 benchmark functions, and most of them have better performance compared with the original DE in the same condition.
