An improved self-adaptive differential evolution algorithm and its application

详细信息    查看全文

• 作者：Wu Deng ; Xinhua Yang ; Li Zou ; Meng Wang ; Yaqing Liu ; Yuanyuan Li
• 关键词：Differential evolution ; Dynamic multi-population parallel ; Chaotic-local-search strategy ; Adaptive parameter adjustment ; High-dimensional complex problem
• 刊名：Chemometrics and Intelligent Laboratory Systems
• 出版年：2013
• 出版时间：15 October, 2013
• 年：2013
• 卷：128
• 期：Complete
• 页码：66-76
• 全文大小：640 K

文摘

Because of the deficiencies in the global searching ability and convergence speed of the differential evolution (DE) algorithm in solving high-dimensional problems, this paper proposes an improved self-adaptive differential evolution algorithm with multiple strategies (ISDEMS) algorithm using a different search strategy and a parallel evolution mechanism. In the ISDEMS algorithm, the population is dynamically divided into multiple populations according to the fitness value of the individuals. Multiple strategies are used to improve the diversity of the individuals, to avoid premature convergence and to ensure efficiency in exchanging information among sub-populations. In addition, a self-adaptive adjustment method is introduced to automatically adjust the scaling and crossover factors during the running time. It is helpful to improve the robustness of the ISDEMS algorithm. To prove the validity of the ISDEMS algorithm for solving complex problems, thirteen benchmark problems and one real-life problem are selected to validate the performance of the ISDEMS algorithm. The experiment results show that the ISDEMS algorithm is better in terms of search precision and convergence performance than the DE, ACDE and SACDE algorithms from the literature.