Abstract: The traditional structure of population is modified based on Differential Evolution（DE） and a new strategy of population setting is proposed, which is sorted based on the fitness values of individuals. A new method with saltatory and sampling in a nonrandom order is used to select candidates for mutation operation, and a method of survival of the fittest is used in individual selection operation. Thus, the ordered-sampling differential evolution algorithm is proposed, which has a better performance both in convergence rate and robustness compared with traditional differential evolution and particle swarm optimization via a benchmark function simulation test.

Key words: differential evolution, ordering, sampling, individual sampling

摘要： 基于传统的差分演化，对其种群的内部结构进行调整，提出了一种基于个体适应度排序的种群设置策略。并通过个体采样方式来选择个体参与变异步骤，结合优胜劣汰的选择策略，提出了基于个体排序的采样差分演化算法。通过优化测试函数的仿真试验，与传统差分演化算法和粒子群算法相比较，基于排序的采样差分演化算法在收敛速度和鲁棒性等方面有较好的优势。

关键词: 差分演化, 排序, 采样, 个体采样