摘要
针对不可分解函数求解问题,基于合作式协同进化(cooperative co-evolutionary,CC)框架,发展一种双系统协同进化算法。该算法给出一种双系统A,B的CC框架新结构形式及其相应的协调机制,以增加算法的多样性和收敛性;给出双系统A,B各自求解的两种算法,例如差异进化、改进粒子群算法选择原则和匹配方式,使该两种算法具互补性,并且与双系统A,B各自角色相匹配,目的是提高基于CC框架双系统算法的计算性能。经不可分解函数集(维数D=1 000)测试表明,本文算法计算性能(计算精度和标准差)与其他3种典型算法相比,对于其中某些函数求解占优,总体上4种算法对函数集的求解各有所长,具有互补性。
Aiming at solving the non-separable function optimization problem, a dual-system cooperative coevolutionary differential evolution particle swarm optimization algorithm （DCCDE/PSO for short） is developed based on the dual-system cooperative co-evolutionary （CC） framework. The proposed algorithm gives a new CC framework of the dual-system A and B and its corresponding coordination mechanism for improving the diversity and convergence, and gives two algorithms for example differential evolution （DE）, the improved particle swarm optimization （PSO） that it solves the systems A and B respectively, as well as complementary and matches with the roles that the systems A and B play in the dual system. The purpose is to improve computational performance of the dual system algorithm based on the CC framework. The numerical experimental results of non separable Benchmark functions （1000 dimensional） show that the performance （computational accuracy and standard deviation） of the proposed DCCDE/PSO compared favorably against other three representative algorithms has advantages for some of functions and as a whole the four algorithms had theirsetves＇s strengths for the Benchmark functions and each complemented the other.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2016年第11期2660-2669,共10页
Systems Engineering and Electronics
基金
国家自然科学基金(61472062)资助课题
关键词
不可分解函数优化
合作式协同进化
双系统框架
现代启发式算法
算法选择和匹配
non-separable function optimization
cooperative co-evolution
dual-system framework
modern heuristic algorithm
algorithm selection and match
