摘要
为了提高进化算法的全局收敛性,本文提出了一种改进的双精英种群协同进化算法。该算法根据生物学遗传多样性的理论,让两类精英种群同时进化。两类种群之间既分工合作又相对独立,不同类型的精英种群采用不同的进化策略,分别担负着全局最优解的深度与广度的搜索任务。为了避免算法陷入局部最优解,每隔一定代数就要淘汰一些较差的种群,并让这些种群重新进化;在与四个优秀算法进行比较的实验中得出结论,新算法综合领先于这四个算法。
To enhance the global astringency of evolutionary algorithm,a new double elite population coevolutionary algorithm has been put forward in this paper. According to the theory of biological diversity,this algorithm could have the two types of elite populations evolve simultaneously. These two types could not only collaborate but also remain independent. The elite populations of diverse types implement different evolutionary strategies and undertake the task of searching the depth and width of global optimal solution respectively.To avoid the algorithm traps into local optimal solution,those relatively poor populations should be eliminated naturally after certain evolutionary generations,and making these populations to evolve again. Comparing with four excellent algorithms by experiments,this paper concludes that the new algorithm is comprehensively ahead of these four algorithms.
出处
《南京工业职业技术学院学报》
2016年第1期21-25,共5页
Journal of Nanjing Institute of Industry Technology
关键词
淘汰
全局收敛性
小生境
大生境
配子
eliminate
global convergence
microhabitats
macrohabitats
gamete