摘要
在差分进化算法的优化过程中,不断生成更优的解并采用达尔文的"适者生存"思想进行择优保留,这样的遗弃会导致个体有效成分缺失,并失去对新空间的探索开发能力,降低种群多样性,进而使算法早熟收敛并陷入局部最优,因此需要改进差分进化算法并权衡算法的空间探索和开发能力,提高解的精确度和算法收敛速度.为此,基于高斯扰动和免疫搜索策略的差分进化算法被提出.首先,通过生物免疫系统的信息处理机制实现自适应地修正差分进化算法中的缩放因子和交叉因子,以满足优化过程中对这两个参数的取值要求;然后,通过基于高斯扰动的交叉操作算子增加种群的多样性,扩展算法的探索空间,以避免陷入局部最优,进而提高算法的性能.实验结果表明,该优化算法具有良好的寻优性能.
During the evolution process of differential evolution algorithm,good solutions are generated and the＇survival of the fittest＇theory of Darwin is employed to select the better solutions,which results in failures of the abandoned individual ＇ s effective component and the reduction of population diversity.Thus the differential evolution algorithm is not able to explore new space and traps in local optima.So the differential evolution algorithm has been shown to have certain weaknesses,especially if the global optimum should be located using a limited number of function evaluations.In order to remedy these defects of the differential evolution algorithm mentioned above,weighting space exploration and exploitation is employed for improving it to enhance the convergence speed and solution quality.In this paper,improved differential evolution algorithm based on Gaussian disturbance and immune search startegy is proposed to solve the global optimization problems.Our approach combines several features of previous evolution algorithms in a unique manner.In the novel approach,firstly,two parameters,scaling factor and crossover rate,are self-adapted by immune system.Secondly,in the crossover phase the best vector is disturbed by Gaussian probability distribution.Thirdly,the trial vector is obtained by crossover between the mutation vector and the disturbed best vector.In the optimizing process Gaussian disturbance increase the variety of the individual,which can make the algorithm avoid trapping into the local optima and improve its performance.It is shown empirically that the novel improved differential evolution algorithm has high performance in solving the benchmark functions.The Gaussian disturbance is also employed for local optimiation to avoid population diveristy descent and individual stagnant evolution.The results of experiments show that the gaussian disturbance in the crossover phase can improve the ability of searching an optimum solution and increase the convergence speed.
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第2期202-209,共8页
Journal of Nanjing University（Natural Science）
基金
江苏省科技支撑计划(BE2012112)
淮安市科技支撑计划(工业)(HAG2011044
HAG2011045)
关键词
差分进化
高斯扰动
局部优化
免疫搜索策略
全局收敛
differential evolution
gaussian disturbance
local optimization
immune search strategy
global convergence
作者简介
通讯联系人，E-mail：ajason509@163．com
