[1]赵新超,熊卿,冯帅.均匀邻域对位的自适应差分进化算法[J].集美大学学报(自然版),2021,26(1):72-81.
 ZHAO Xinchao,XIONG Qing,FENG Shuai.Self-adaptive Differential Evolution Algorithm Based on Uniform Neighborhood Opposition[J].Journal of Jimei University,2021,26(1):72-81.
点击复制

均匀邻域对位的自适应差分进化算法()
分享到:

《集美大学学报(自然版)》[ISSN:1007-7405/CN:35-1186/N]

卷:
第26卷
期数:
2021年第1期
页码:
72-81
栏目:
数理科学与信息工程
出版日期:
2021-01-28

文章信息/Info

Title:
Self-adaptive Differential Evolution Algorithm Based on Uniform Neighborhood Opposition
作者:
赵新超熊卿冯帅
(北京邮电大学理学院,北京 100876)
Author(s):
ZHAO XinchaoXIONG QingFENG Shuai
(School of Science,Beijing University of Posts and Telecommunications,Beijing 100876,China)
关键词:
差分进化对位学习均匀变异多阶段扰动
Keywords:
differential evolutionopposition-based learninguniform mutationmultiple stages perturbation
摘要:
针对对位差分进化算法依然存在探索能力差和早熟收敛问题,提出一种基于均匀邻域对位的自适应差分进化算法。该算法在对位点所在局部邻域作适应性的小幅均匀变异操作,用以扩大对位点的搜索区域,从而提高跳出局部陷阱的概率;在对位点均匀变异操作中,变异步长利用当前群体中所有个体在每一维度的最大最小值的差距作自适应的调节,通过实时利用群体信息平衡了全局搜索与局部勘探的关系,提高算法的收敛速度;在算法搜索过程中引入多阶段扰动策略,以进一步增加算法群体实时多样性与算法所处搜索阶段的适应性,算法后期在一定程度上加强了对当前解所在邻域内的精细搜索。采用CEC 2014中不同类型基准测试函数进行仿真实验,并与其他差分算法进行对比,所提算法在10个测试函数上都取得最优的平均结果,证明所提算法具有更稳定、更优异的算法性能和更好的收敛精度。
Abstract:
In order to solve the problem of poor exploration ability and premature convergence in the oppositionbased differential evolution algorithm,this paper proposed an oppositionbased differential evolution algorithm with neighborhood-based self-adaptive uniform mutation.The algorithm performed an adaptive small uniform mutation operation in the local neighborhood to expand the search area of the opposition point,thereby increasing the probability of jumping out of the local trap.As well as the uniform mutation operation was concerned,the step size of the mutation was adaptively adjusted by the difference between maximum and minimum of all individuals in the current population in each dimension.The relationship between the global exploration and the local exploitation was well balanced by the online population information and individual information,so the algorithmic convergence speed was improved.A multiple stage perturbation strategy was also proposed to further balancing the real-time diversity of the algorithm and the adaptability of the search stage in which the algorithm was located.In the latter stage of the algorithm,the fine search in the neighborhood of the current solution was strengthened to some extent.The simulation experiments were carried out with different types of benchmark functions of CEC 2014.The proposed algorithm was compared with other differential algorithms and achieves the best average result on 10 test functions.It was proved that the proposed algorithm has more stable,better algorithm performance and better convergence accuracy.
更新日期/Last Update: 2021-03-25