混合差分进化与二次插值改进哈里斯鹰算法  

Hybrid Differential Evolution and Quadratic Interpolation Improved Harris Hawks Optimization Algorithm

作　　者：高建瓴 喻明毫 GAO Jian-ling;YU Ming-hao(School of Big Data and Information Engineering,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学大数据与信息工程学院,贵阳550025

出　　处：《小型微型计算机系统》2023年第7期1352-1359,共8页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(62063002)资助。

摘　　要：针对基本哈里斯鹰优化算法(Harris Hawks Optimization,HHO)存在易陷入局部最优值、收敛精度低和收敛速度慢的问题,提出混合差分进化以及二次插值法改进的哈里斯鹰算法.首先,在初始化种群阶段使用Sobol产生均匀随机数,增强种群遍历性及多样性;其次,在探索阶段融合差分进化算法加速HHO算法的寻优过程以及精度,同时提高全局搜索能力;最后,开发阶段使用二次插值法改进包围策略,提高算法局部搜索能力.同时改进算法中还加入了简单的精英制度,通过子代与父代适应度比较选出较优的一代作为下一种群.通过对测试函数集优化求解验证,表明改进算法具有较好的寻优能力及收敛速度,同时能有效避免陷入局部最优解.In view of the problem of the basic Harris Hawks Optimization algorithm,which is prone to local optimal value,low convergence accuracy and slow convergence speed,the hybrid differential evolution(DE)and quadratic interpolation improved Harris Hawks Optimization algorithm is proposed..Firstly,Sobol is used to produce uniform random numbers in the initial population stage to enhance population traversal and diversity,secondly,the fusion differential evolution algorithm accelerates the hysteresis process and precision of HO algorithm in the exploration stage,and improves the global search capability,and finally,the development stage uses the second interpolation method to improve the siege strategy and improve the algorithm local search capability.At the same time,the improved algorithm also adds a simple elite system,through the comparison of the adaptability of children and parents to select the better generation as the next group.Through the optimization and verification of the test function set,it is shown that the improved algorithm has better search ability and convergence speed,and can effectively avoid falling into local optimization.

关 键 词：哈里斯鹰算法 Sobol序列 差分进化 二次插值法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]