基于佳点集和莱维飞行原理的蜉蝣优化算法  

Mayfly optimization algorithm based on good-point set and Lévy flight principle

作　　者：吴霄 江海新[1] 吴芸[1] 吴雪颜 江佳玉 童林 WU Xiao;JIANG Haixin;WU Yun;WU Xueyan;JIANG Jiayu;TONG Lin(School of Science,Jiujiang University,Jiujiang 332005,China)

机构地区：[1]九江学院理学院,江西九江332005

出　　处：《高师理科学刊》2022年第3期36-41,51,共7页Journal of Science of Teachers＇College and University

基　　金：江西省教育厅科技项目(GJJ201814,GJJ211823,GJJ211825);江西省2021年大学生创新创业训练计划项目(s202111843031,s202111843039);九江学院2021年大学生创新创业训练计划项目(x202111843141,x202111843144,x202111843150)。

摘　　要：针对蜉蝣优化算法(Mayfly Algorithm,MA)在求解多峰函数时容易发生早熟收敛而陷入局部极值的缺陷,提出了一种基于佳点集和莱维飞行原理的蜉蝣优化算法(JLMA).该算法首先采用佳点集初始化替代随机初始化,以增强初始种群的遍历性;然后修改了雄性蜉蝣的位置更新公式,消除了速度项对收敛速度的影响;最后,利用莱维飞行改变雄性蜉蝣位置的移动方向,防止算法陷入局部最优.在12个测试函数上的实验结果表明,JLMA算法能跳出局部最优,提高解的精度,寻优效果更好.To address the shortcoming that the mayfly algorithm(MA)is prone to premature convergence and falling into local stagnation when solving multimodal functions,proposed a mayfly algorithm based on good-point and Lévy flight(JLMA).The algorithm first adopts good-point set initialization instead of random initialization to increase the traversal of the initial population.Then modifies the position update formula of male mayflies to eliminate the influence of the velocity term on the convergence speed.And finally uses Lévy flight to change the moving direction of male mayfly positions to prevent the algorithm from falling into local optimum.Our experiments on 12 test functions show that the JLMA algorithm can jump out of the local optimum,improve the accuracy of the solution,and have a better result in finding the optimum.

关 键 词：蜉蝣优化算法 佳点集 莱维飞行 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]