一种求解高维优化问题的改进灰狼算法  

Improved grey wolf algorithm to solve high-dimensional optimization problems

作　　者：李煜[1] 林笑笑 刘景森[3] Li Yu;Lin Xiaoxiao;Liu Jingsen(Research Institute of Management Science and Engineering,Henan University,Kaifeng 475004,China;School of Business,Henan University,Kaifeng 475004,China;Institute of Intelligent Networks System,Henan University,Kaifeng 475004,China)

机构地区：[1]河南大学管理科学与工程研究所,河南开封475004 [2]河南大学商学院,河南开封475004 [3]河南大学智能网络系统研究所,河南开封475004

出　　处：《系统工程学报》2024年第2期200-216,共17页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(72104069);河南省重点研发与推广专项资助项目(222102210065);教育部青年基金人文社会科学基金资助项目(15YJC630079).

摘　　要：为求解高维优化问题,提出基于反向学习和衰减因子的灰狼优化算法(grey wolf algorithm based on opposition learning and reduction factor,ORGWO).设计一种灰狼反向学习模型,模型考虑问题搜索边界信息和种群历史搜索信息,初始种群阶段增加反向学习,增强种群多样性.根据算法各个阶段不同特征引入衰减因子,平衡全局和局部勘探能力.选取8个高维函数和23个不同特征的优化函数对算法性能进行测试,进一步使用收敛性分析,寻优成功率,CPU时间,Wilcoxon秩和检验来评估改进算法,实验结果表明,ORGWO算法在求解高维问题上具有较好的精度,鲁棒性和更快的收敛速度.This paper proposes a grey wolf algorithm based on opposition-based learning and an attenuation factor to solve high dimensional optimization problems(ORGWO).An opposition-based learning model is designed,considering the boundary search information and the population historical search information.In the initial population stage,an opposition-based learning is added,which enhances the diversity of the pop-ulation.An attenuation factor is introduced to balance the global and local exploration capabilities.Eight high-dimensional test functions and twenty-three benchmark functions with different characteristics are used to test the performance of the algorithm.The convergence analysis,success rate,CPU time,and Wilcoxon rank sum test are used to evaluate the ORGWO.The experimental results show that the ORGWO has better precision,robustness and faster convergence speed in solving high dimensional problems.

关 键 词：灰狼优化算法 反向学习 衰减因子 高维优化问题 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]