求解全局优化问题的多策略改进灰狼算法  

Multi-strategy improved grey wolf algorithm for solving the global optimization problem

作　　者：陈敏 陈晔 牛兴龙 武志博 Chen Min;Chen Ye;Niu Xinglong;Wu Zhibo(School of Electrical and Control Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学电气与控制工程学院,太原030051

出　　处：《国外电子测量技术》2022年第11期22-29,共8页Foreign Electronic Measurement Technology

基　　金：山西省高等学校科技创新项目(18005548);住房和城乡建设部科技技术计划(2021-k-136)项目资助。

摘　　要：针对原始灰狼算法在求解全局优化问题时,存在后期收敛速度慢、易陷入局部最优、优化效率低等缺点,提出一种多策略改进的灰狼优化算法(multi-strategy improved grey wolf optimization, MSI_GWO)。从参数、搜索机制以及最优解扰动3个方面引入3种改进策略。对控制参数a,采用非线性调整策略,改进算法的勘探和开发能力;采用动态权重策略进行位置更新,提高算法的收敛性;设计小波最优解扰动策略,提高种群多样性,同时避免算法陷入局部最优。对MSI_GWO算法的寻优性能进行验证,选取单峰多峰共9个测试函数完成仿真实验,并与其他改进灰狼优化算法以及粒子群、天鹰优化算法、蜜獾算法进行比较,结果表明,MSI_GWO算法在收敛速度和寻优效率方面均为最优的。Aiming at the shortcomings of the original grey wolf algorithm in solving global optimization problems, such as slow convergence, easy local optimum and low optimization efficiency, a multi-strategy improved grey wolf optimization algorithm(multi-strategy improved grey wolf optimization, MSI_GWO) is proposed. Three improvement strategies are introduced from parameter, search mechanism and optimal solution disturbance: nonlinear adjustment strategy for control parameter a to improve the exploration and development capabilities of the algorithm;dynamic weight strategy to update position to improve the convergence of the algorithm;wavelet optimal solution disturbance strategy to improve population diversity and avoid the algorithm from local optimum. The optimal performance of MSI_GWO algorithm is verified, and nine test functions are selected to complete the simulation experiment, comparing it with other improved grey wolf optimization algorithms particle swarm optimization, aquila optimizer and honey badger algorithm. The results show that MSI_GWO algorithm is optimal in terms of convergence speed and optimal efficiency.

关 键 词：改进灰狼算法 控制参数 动态权重 小波函数 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]