1 简介
本文提出了一种基于自然启发的群体求解全局优化问题的元启发式算法——金鹰优化器( Golden Eagle Optimizationr,GEO )。GEO的核心灵感是金鹰在其螺旋轨迹的不同阶段调谐速度的智能,用于狩猎。它们在狩猎的初始阶段表现出更多的巡游和寻找猎物的倾向,在最后阶段表现出更多的攻击倾向。一只金鹰调整这两个组件,以在可行的区域以最短的时间捕获最好的猎物。这种行为通过数学建模来突出对全局优化方法的探索和利用。利用33个基准测试函数和一个可扩展性测试对本文算法的性能进行了测试和确认。将结果与其他6种著名算法进行了比较,揭示了GEO的优越性,表明它能有效地找到全局最优,避免局部最优。多目标金鹰优化器( MOGEO )也被提出用于求解多目标问题。MOGEO的性能也在10个多目标基准函数上进行了测试和验证。并与其他两种多目标算法的结果进行了比较,结果表明能更好地逼近真实Pareto最优解。
2 部分代码
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Multi-Objective Golden Eagle Optimizer (MOGEO) source codes version 1.0 % % Original paper:Abdolkarim Mohammadi-Balani, Mahmoud Dehghan Nayeri, %Adel Azar, Mohammadreza Taghizadeh-Yazdi, %G5olden Eagle Optimizer: A nature-inspired %metaheuristic algorithm, Computers & Industrial Engineering. % To use this code in your own project % remove the line for 'GetFunctionDetails' function % and define the following parameters: % fun : function handle to the .m file containing the objective function % the .m file you define should accept 'x' as input and return % a column vector containing objective function values % nobj : number of objectives % nvars : number of decision/design variables % lb : lower bound of decision variables (must be of size 1 x nvars) % ub : upper bound of decision variables (must be of size 1 x nvars) % % MOGEO will return the following: % x : best solution found % fval : objective function value of the found solution % % You can alternatively download the GEO toolbox from: % %% Inputs FunctionNumber = 7; % 1-10 options.PopulationSize = 200; options.ArchiveSize = 100; options.MaxIterations = 1000; options.FunctionNumber = FunctionNumber; %% Run Multi-Objective Golden Eagle Optimizer [fun,nobj,nvars,lb,ub] = GetFunctionDetails (FunctionNumber); options.AttackPropensity = [0.5 , 2]; options.CruisePropensity = [1 , 0.5]; [x,fval] = MOGEO (fun,nobj,nvars,lb,ub, options); img =gcf; %获取当前画图的句柄 print(img, '-dpng', '-r600', './运行结果.png') %即可得到对应格式和期望dpi的图像
3 仿真结果
4 参考文献
[1]Mohammadi-Balani Abdolkarim et al. Golden eagle optimizer: A nature-inspired metaheuristic algorithm[J]. Computers & Industrial Engineering, 2021, 152