clear
clc
close all
Function_name='F13'; % 测试函数编号（F1~F23）
[lb,ub,dim,fobj]=Get_Functions_details(Function_name); % 获取目标函数对应参数
SearchAgents_no=30; %种群规模
Max_iteration=500; %最大迭代次数
%% 模型训练
for i = 1:80 % 次数可以修改运行次数
disp(['第',num2str(i),'次实验']);
num1 = 5; % 头部混沌变异改进方案选择：1-10，tent、Logistic、Cubic、chebyshev、Piecewise、sinusoidal、Sine,ICMIC, Circle,Bernoulli
num2 = 3; % 身体融合变异改进方案选择：1-15，高斯变异，t分布扰动，自适应t分布，柯西变异扰动，差分变异扰动，高斯随机游走，莱维飞行,三角形游走,螺旋飞行，黄金正弦，正余弦,透镜成像反向学习,纵横交叉，动态反向学习，随机游走
num3 = 11; % 尾部拼接变异改进方法选择：
[Best_score(i,:),Best_pos(i,:),cg_curve(i,:)]=ESOA(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 改进后白鹭群优化算法
num1 = 0; % 无改进
num2 = 0; % 无改进
num3 = 0; % 无改进
[Best_score1(i,:),Best_pos1(i,:),cg_curve1(i,:)]=ESOA(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 白鹭群优化算法
[Best_score2(i,:),Best_pos2(i,:),cg_curve2(i,:)]=ALO(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 蚁狮优化算法
[Best_score3(i,:),Best_pos3(i,:),cg_curve3(i,:)]=PSO(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 粒子群优化算法
[Best_score4(i,:),Best_pos4(i,:),cg_curve4(i,:)]=HHO(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 哈里斯鹰优化算法
[Best_score5(i,:),Best_pos5(i,:),cg_curve5(i,:)]=GWO(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 灰狼优化算法
[Best_score6(i,:),Best_pos6(i,:),cg_curve6(i,:)]=HLOA(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 角蜥蜴优化算法
[Best_score7(i,:),Best_pos7(i,:),cg_curve7(i,:)]=GOOSE(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 鹅优化算法
end
%% 结果绘图
figure('Position', [500 500 350 500])
%绘制收敛曲线
CNT=20; % 绘制图形随机选点的数量(可自行设置)
k=round(linspace(1,Max_iteration,CNT)); %随机选CNT个点
iter=1:1:Max_iteration;
Cg_curve=mean(cg_curve);
Cg_curve1=mean(cg_curve1);
Cg_curve2=mean(cg_curve2);
Cg_curve3=mean(cg_curve3);
Cg_curve4=mean(cg_curve4);
Cg_curve5=mean(cg_curve5);
Cg_curve6=mean(cg_curve6);
Cg_curve7=mean(cg_curve7);
semilogy(iter(k),Cg_curve(k),'-*','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve1(k),'-p','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve2(k),'-d','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve3(k),'-s','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve4(k),'-v','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve5(k),'-+','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve6(k),'-h','LineWidth',1)
hold on
semilogy(iter(k),Cg_curve7(k),'-x','LineWidth',1)
legend('IESOA','ESOA','ALO','PSO','HHO','GWO','HLOA','GOOSE','Location','best')
grid off
xlabel('Number of iterations')
ylabel('Objective function value')
title('F1 Comparison of evolution curves of each optimization algorithm')
%% 输出结果
disp('======IESOA算法结果==========');
display(['IESOA算法100次实验最优适应度值(Best) : ', num2str(min(Best_score))]);
display(['IESOA算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score))]);
display(['IESOA算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score))]);
display(['IESOA算法100次实验标准差（std） : ', num2str(std(Best_score))]);
disp('======ESOA算法结果============');
display(['ESOA算法100次实验最优适应度值(Best) : ', num2str(min(Best_score1))]);
display(['ESOA算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score1))]);
display(['ESOA算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score1))]);
display(['ESOA算法100次实验标准差（std） : ', num2str(std(Best_score1))]);
disp('======ALO算法结果==========');
display(['ALO算法100次实验最优适应度值(Best) : ', num2str(min(Best_score2))]);
display(['ALO算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score2))]);
display(['ALO算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score2))]);
display(['ALO算法100次实验标准差（std） : ', num2str(std(Best_score2))]);
disp('======PSO算法结果==========');
display(['PSO算法100次实验最优适应度值(Best) : ', num2str(min(Best_score3))]);
display(['PSO算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score3))]);
display(['PSO算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score3))]);
display(['PSO算法100次实验标准差（std） : ', num2str(std(Best_score3))]);
disp('======HHO算法结果==========');
display(['HHO算法100次实验最优适应度值(Best) : ', num2str(min(Best_score4))]);
display(['HHO算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score4))]);
display(['HHO算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score4))]);
display(['HHO算法100次实验标准差（std） : ', num2str(std(Best_score4))]);
disp('======GWO算法结果==========');
display(['GWO算法100次实验最优适应度值(Best) : ', num2str(min(Best_score5))]);
display(['GWO算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score5))]);
display(['GWO算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score5))]);
display(['GWO算法100次实验标准差（std） : ', num2str(std(Best_score5))]);
disp('====HLOA算法结果==========');
display(['HLOA算法100次实验最优适应度值(Best) : ', num2str(min(Best_score6))]);
display(['HLOA算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score6))]);
display(['HLOA算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score6))]);
display(['HLOA算法100次实验标准差（std） : ', num2str(std(Best_score6))]);
disp('=====GOOSE算法结果==========');
display(['GOOSE算法100次实验最优适应度值(Best) : ', num2str(min(Best_score7))]);
display(['GOOSE算法100次实验最优解对应的平均适应度值(mean) : ', num2str(mean(Best_score7))]);
display(['GOOSE算法100次实验最差适应度值(wrost) : ', num2str(max(Best_score7))]);
display(['GOOSE算法100次实验标准差（std） : ', num2str(std(Best_score7))]);
以上为主函数 main_100
以下为原ESOA函数
function [y_global_best, x_global_best, Convergence_curve]=ESOA(num1,num2,num3,SearchAgents_no, Max_iter, lb, ub, dim, fobj)
%% 运行群智能
func = fobj;
beta1 = 0.9;
beta2 = 0.99;
%Initialize the positions of search agents初始化搜索代理位置
lb=lb.*ones(1,dim);
ub=ub.*ones(1,dim);
%% 改进方式1：头部变异|混沌初始化
x=repmat(lb, SearchAgents_no, 1)+chaos(num1,SearchAgents_no,dim) .* repmat((ub-lb), SearchAgents_no, 1);
Convergence_curve=zeros(1,Max_iter);
w = random('Uniform', -1, 1, SearchAgents_no, dim);
%g = random('Uniform', -1, 1, SearchAgents_no, dim);
m = zeros(SearchAgents_no, dim);
v = zeros(SearchAgents_no, dim);
y = zeros(SearchAgents_no,1);
for i=1:SearchAgents_no
y(i) = func(x(i,:));
end
p_y = y;
x_hist_best = x;
g_hist_best = x;
y_hist_best = ones(SearchAgents_no)*inf;
x_global_best = x(1, :);
g_global_best = zeros(1, dim);
y_global_best = func(x_global_best);
hop = ub - lb;
l=0;% Loop counter
%% 改进方式2：融合变异
% Main loop
while l<Max_iter
for i=1:SearchAgents_no
if rand < 1-sqrt(l/Max_iter) && num2 ~= 0 % 满足条件执行变异
x =integration(x,i,x_global_best,num2,l,Max_iter,lb,ub);
else
p_y(i) = sum(w(i, :) .* x(i, :));
p = p_y(i) - y(i);
g_temp = p.*x(i, :);
% Indivual Direction
p_d = x_hist_best(i, :) - x(i, :);
f_p_bias = y_hist_best(i) - y(i);
p_d = p_d .* f_p_bias;
p_d = p_d ./ ((sum(p_d)+eps).*(sum(p_d)+eps));
d_p = p_d + g_hist_best(i, :);
% Group Direction
c_d = x_global_best - x(i, :);
f_c_bias = y_global_best - y(i);
c_d = c_d .* f_c_bias;
c_d = c_d ./ ((sum(c_d)+eps).*(sum(c_d)+eps));
d_g = c_d + g_global_best;
% Gradient Estimation
r1 = rand(1, dim);
r2 = rand(1, dim);
g = (1 - r1 - r2).*g_temp + r1 .* d_p + r2 .* d_g;
g = g ./ (sum(g) + eps);
m(i,:) = beta1.*m(i,:)+(1-beta1).*g;
v(i,:) = beta2*v(i,:)+(1-beta2)*g.^2;
w(i,:) = w(i,:) - m(i,:)/(sqrt(v(i,:))+eps);
% Advice Forward
x_o = x(i, :) + exp(-l/(0.1*Max_iter)) * 0.1 .* hop .* g;
Flag4ub=x_o>ub;
Flag4lb=x_o<lb;
x_o = (x_o.*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;
y_o = func(x_o);
% Random Search
r = random('Uniform', -pi/2, pi/2, 1, dim);
x_n = x(i, :) + tan(r) .* hop/(1 + l) * 0.5;
Flag4ub=x_n>ub;
Flag4lb=x_n<lb;
x_n = (x_n.*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;
y_n = func(x_n);
% Encircling Mechanism
d = x_hist_best(i, :) - x(i, :);
d_g = x_global_best - x(i, :);
r1 = rand(1, dim);
r2 = rand(1, dim);
x_m = (1-r1-r2).*x(i, :) + r1.*d + r2.*d_g;
Flag4ub=x_m>ub;
Flag4lb=x_m<lb;
x_m = (x_m.*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;
y_m = func(x_m);
% Discriminant Condition
x_summary = [x_m; x_n; x_o];
y_summary = [y_m, y_n, y_o];
y_summary(isnan(y_summary)) = inf;
ind = y_summary==min(y_summary);
y_i = min(y_summary);
x_i = x_summary(ind, :);
x_i = x_i(1, :);
if y_i < y(i)
y(i) = y_i;
x(i, :) = x_i;
if y_i < y_hist_best(i)
y_hist_best(i) = y_i;
x_hist_best(i, :) = x_i;
g_hist_best(i, :) = g_temp;
if y_i < y_global_best
y_global_best = y_i;
x_global_best = x_i;
g_global_best = g_temp;
end
end
else
if rand()<0.3
y(i) = y_i;
x(i, :) = x_i;
end
end
end
end
%% 改进方式3：尾部变异
if num3 ~= 0
for i=1:size(x,1)
NEW_pos =integration(x,i,x_global_best,num3,l,Max_iter,lb,ub);
Flag4ub=NEW_pos(i,:)>ub;
Flag4lb=NEW_pos(i,:)<lb;
NEW_pos(i,:)=(NEW_pos(i,:).*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;
fitness= fobj(NEW_pos(i,:));
if(y(i)>fitness)
y(i)=fitness;
x(i,:) = NEW_pos(i,:);
end
if(y_global_best>fitness)
y_global_best=fitness;
x(i,:) = NEW_pos(i,:);
x_global_best=x(i,:);
end
end
end
l=l+1;
% fprintf("%d, %f\n", l, y_global_best)
Convergence_curve(l) = y_global_best;
end
end
现在这个函数运行后显示
random 需要以下项之一:
SimBiology
Statistics and Machine Learning Toolbox
出错 ESOA (第 21 行)
w = random('Uniform', -1, 1, SearchAgents_no, dim);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
出错 main_100 (第 16 行)
[Best_score(i,:),Best_pos(i,:),cg_curve(i,:)]=ESOA(num1,num2,num3,SearchAgents_no,Max_iteration,lb,ub,dim,fobj); % 改进后白鹭群优化算法
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
请对主函数进行修改，以保证正常运行
本文介绍了融合黄金正弦函数和随机游走策略的哈里斯鹰优化算法，该算法结合鹰群捕食行为，通过MATLAB实现，增强了全局搜索能力和收敛速度。代码示例中，定义了优化问题结构体，初始化鹰群，迭代优化，更新个体位置，利用黄金正弦权重和随机游走更新策略，并进行边界处理，以提高优化效果。
订阅专栏 解锁全文
扫一扫
3022
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
