SA-IPSO.zip

% ----------------------- README ------------------------------------------ % ------------------- 原创声明 ------------------------------------------- % -------------- 最后一次修改：2023/7/10 ----------------------------------- % ------------------- 欢迎关注₍^.^₎♡ ------------------------------------- % -------------- 项目：模拟退火-粒子群自适应优化算法应用于函数寻优 ------------ % -------------- 微信公众号:KAU的云实验台 ---------------------------------- % -------------- CSDN:KAU的云实验台 --------------------------------------- % ------------------------------------------------------------------------- %% 自适应改进策略： % 1 非线性修改惯性权重，根据全局最优点的距离调整 % 2 混沌映射初始化粒子-Logistic映射 % 3 学习因子异步动态变化 %% 适应度越小越好 %% 清空环境 clc clear close all fun = @Rosenbrock_;%Sphere_ Griewank_ Rastrigin_ Rosenbrock_ %种群 % SP % popmax=100;popmin=-100; % GR(这个函数需把粒子的Vmax Vmin改一下 因为范围太大) % popmax=600;popmin=-600; % RA % popmax=5.12;popmin=-5.12; % RO popmax=30;popmin=-30; %% 参数初始化 %粒子群算法中的两个参数 maxgen=300; % 进化次数 sizepop=100; %种群规模 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 动态学习因子 c1fun = @(t) 2*sin(pi/2*(1-t/maxgen))^2; c2fun = @(t) 2*sin(pi*t/(2*maxgen))^2; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 速度上下限 Vmax=0.5; Vmin=-0.5; % 变量维度 D=2; % 线性递减惯性权重 wmax=0.9; wmin=0.4; % 模拟退火参数 lamda = 0.8;% 退火系数 %% 产生初始粒子和速度 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进2 %%%%%%%%%%%%%%%%%%%%%%%%%%% % 混沌初始化 Z = zeros(sizepop, D); % 随机生成一个d维向量 Z(1, :) = rand(1, D); % 利用logistic生成sizepop个向量 for i=2:sizepop Z(i,:) = 4.0*Z(i-1,:).*(1-Z(i-1,:)); end % 将z的各个分量载波到对应变量的取值区间 pop = zeros(sizepop, D); for i=1:sizepop pop(i,:) = popmin + (popmax - popmin).*Z(i,:); V(i,:)=(Vmax-Vmin).*Z(i,:)+Vmin; %初始化速度 %计算适应度 fitness(i)=fun(pop(i,:)); %染色体的适应度 end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进2 %%%%%%%%%%%%%%%%%%%%%%%%%%% %% 个体极值和群体极值 [bestfitness bestindex]=min(fitness); zbest=pop(bestindex,:); %全局最佳 gbest=pop; %个体最佳 fitnessgbest=fitness; %个体最佳适应度值 fitnesszbest=bestfitness; %全局最佳适应度值 favg = sum(fitness)/sizepop; fmax = max(fitness); % 退火初温 T=-fitnesszbest/log(0.2); %% 迭代寻优 for i=1:maxgen c1 = c1fun(i); c2 = c2fun(i); % 计算粒子的退火算法适应度 for ii=1:sizepop Tfit(ii) = exp( - (fitness(ii) - fitnesszbest)/T); end SumTfit = sum(Tfit); Tfit = Tfit/SumTfit; pBet = rand(); % 将满足条件的第iii个粒子代替全局最优 for iii=1:sizepop ComFit(iii) = sum(Tfit(1:iii)); if pBet <= ComFit(iii) zbest_plus = pop(iii,:); break; end end % 更新粒子速度 for j=1:sizepop %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进3 %%%%%%%%%%%%%%%%%%%%%%%%% % 惯性权重非线性递减 if fitness(j) <= favg % 若适应度优于平均值 惯性权重小 w = wmin + (fmax-fitness(j))*(wmax-wmin)/(fmax-favg); else w = wmax; end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 改进3 %%%%%%%%%%%%%%%%%%%%%%%%% %速度更新 V(j,:) = w*V(j,:) + c1*rand*(gbest(j,:) - pop(j,:)) + c2*rand*(zbest_plus - pop(j,:)); V(j,find(V(j,:)>Vmax))=Vmax; V(j,find(V(j,:)<Vmin))=Vmin; %位置更新 pop(j,:)=pop(j,:)+V(j,:); pop(j,find(pop(j,:)>popmax))=popmax; pop(j,find(pop(j,:)<popmin))=popmin; %适应度值 fitness(j)=fun(pop(j,:)); end % 适应度，平均适应度，最劣 favg = sum(fitness)/sizepop; fmax = max(fitness); % 更新种群 for j=1:sizepop %个体最优更新 if fitness(j) < fitnessgbest(j) gbest(j,:) = pop(j,:); fitnessgbest(j) = fitness(j); end %群体最优更新 if fitness(j) < fitnesszbest zbest = pop(j,:); fitnesszbest = fitness(j); end end T = T*lamda; yy1(i)=fitnesszbest; end fitnesszbest1 = fitnesszbest; zbest1 = zbest; %% 结果分析 save SA-IPSO.mat yy1 zbest1 fitnesszbest1 plot(yy1,'Color',[021 151 165]/255,'LineWidth',3) title('SA-IPSO最优个体适应度','fontsize',12); xlabel('进化代数','fontsize',12);ylabel('适应度','fontsize',12); disp('*************************************************'); disp('目标函数取最小值时的自变量：'); zbest disp('目标函数的最小值为：') fitnesszbest disp('**************************************************');