时序预测 | MATLAB实现基于PSO-LSTM、LSTM时间序列预测对比

效果一览

基本描述

MATLAB实现基于PSO-LSTM、LSTM时间序列预测。
1.Matlab实现PSO-LSTM和LSTM神经网络时间序列预测；
2.输入数据为单变量时间序列数据，即一维数据；
3.运行环境Matlab2020及以上,依次运行Main1LSTMTS、Main2PSOLSTMTS、Main3CDM即可，其余为函数文件无需运行,所有程序放在一个文件夹,data为数据集；
LSTM（长短时记忆模型）与粒子群算法优化后的LSTM（PSOLSTM）对比实验，可用于风电、光伏等负荷预测，时序预测，数据为单输入单输出，PSO优化超参数为隐含层1节点数、隐含层2节点数、最大迭代次数和学习率。

4.命令窗口输出MAE、MAPE、RMSE和R2；

程序设计

• 完整程序和数据下载：私信博主回复MATLAB实现基于PSO-LSTM、LSTM时间序列预测对比。

for i=1:PopNum%随机初始化速度,随机初始化位置for j=1:dimif j==dim% % 隐含层节点与训练次数是整数 学习率是浮点型pop(i,j)=(xmax(j)-xmin(j))*rand+xmin(j);elsepop(i,j)=round((xmax(j)-xmin(j))*rand+xmin(j));  %endend
end% calculate the fitness_value of Pop
pbest = pop;
gbest = zeros(1,dim);
data1 = zeros(Maxstep,PopNum,dim);
data2 = zeros(Maxstep,PopNum);
for i = 1:PopNumfit(i) = fitness(pop(i,:),p_train,t_train,p_test,t_test);f_pbest(i) = fit(i);
end
g = min(find(f_pbest == min(f_pbest(1:PopNum))));
gbest = pbest(g,:);
f_gbest = f_pbest(g);%-------- in the loop -------------
for step = 1:Maxstepmbest =sum(pbest(:))/PopNum;% linear weigh factorb = 1-step/Maxstep*0.5;data1(step,:,:) = pop;data2(step,:) = fit;for i = 1:PopNuma = rand(1,dim);u = rand(1,dim);p = a.*pbest(i,:)+(1-a).*gbest;pop(i,:) = p + b*abs(mbest-pop(i,:)).*...log(1./u).*(1-2*(u >= 0.5));% boundary detectionfor j=1:dimif j ==dimif pop(i,j)>xmax(j) | pop(i,j)<xmin(j)pop(i,j)=(xmax(j)-xmin(j))*rand+xmin(j);  %endelsepop(i,j)=round(pop(i,j));if pop(i,j)>xmax(j) | pop(i,j)<xmin(j)pop(i,j)=round((xmax(j)-xmin(j))*rand+xmin(j));  %endendendfit(i) = fitness(pop(i,:),p_train,t_train,p_test,t_test);if fit(i) < f_pbest(i)pbest(i,:) = pop(i,:);f_pbest(i) = fit(i);endif f_pbest(i) < f_gbestgbest = pbest(i,:);f_gbest = f_pbest(i);endendtrace(step)=f_gbest;step,f_gbest,gbestresult(step,:)=gbest;
end
or i=1:N%随机初始化速度,随机初始化位置for j=1:Dif j==D% % 隐含层节点与训练次数是整数 学习率是浮点型x(i,j)=(xmax(j)-xmin(j))*rand+xmin(j);elsex(i,j)=round((xmax(j)-xmin(j))*rand+xmin(j));  %endendv(i,:)=rand(1,D);
end%------先计算各个粒子的适应度，并初始化Pi和Pg----------------------
for i=1:Np(i)=fitness(x(i,:),p_train,t_train,p_test,t_test);y(i,:)=x(i,:);end
[fg,index]=min(p);
pg = x(index,:);             %Pg为全局最优%------进入主要循环，按照公式依次迭代------------for t=1:Mfor i=1:Nv(i,:)=w*v(i,:)+c1*rand*(y(i,:)-x(i,:))+c2*rand*(pg-x(i,:));x(i,:)=x(i,:)+v(i,:);for j=1:Dif j ~=Dx(i,j)=round(x(i,j));endif x(i,j)>xmax(j) | x(i,j)<xmin(j)if j==Dx(i,j)=(xmax(j)-xmin(j))*rand+xmin(j);  %elsex(i,j)=round((xmax(j)-xmin(j))*rand+xmin(j));  %endendendtemp=fitness(x(i,:),p_train,t_train,p_test,t_test);if temp<p(i)p(i)=temp;y(i,:)=x(i,:);endif p(i)<fgpg=y(i,:);fg=p(i);endendtrace(t)=fg;result(t,:)=pg;

参考资料

[1] https://blog.csdn.net/kjm13182345320/article/details/127596777?spm=1001.2014.3001.5501
[2] https://download.csdn.net/download/kjm13182345320/86830096?spm=1001.2014.3001.5501