2.rar_BP神经网络_BP预测_bp 预测_短期预测_网络 电力系统

clc;clear ;close all load datafile x1=data(:,1); x2=data(:,2); x3=data(:,3); x4=data(:,4); y1=data(:,5); %%%数据样本181组:length(x1)=length(x2)=length(x3)=length(x4)=181; meanx1=sum(x1)/length(x1); meanx2=sum(x2)/length(x2); meanx3=sum(x3)/length(x3); meanx4=sum(x4)/length(x4); %%%数据处理——归一化： for i=1:181 x1Norm(i)=(x1(i,1)-meanx1)/std(x1); x2Norm(i)=(x2(i,1)-meanx2)/std(x2); x3Norm(i)=(x3(i,1)-meanx3)/std(x3); x4Norm(i)=(x4(i,1)-meanx4)/std(x4); end %前150作为训练样本，训练神经网络 for m1=1:150 Intput(1,m1)=x1Norm(m1); end for m2=1:150 Intput(2,m2)=x2Norm(m2); end for m3=1:150 Intput(3,m3)=x3Norm(m3); end for m4=1:150 Intput(4,m4)=x4Norm(m4); end for m5=1:150 Tempout(m5)=y1(m5); end %%后31个样本用来测试神经网络 k=1; for n1=151:181 Testinput(1,k)=x1Norm(n1); k=k+1; end k=1; for n2=151:181 Testinput(2,k)=x2Norm(n2); k=k+1; end k=1; for n3=151:181 Testinput(3,k)=x3Norm(n3); k=k+1; end k=1; for n4=151:181 Testinput(4,k)=x4Norm(n4); k=k+1; end k=1; for n5=151:181 Testoutput(k)=y1(n5); k=k+1; end %%%训练样本：输入：Intput；输出：Tempout %%%测试样本：输入：Testinput；输出：Testoutput tic [input,minp,maxp,output,mint,maxt]=premnmx(Intput,Tempout); net=newff(minmax(Intput),[9,1],{'tansig','purelin'},'traingdm'); inputweights=net.IW{1,1}; inputbias=net.b{1}; layerweights=net.LW{2,1}; layerbias=net.b{2}; net.trainParam.epochs=5000; net.trainParam.goal=0.01; net.trainParam.show=50; net.trainParam.lr=0.1; net.trainParam.mc=0.9; net=train(net,input,output); out=sim(net,Testinput); yuce= postmnmx(out,mint,maxt);%help " postmnmx" e=Testoutput-yuce; MSE=mse(e); %Testoutput(1)=yuce(1); %%%画图 figure; subplot(2,1,1); plot((1:31),Testoutput,'-*',(1:31),yuce,'-o'); ylabel('电力负荷'); title('bp负荷预测'); legend('实际负荷','预测负荷'); subplot(2,1,2); erate=(Testoutput-yuce)./Testoutput; stem(erate); ylabel('相对误差值'); title('预测相对误差'); toc