sss.asv

神经网络源代码

p=[0.3   1.0   4     20    20    45     12    23     6.5    1.6   0.5   0.5    1.65  0.375 ;
   0.923 0.934 0.924 0.916 0.927 0.9555 0.952 0.9575 0.9585 0.948 0.948 0.9445 0.943 0.9455]';
t=[15.17 19.17 13.10 8.83 15.10  32  30  32  33  30 30 26 24 24;
   1.7   1.4   1.4   1.5  1.1    2.0 5.0 3.0 4.5 8  20 30 10 12]'; 
u=t;
%训练样本归一化
for i=1:14
    P(i,:)=(p(i,:)-min(p(i,:)))/(max(p(i,:))-min(p(i,:)));
    
end
for i=1:14
    T(i,:)=(t(i,:)-min(t(i,:)))/(max(t(i,:))-min(t(i,:)));
end
threshold=[0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1;0 1];
net=newff(threshold,[29,14],{'tansig','logsig'},'trainlm');
net.trainParam.epochs=1000;
net.trainParam.goal=0.0001;
LP.lr=0.1;
%net.trainParam.show=20;
net=init(net);
net=train(net,P,T);
Y=sim(net,P);
error=T-Y;
%反归一化
predict=zeros(14,2);
for i=1:14
   predict(i,1)=Y(i,1)* (max(u(i,:))-min(u(i,:)))+ min(u(i,:));

   predict(i,2)=Y(i,2)* (max(u(i,:))-min(u(i,:)))+ min(u(i,:));
end
%测试样本归一化
p_test=[2.0   20    27     1.6   1.6;
        0.919 0.921 0.9515 0.943 0.947]';
t_test=[12.55 11.58 29  26 29;
        1.6   1.4   2.8 10 10]';
for i=1:5
    P_test(i,:)=(p_test(i,:)-min(p_test(i,:)))/(max(p_test(i,:))-min(p_test(i,:)));
    
end
for i=1:5
    T_test(i,:)=(t_test(i,:)-min(t_test(i,:)))/(max(t_test(i,:))-min(t_test(i,:)));
end
%预测    
out=sim(net,P_test);