神經網絡辨識系統

clc;close all;clear;

y(1)= 0;
k=1;
u(k)= sin(2*pi*k/250);
f(k)= 0.6*sin(pi*u(k))+0.3*sin(3*pi*u(k))+0.1*sin(5*pi*u(k));
y(k+1)= 0.3*y(k)+f(k);

for k=2:100
    u(k)= sin(2*pi*k/250);
    f(k)= 0.6*sin(pi*u(k))+0.3*sin(3*pi*u(k))+0.1*sin(5*pi*u(k));
    y(k+1)= 0.3*y(k)+0.6*y(k-1)+f(k); 
end

P= [u(2:100);y(2:100);y(1:99)];
T= [y(3:101)];
net= newff(minmax(P),[20,20,1],{'tansig','tansig','purelin'},'trainlm','learngdm','msereg');
[net,tr] = train(net,P,T);

Q= [u(2:100);y(2:100);y(1:99)];
nety= sim(net,Q);
plot(1:99,y(1:99),'b',3:101,nety(1:99),'r');
legend('系統輸出','系統辨識');


% clc;close all;clear;
% 
% y(1)= 0;
% 
% for k=1:100
%     u(k)= sin(2*pi*k/25)+sin(2*pi*k/10);
%     f(k)= u(k).^3;
%     y(k+1)= y(k)/(1+y(k).^2)+f(k);
% end
% 
% P= [u(1:100);y(1:100)];
% T= [y(2:101)];
% net= newff(minmax(P),[20,20,1],{'tansig','tansig','purelin'},'trainlm','learngdm','msereg');
% [net,tr] = train(net,P,T);
% 
% % z= zeros(1,100);
% Q= [u(1:100);y(1:100)];
% nety= sim(net,Q);
% figure;
% plot(1:100,y(1:100),'b',2:101,nety(1:100),'r');
% legend('系統輸出','系統辨識');