Scribd
Upload
21 views

Code Mat

sdf

Uploaded by

nsh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
21 views

Code Mat

sdf

Uploaded by

nsh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 2

clear all

close all
clc
V=[1 1 1 1 1];
EP=[];

for i=1:31
PRBS(i,1)=V(5);
EP=[EP;V(5);V(5);V(5);V(5)];
f=V(3)V(5)(-1);
V(5)=V(4);
V(4)=V(3);
V(3)=V(2);
V(2)=V(1);
V(1)=f;
end
figure(1)
plot(PRBS,'o')
figure(2)
plot(EP,'*')

%y(i)=1.5*y(i-1)-0.75*y(i-2)+1*EP(i-1)+0.25*EP(i-2)
y(1,1)=0;
y(2,1)=1*EP(1);
for i=3:124
y(i,1)=1.5*y(i-1)-0.75*y(i-2)+1*EP(i-1)+0.25*EP(i-2);
end;
figure(3)
plot(y,'r')
title('ideal output')

%calculte a1,a2,b1 and b2 using LSE method, define Y and X==>


%theta=inv(X'*X)*X'*Y
Y=y(3:124);
X=[y(2:123) y(1:122) EP(2:123) EP(1:122)];
theta=inv(X'*X)*X'*Y;

%output with noise


yn=y+randn(124,1);
figure(4)
plot(1:124,y,'r',1:124,yn,'k')
legend('y','yn')

%LSE method with noise


Yn=yn(3:124);
Xn=[yn(2:123) yn(1:122) EP(2:123) EP(1:122)];
thetan =inv(Xn'*Xn)*Xn'*Yn;

%test of the identified model

yst(1,1)=0;
yst(2,1)=1*1;
for i=3:100
yst(i,1)=1.5*yst(i-1)-0.75*yst(i-2)+1*1+0.25*1;
end;
ystLSE(1,1)=0;
ystLSE(2,1)=thetan(3)*1;
for i=3:100
ystLSE(i,1)=thetan(1)*yst(i-1)+thetan(2)*yst(i-2)+thetan(3)*1+thetan(4)*1
end;
legend('yst','ystLSE');
uGLS=EP; yGLS=yn; precision=1e-4; e11=1000; e22=1;
while abs(e11-e22)>precision
%calculate thetaGLS
YGLS=yGLS(3:124);
XGLS=[yGLS(2:123) yGLS(1:122) uGLS(2:123) uGLS(1:122)];
thetaGLS=inv(XGLS'*XGLS)*XGLS'*YGLS;
%estimate the residu
e=YGLS-XGLS*thetaGLS; %e:122x1
%e(i)+f1*e(i-1)=xi(i)==>e(i)=-f1e(i-1)+xi(i)
E1=e(2:122);
Xe=-e(1:121);
f=inv(Xe'*Xe)*Xe'*E1;
%update U and Y
ys(i,1)=yGLS(i);
us(i,1)=uGLS(i);
for i=2:124
ys(i,1)=yGLS(i)+f*yGLS(i-1);
us(i,1)=uGLS(i)+f*uGLS(i-1);
end;
yGLS=ys; uGLS=us; e11=e22; e22=e'*e;
end;

ystGLS(1,1)=0;
ystGLS(2,1)=thetaGLS(3)*1;
for i=3:100
ystGLS(i,1)=thetaGLS(1)*yst(i-1)+thetaGLS(2)*yst(i-2)+thetaGLS(3)*1+
(thetaGLS(4))*1;
end;

%gradient methode
yGRAD=yn; uGRAD=EP; parGRAD=thetan; epsGRAD=1e-5; alpha=1e-5; eGRAD=10;
while eGRAD>epsGRAD
grad=[0;0;0;0];
ymGRAD(1,1)=0;
ymGRAD(2,1)=parGRAD(3)*uGRAD(1);
for i=3:124
ymGRAD(i,1)=parGRAD(1)*ymGRAD(i-1,1)+parGRAD(2)*ymGRAD(i-
2,1)+parGRAD(3)*uGRAD(i-1,1)+parGRAD(4)*uGRAD(i-2,1);
sensib=2*[-ymGRAD(i-1,1); -ymGRAD(i-2,1); -uGRAD(i-1,1); -uGRAD(i-2,1)];
end;
parGRADkp1=parGRAD-alpha*grad;
eGRAD=abs(max(parGRADkp1-parGRAD));
parGRAD=parGRADkp1;
end;
parGRAD

y_gr(1,1)=0;
y_gr(2,1)=1;
for i=3:100
y_gr(i,1)=parGRAD(1)*yst(i-1)+parGRAD(2)*yst(i-2)+parGRAD(3)*1+parGRAD(4)*1;

end;
figure(5)
plot(1:100,yst,'r',1:100,ystLSE,'k',1:100,ystGLS,'b',1:100,y_gr,'g')
legend('process','LSE','GLS','gradient')

You might also like