Why the order of the system is reduced while using PEM in System Identification?

1 view (last 30 days)
Hi All,
I used the following code for System Identification using PEM and iddata. But the initial system used to generate the step response was a 4th order system and the system identified using the following commands gave a 3rd order system. Why does this disparity happen?
Regards, Raghu
clc
clear all
close all
A=[-0.6129 0 .0645 0 ; 0 .7978 0 -.4494; -6.4516 0 -.7419 0;0 0.4494 0 .8876];
B=[.0323;.8989;.1290;.2247];
C=[.96774 .1124 1.6129 .4719];
D=0;
% statespace model
sys1=ss(A,B,C,D);
sys5=c2d(sys1,.001,'zoh')
step(sys1);
%figure;
resp=step(sys5);
[m,n]=size(resp);
k=10000/2;
% dividing data into two datasets
for j=1:k
sys2(j,1)=resp(j,1);
sys3(j,1)=resp(k,1);
k=k+1;
end
% modelling system
u=ones((m-1)/2,1);
sys4=iddata(sys2,u,.001);
sysid=pem(sys4);
% system parameters from the model
figure,step(sysid)
[ad bd cd dd]=ssdata(sysid)
ssfinal=ss(ad,bd,cd,dd,.001)
figure,plot(sysid)
figure,plot(sys3)

Answers (1)

Arkadiy Turevskiy
Arkadiy Turevskiy on 27 Mar 2014
If you follow the process described here

Categories

Find more on Linear Model Identification in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!