familiar with system identification toolbox?

I am using the system identification toolbox to find out the transfer function of a set of measured data. I use the simple example as below.but what I get as the transfer function in this toolbox is completely different from my TF. does anyone know what happens to the data and how exactly this toolbox works?

if true
  T1=1;
Ks=1;
s=tf('s');
sys= Ks/(1+T1*s);
[y t]=step(sys);
plot(t,y)

end

after this code,import t and y as the input and output data in ident tool and see what happens in TF estimation!

Answer 1

[Arkadiy Turevskiy]

a little bit familiar:)

System Identification Toolbox lets you identify transfer functions and other models from input-output test data. "Input data" means the excitation signal you apply to the system, not the time vector you use. Step response by definition means system response to step input, so the input signal is 0 at time 0 and then is equal to 1.

So to get the results you expect:

u=ones(size(t));

Then import u and y (not t and y)as input and output data, starting time:0, sampling time: t(2)-t(1).

Then estimate a transfer function, and you will get your 1/(s+1).

[sellyEpp]

perfect! Thanks!

[sellyEpp]

How is the Validation data used? does it have a significant effect on the results?if you could explain it with the same example, that would be great!

[Arkadiy Turevskiy]

Validation data is used to make sure the model you get is good. If you use just one data set, you will always be able to find a model that gives you a perfect fit for that particular dataset. That's where the validation dataset comes in - you test your model on it to see if this still gives you good results.