[System Identification] Estimate and validate transfer function

Hi!

There are 4 sets of data in the attached mat file:

• Data_estimate_dry
• Data_validate_dry
• Data_estimate_wet
• Data_validate_wet

Note that for the wet, the output data (y1) decreases at a much faster rate than the dry.

I used the following to estimate the transfer function based on dry set of data:

Opt = tfestOptions('Display','on');

mtf = tfest(Data_estimate_dry, 2,2,13,Opt);

Used another set to check the fit:

figure, compare(Data_validate_dry,mtf)

The resulting fit looks good with 88% fit.

Then I used wet data to run it through the estimated transfer function:

figure, compare(Data_validate_wet,mtf)

As expected, the fit is not good. However, during the decay phase, I expected to see a big different between simulated data and the wet data.

Is this due to different input signals for wet and dry data?

[Mathieu NOE]

mayba, have you tried to compare both tf between dry and wet data ?