Sunday, May 11, 2008

Systems Identification - Frequency response method

I am improving a matlab script I wrote to identify non linear systems using the frequency response method. I wrote a simple script but I think it is using way too much resources for something that isn't that complicated.

I need to learn more about the FFT. from what ive learned so far, when FFTing a signal, the higher the sample rate, the better the resolution in frequency (makes sense, given the shanon/nyquist theorem) and the greater the number of periods the more accurate is the result.

The system I am trying to identify, supposably, has two complex poles at 0.25 +/- 0.9 so I need good frequency resolution between 10^(-1) and 1 rad/s

I used a 1kHz sample rate and this is the result I get:



When I change the sample rate to 10kHz this is what I get:
Posted by Donato Azevedo at 7:30 PM

1 comments:

Gabe said...

I work in mechanism controls and I've written a few system identification scripts for simple 2nd order systems but nothing larger than that - at least not yet.

Does your frequency domain system identification tool handle arbitrary systems or is there a limit like my tools for 2nd order systems?

I have a wiki at http://wikis.ControlTheoryPro.com and if you have something you are willing to share I'd like to get it up on the site.

Thanks
Gabe

2:19 PM

Post a Comment

Subscribe to: Post Comments (Atom)