Proper Use of System Identification Toolbox

Hi Guys,

Fundamental Science Question: I can see that people use the System Identification Toolbox to produce models using experimental (or generally 'real world') data (in the form of time series)to produce a model for control studies.

Can someone use a time series that is produced from a simulation in order to produce a model with System Identification Toolbox? Suppose that no experimental or real world data is available. Does this make sense at all?

Background: I have a model which is a DAE system (pure-no transformation whatsoever) and I want to study the control part a bit. The system is very difficult to get it transformed with Laplace etc and get a transfer function out of it. Hence I thought that I could use the System Identification toolbox for this job.

Thanks,

Warm regards,

Chris

Answer 1

[Arkadiy Turevskiy]

Yes, it makes a perfect sense. People use system identification techniques to get easier/more computationally efficient models from complex models - this is a common use case.

Take a look at these two examples: this one and this one .

So to answer your question - yes, absolutely, you can use System Identification Toolbox to create a "simpler"model to approximate your complex one.

[Christos]

Thank you very much Arkadiy. That's all I wanted to hear :).