Question about estimating an LTI system matrices using system identification toolbox.

Hi !

I would like to ask about the the system idintification toolbox.

In my problem I used linear parameter varying model to generate data.

so that I have vector of inputs and outputs.

Then I wanted to build a linear time invatiant system like:

So I used the system identification toolbox, however all the fit estimation error comes out as negative numbers (- 11% , -3, ...)

I use state space model to estimate these matrices with model order=2

is there any other way to estimate A,B, C using the input and output vectors? that might lead to better fit to estimation percentage?

Thanks.