Thread Subject: SystemID: Preparation of nonuniformly-sampled data

Hi,

I have a project going on, where have two log files one containing
input, the other one the output. Both are not uniformly sampled, which
would not yet be too bad, as this is supposed to work with an iddata
object. But my problem is, that the sampling instants of the two data
files are not equal. So I can't properly choose in- and outputs.

Any suggestions?

Thanks,
Simo

Simo Kauth schrieb:
> Hi,
>
> I have a project going on, where have two log files one containing
> input, the other one the output. Both are not uniformly sampled, which
> would not yet be too bad, as this is supposed to work with an iddata
> object. But my problem is, that the sampling instants of the two data
> files are not equal. So I can't properly choose in- and outputs.
>
> Any suggestions?
>
> Thanks,
> Simo

Now it seems, that it won't even work with in- and output in same iddata
  object and with same SamplingInstants. If I try to import the object,
it tells me "The sampling interval could not be evaluated. For
Multiexperiment data use cell array."

This isn't very helpful, as I thought that for nonuniformly sampled data
it would be sufficient to proceed as I red in the help file:

"For nonuniformly-sampled data, specify Ts as [],
and set the value of the SamplingInstants property as a
column vector containing individual time values.
For example:data = iddata(y,u,Ts,[],'SamplingInstants',TimeVector)"

And also I don't have multple experiment data.

Still no suggestions?

Simo Kauth <kauth@hrz.tu-chemnitz.de> wrote in message
<g7ubsf$af9$1@anderson.hrz.tu-chemnitz.de>...
> Simo Kauth schrieb:
> > Hi,
> >
> > I have a project going on, where have two log files one
containing
> > input, the other one the output. Both are not uniformly
sampled, which
> > would not yet be too bad, as this is supposed to work
with an iddata
> > object. But my problem is, that the sampling instants of
the two data
> > files are not equal. So I can't properly choose in- and
outputs.
> >
> > Any suggestions?
> >
> > Thanks,
> > Simo
>
> Now it seems, that it won't even work with in- and output
in same iddata
> object and with same SamplingInstants. If I try to
import the object,
> it tells me "The sampling interval could not be evaluated.
For
> Multiexperiment data use cell array."
>
> This isn't very helpful, as I thought that for
nonuniformly sampled data
> it would be sufficient to proceed as I red in the help file:
>
> "For nonuniformly-sampled data, specify Ts as [],
> and set the value of the SamplingInstants property as a
> column vector containing individual time values.
> For example:data =
iddata(y,u,Ts,[],'SamplingInstants',TimeVector)"
>
> And also I don't have multple experiment data.
>
> Still no suggestions?

The best thing you can do is get a copy of the SysId toolbox
documentation, it is free and available at the mathworks
site and run through it. The toolbox lets you use
non-uniform sampling intervals, but I haven't played enough
with it to give you any extra pointers, but heck, L.Ljung
wrote most of the documentation so if you go through it you
should be in good hands.