Re: MIMO models in Control 2.0.0

[Lukas Reichlin]

On 21.03.2011, at 13:58, Richard Balogh wrote:

> Hello, 
> 
> if I understand it well, it is not yet possible to construct MIMO models in
> Control 2.0.0 package.
> Or is there a way how to model the step response of the DC motor with two
> inputs (voltage, load) and one output (omega)? 
> 
> I need to model it and to inspect its responses - see also attached figure.
> Any advices, please?
> 
> http://octave.1599824.n4.nabble.com/file/n3393485/SimulinkDCM.png 
> 
> 
> Richard Balogh

Hi Richard,

MIMO models are possible both in state-space as well as in transfer functions. 
However, there are some issues with tf models:

The time responses are computed in state-space. Therefore, tf models are 
converted internally to state-space first. Since the tf2ss routine (taken from 
control-1.0.11) is not capable to convert MIMO models, you will get an error 
message for MIMO tf models in combination with time response commands. 
Frequency responses are *not* affected at all.

Because of the same "old" tf2ss routine, tf model interconnections are limited 
to SISO models. If you're interested in details, have a look at the file 
@tf/__sys_connect__.m, you get the path by "which tf".

Most of the existing conversion routines from SLICOT are limited to proper 
models. I need routines which are able to convert non-proper (more zeros than 
poles) transfer functions to descriptor state-space models and back. Today I 
don't understand its theory. As a stopover, I'll use the SLICOT routines in a 
future release.

OK, enough of the technobabble :-) I recommend that you use state-space models 
to describe MIMO systems.

Regards,
Lukas