NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/softsys/matlab/231

Text File | 1993-01-28 | 3.0 KB | 82 lines

Newsgroups: comp.soft-sys.matlab
Path: sparky!uunet!ukma!usenet.ins.cwru.edu!magnus.acs.ohio-state.edu!csn!news.den.mmc.com!stote
From: [e-mail] (M. Stote)
Subject: Re: Robust Controller Synthesis
Message-ID: <[e-mail]>
Sender: stote@tomcat (M. Stote)
Nntp-Posting-Host: tomcat.den.mmc.com
Organization: Martin Marietta Astronautics Group
References:  <[e-mail]>
Date: Thu, 28 Jan 1993 23:19:31 GMT
Lines: 69

In article <[e-mail]>, [e-mail] (-BJS-) writes:
|>
|>
|>    I have a question for those of you readers who have performed
|> optimal controller synthesis using the robust control and
|> mutools toolboxes available in matlab.  I am attempting to
|> design a controller for an "improper" system, i.e., the
|> frequency response does not roll-off at high frequencies.  Is
|> there any such requirement on the plant in order to use the
|> 'hinfsyn', 'h2syn', musynfit, etc. functions.
|>

  The only requirement is that the systems be `at least'
proper.  That is they cannot go to infinity as freq goes to
infinity.


|>    The problem I am encountering does not seem to be related to
|> this; I have problems with the row rank test on the matrix [a
|> b1; c2, d21].  This matrix ends up being (7x5) and thus cannot
|> have full row rank.  Obviously, I have formulated the problem
|> incorrectly, but I am not sure where I've gone astray.  There
|> are three measurements available to the controller and the "Z"
|> vector has one measurement.  The performance specification (Wp)
|> is second order, is this okay??  Can anyone give me possible
|> avenues to explore which I might have missed??  The requirement
|> indicates that the number of n+p must be equal to or less than
|> m1+m2, where;
|>
|>     n  = number of states,
|>     p  = number of outputs,
|>     m1 = number of exogenous inputs,
|>     m2 = number of controls
|>
|>
|>    Any ideas would be greatly appreciated either by e-mail or via
|> posting.  Thanks in advance.
|>

The basic requirements are that:
   p2 (sensed outputs) <= m1 (exog inputs)
   p1 (regulat outputs) >= m2 controls

In addition the sensor outputs must contain some exogenous
input signal over the entire frequency range.  Failure to do this
is usually what causes the rank condition failure you described.

Also be careful of zeros at s=0 or almost 0 in the weighting functions
on the exogenous inputs to the sensor signals.

                     Befuddled but not Defeated,
|>
|>                     -BJS-
|> --
|>
|> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
|> Brian J. Scallan,                  [e-mail]
|>
|> When angry, count four; when very angry, swear.
|> Pudd'nhead Wilson - M. Twain
|>
|> The opinions expressed
|> should not be interpreted as those
|> of my employer or any other (semi-)rational entity.            {-BJS-}
|> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

--
Mike Stoughton
[e-mail]