Dear Professor Skogestad,
Thank you for your prompt response regarding my query about the
availability in DOS format of the m-files featured in your book.
I am wondering if you can assist me with some problems I am having in applying
one of the methods described in the book.
I am attempting to design a 2 degrees-of-freedom controller for a
hydraulic robot manipulator following the procedure you described. I
have achieved a pretty good loop shape, but on using coprimeunc.m two
problems occur
a} Warnings for poorly conditioned matrices
b} At different times either of the Riccatti systems are
ill-posed. Both my (A,Q,R) systems are minimal, ie
controllable and observable, so I am somewhat confused
I am hoping that you will be able to provide brief answers to the following
1 How well does the coprime factorisation method work on
non-minimum phase systems
2 My nominal model is actually a very accurate 8th order model (5
zeros) obtained by using invfreqs.m to identify transfer function
measurements taken from a non-linear simulation. Would taking a
lower order nominal model alleviate my conditioning problems.
(companion form state-space realisations of MIMO systems do have
a lot of noughts). Could the ill-conditioning be leading
directly to the ill-posedness.
Thank you for your attention. I look forward to hearing from you.
Much regards
A. H. Fraser, Development Engineer (Control Systems)
UK Robotics Ltd.
Derwent House,
Clarence Avenue,
Trafford Park,
Manchester,
M17 1QS.
email ainsley.fraser@robotics.uk.co
Tel: +44 (0) 161 876 3200
Fax: +44 (0) 161 876 3201
Received on Wed Sep 11 18:28:05 1996
This archive was generated by hypermail 2.1.8 : Tue Jan 20 2004 - 09:32:14 CET