home *** CD-ROM | disk | FTP | other *** search
- %
- % MRDEMO Robust model reduction demonstration
- %
-
- % R. Y. Chiang & M. G. Safonov 3/88
- % Copyright (c) 1988 by the MathWorks, Inc.
- % All Rights Reserved.
- % ----------------------------------------------------------------
- clc
- disp(' ')
- disp(' ')
- disp(' <<< Robust Basis Free Model Reduction Techniques Demo >>>')
- disp(' ')
- disp(' ')
- disp(' 1). Balanced model reduction techniques')
- disp(' (Schur method vs. Square-root method)')
- disp(' ')
- disp(' 2). Optimal Descriptor Hankel model reduction & ')
- disp(' anticausal projection w/o balancing')
- disp(' ')
- disp(' 3). Schur Balanced Stochastic Trunction /')
- disp(' Relative Error Method')
- disp(' ')
- disp(' 4). Comparison of 1, 2, and 3.')
- disp(' ')
- disp(' 5). Go to the main menu')
- disp(' ')
- disp(' 0). Quit ..')
- disp(' ')
- Demono = input(' Select one of the above options: ');
- % ----------------------------------------------------------------------
- if Demono == 0
- clc
- disp(' ')
- disp(' ')
- disp(' ')
- disp(' ')
- disp(' * * * * * * * * * * * * * * * * *')
- disp(' * End of MRDEMO ...... *')
- disp(' * *')
- disp(' * Good luck with your design !! *')
- disp(' * * * * * * * * * * * * * * * * *')
- break
- end
- if Demono == 5
- rctdemo1
- end
- % ----------------------------------------------------------------------
- clc
- disp(' ' )
- disp(' << Motivation of Robust Model Reduction >>')
- disp(' ')
- disp(' Robust model reduction techniques have become more and more important.')
- disp(' ')
- disp(' This is due to the following:')
- disp(' ')
- disp(' 1). Complexity of the design problem (e.g. large size plant)')
- disp(' ')
- disp(' 2). Complexity of the design algorithm')
- disp(' ')
- disp(' 3). Large size of the controller generated by modern control')
- disp(' techniques -- e.g. LQG, LQG/LTR, H2, H-inf ..etc..')
- disp(' ')
- disp(' Model reduction algorithms must be both numerically robust and')
- disp(' ')
- disp(' be able to address closed-loop "robustness" issues.')
- disp(' ')
- disp(' ')
- disp(' (strike a key to continue ...)')
- pause
- clc
- disp(' The model reduction algorithms included here have the following features:')
- disp(' ')
- disp(' 1). They all by-pass the ill-conditioned "balanced transformation"')
- disp(' ')
- disp(' 2). They employ Schur decomposition as an intermediate step')
- disp(' ')
- disp(' for robust computation of orthonormal bases for eigenspaces.')
- disp(' ')
- disp(' 3). They all have infinity-norm error bounds:')
- disp(' ')
- disp(' Methods Infinity-Norm Error Bounds')
- disp(' ---------------------------- -------------------------------')
- disp(' Schur/Square-root Balanced twice of the tails of Hankel')
- disp(' (schur.m & balmr.m) singular values')
- disp(' ')
- disp(' Optimal Descriptor Hankel same as above')
- disp(' without balancing (ohklmr.m)')
- disp(' -1 ^ n hsv')
- disp(' Balanced Stochastic Truncation/ |G (G-G) | <= 2 Sum -------')
- disp(' Relative Error Method (reschmr.m) inf k+1 1-hsv')
- disp(' ')
- disp(' (strike a key to continue ...)')
- pause
- % ---------------------------------------------------------------------
- if Demono == 1
- baldemo
- mrdemo
- end
- % ----------------------------------------------------------------------
- if Demono == 2
- ohkdemo
- mrdemo
- end
- % ----------------------------------------------------------------------
- if Demono == 3
- remdemo
- mrdemo
- end
- % ----------------------------------------------------------------------
- if Demono == 4
- bhrdemo
- mrdemo
- end
- %
- % --------- End of MRDEMO.M --- RYC/MGS %
-