Quantitative criteria are presented for model simplification, or order reduction, such that the reduced order model may be used to synthesize and evaluate a control law, and the stability robustness obtained using the reduced order model will be preserved when controlling the full-order system. The error introduced due to model simplification is treated as modeling uncertainty, and some of the results from multivariate robustness theory are brought to bear on the model simplification problem. A numerical procedure developed previously is shown to lead to results that meet the necessary criteria. The procedure is applied to reduce the model of a flexible aircraft. Also, the importance of the control law itself, in meeting the modeling criteria, is underscored. An example is included that demonstrates that an apparently robust control law actually amplifies modest modeling errors in the critical frequency region, and leads to undesirable results. The cause of this problem is associated with the canceling of lightly damped transmission zeroes in the plant. An attempt is made to expand on some of the earlier results and to further clarify the theoretical basis behind the proposed methodology.
Multivariable frequency weighted model order reduction for control synthesis
1989-12-15
Conference paper
No indication
English
Multivariable frequency-weighted order reduction
AIAA | 1989
|Optimal model reduction and frequency-weighted extension
AIAA | 1990
|Multivariable Flex Model Reduction (AAS 03-600)
British Library Conference Proceedings | 2004
|Frequency Stability Criteria of Multivariable Nonlinear Control System
British Library Online Contents | 1999
|