Improved mu-analysis results by using low order uncertainty modelling techniques

Abstract

The paper describes some recent enhancements of symbolic factorization techniques for multivariate Laurent polynomial matrices, which are very effective as a preprocessing step for generating linear fractional representations of low order. The symbolic methods are applied to realize a linear fractional representation of full accuracy and probably minimal order for a very complex linear parametric representation of a Research Civil Aircraft Model. The µ-analysis results obtained with the very accurate and low-order linear fractional representation coincide with the robust stability analysis results obtained from a computationally demanding, optimization-based, worst-case search. Furthermore, it is shown that the quality of the µ-analysis results depends on the accuracy and order of the underlying model, i.e. the tightest upper and lower bounds are obtained using the most accurate and least-order linear fractional representation.