Model reduction methods for unstable uncertain systems

We propose methods for reduction of unstable uncertain systems; in particular the generalization of coprime factors model reduction method introduced by Meyer (1988) to uncertain system models is considered. These reduction methods are applicable to uncertain system realizations that are not stable in a robust l/sub 2/ setting. Error bounds are given in the induced 2-norm of the factorized mapping, where this norm is computed over the unity norm-bounded structured uncertainty set. A second reduction method is considered, namely that of simply scaling the allowable uncertainty set down to a size at which the system is robustly stable, and applying previously developed reduction methods for stable uncertain systems.