A Dynamical Minimal Parametrization of Multivariable Linear Systems and its Application to Optimization and System Identification
1999 (English)In: Proceedings of the 14th IFAC World Congress, 1999, Vol. H, 7-12 p.Conference paper (Refereed)
A minimal state-space parametrization which is dynamically or adaptively changed during the optimization process is proposed. The basis is an overparametrized model structure which can describe all systems of given dimensions. This reduces the search space to the minimal dimension. The construction of this local parametrization can be done efficiently and in such a way that the numerical operations in each optimization iteration becomes well conditioned.
Place, publisher, year, edition, pages
1999. Vol. H, 7-12 p.
State-space models, Parametrization, Optimization, Linear systems, System identifcation
Engineering and Technology Control Engineering
IdentifiersURN: urn:nbn:se:liu:diva-91180ISBN: 978-0080432137OAI: oai:DiVA.org:liu-91180DiVA: diva2:617269
14th IFAC World Congress, Beijing, China, July, 1999