Approximate Modelling by Means of Orthonormal Functions
1990 (English)Report (Other academic)
ARX, AR and FIR modeling are generalized by repacing the delay operator by discrete Laguerre/Kautz filters. The aim is to obtain useful low order approximate models of complex systems by using a priori information about the dominating time constants of the system. An important characteristic of these models are that they can be written in a linear regressions form. Hence, the least squares methods can be applied for system identification. By deriving naturally associated state-space realizations of Laguerre/Kautz models, we obtain model representations that are more directly suitable for control design etc. The orthonormal property of Laguerre/Kautz functions is very important, since it guarantees a Toeplitz structure of the corresponding least squares normal equations. This property enables us to determine persistence of excitation conditions and analyze asymp totic statistical properties of Laguerre/Kautz model estimates.
Place, publisher, year, edition, pages
Linköping: Linköping University , 1990.
LiTH-ISY-I, ISSN 8765-4321 ; 1138
Laguerre filter, Kautz filter, Approximation, Complex system, Linear regression, Least square method
IdentifiersURN: urn:nbn:se:liu:diva-55323OAI: oai:DiVA.org:liu-55323DiVA: diva2:316060