Incorporating Term selection into nonlinear block structured system identification

Mohammad Rasouli, David T. Westwick, W. D. Rosehart

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

Subset selection and shrinkage methods locate and remove insignificant terms from identified models. The least absolute shrinkage and selection operator (Lasso) is a term selection method that shrinks some coefficients and sets others to zero. In this paper, the incorporation of constraints (such as Lasso) into the linear and/or nonlinear parts of a Separable Nonlinear Least Squares algorithm is addressed and its application to the identification of block-structured models is considered. As an example, this method is applied to a Hammerstein model consisting of a nonlinear static block, represented by a Tchebyshev polynomial, in series with a linear dynamic system, modeled by a bank of Laguerre filters. Simulations showed that the Lasso based method was able to identify the model structure correctly, or with mild over-modeling, even in the presence of significant output noise.

Original languageEnglish (US)
Title of host publicationProceedings of the 2010 American Control Conference, ACC 2010
PublisherIEEE Computer Society
Pages3710-3715
Number of pages6
ISBN (Print)9781424474264
DOIs
StatePublished - 2010

Publication series

NameProceedings of the 2010 American Control Conference, ACC 2010

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering

Fingerprint Dive into the research topics of 'Incorporating Term selection into nonlinear block structured system identification'. Together they form a unique fingerprint.

Cite this