Bounding L2 Gain System Error Due to Approximations of the Nonlinear Vector Field
2007 (English)In: Proceedings of the IEEE Conference on Computer-Aided Design, 2007, 879-886 p.Conference paper (Refereed)
Typical nonlinear model order reduction approachesneed to address two issues: reducing the order ofthe model, and approximating the vector field. In this paperwe focus exclusively on the second issue, and present resultscharacterizing the repercussions at the system level of vector fieldapproximations. The error assessment problem is formulatedas the L2 gain upper bounding problem of a scaled feedbackinterconnection. Applying the small gain theorem in the proposedsetup, we prove that the L2 gain of the error system is upperbounded by the L2 gain of the vector field approximationerror, provided it is small. In addition, the paper also presentsa numerical procedure, based on the IQC/LMI approach, toperform the error estimation task with less conservatism. Anumerical example is given in this paper to demonstrate thepractical implications of the presented results.
Place, publisher, year, edition, pages
2007. 879-886 p.
IdentifiersURN: urn:nbn:se:kth:diva-82413OAI: oai:DiVA.org:kth-82413DiVA: diva2:498216
IEEE/ACM International Conference on Computer-Aided Design. San Jose, CA, USA. 5-8 November 2007
QC 201205222012-02-112012-02-112012-05-22Bibliographically approved