Title: Fast realization of the modal vector fitting method for rational modeling with accurate representation of small eigenvalues
Authors : Gustavsen, Bjørn
Heitz, Christoph
Published in : IEEE : Transactions on Power Delivery
Volume(Issue) : 24
Issue : 3
Pages : 1396
Pages to: 1405
Publisher / Ed. Institution : Institute of Electrical and Electronics Engineers
Issue Date: 2009
License (according to publishing contract) : Licence according to publishing contract
Type of review: Not specified
Language : English
Subjects : Signal processing
Subject (DDC) : 621.3: Electrical engineering and electronics
Abstract: Admittance-based rational modeling of multiport systems is prone to error magnification in applications with high-impedance terminations. This problem is overcome by the modal vector fitting method (MVF) which is formulated in terms of modal components with inverse least-squares weighting by the eigenvalue magnitude. A direct realization of MVF is very demanding in computation time and memory requirements. This paper overcomes the performance deficiency via three steps: 1) the required number of MVF iterations is reduced by precalculating an improved initial pole set via conventional vector fitting with inverse magnitude weighting; 2) the pole identification step is calculated in an efficient manner by solving for only the few essential unknowns while exploiting the sparse matrix structure; and 3) the residue identification step is calculated efficiently by a row-wise solution procedure that takes advantage of symmetry. The approach is demonstrated to give large savings for the modeling of a frequency-dependent network equivalent.
Departement: School of Engineering
Organisational Unit: Institute of Data Analysis and Process Design (IDP)
Publication type: Article in scientific Journal
DOI : 10.1109/TPWRD.2008.2005462
ISSN: 0885-8977
URI: https://digitalcollection.zhaw.ch/handle/11475/4428
Appears in Collections:Publikationen School of Engineering

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.