Title :
A New Analysis on PVL Model Reduction Using Solutions of Linear Systems by the Lanczos Method
Author_Institution :
Dept. of Math. Sci., Tsinghua Univ., Beijing, China
Abstract :
It is theoretically well known that PVL model reduction is locally effective, but it may perform quite well globally in practice. This phenomenon is not well understood. In this paper, we give a solid theoretical analysis for the convergence of PVL model reduction. We transform it to the solution of parametrized linear systems by the nonsymmetric Lanczos method. We establish a result on its local effectiveness and shed lights on its possible global effectiveness. So our analysis gives more theoretical insights into the convergence or effectiveness of PVL model reduction.
Keywords :
convergence of numerical methods; linear systems; linearisation techniques; reduced order systems; PVL model reduction; global effectiveness; nonsymmetric Lanczos method; parametrized linear systems; Approximation algorithms; Computational modeling; Error analysis; Large-scale systems; Linear systems; Power system modeling; Reduced order systems; Solids; Transfer functions; Vectors;
Conference_Titel :
Image and Signal Processing, 2009. CISP '09. 2nd International Congress on
Conference_Location :
Tianjin
Print_ISBN :
978-1-4244-4129-7
Electronic_ISBN :
978-1-4244-4131-0
DOI :
10.1109/CISP.2009.5304588