Paper Info
Title
New Results And Methods In Balancing/Spectral-Zero-Interpolation Based Model Order Reduction
Abstract
This paper studies spectral-zero-interpolation based methods for model order reduction (MOR). We focus on symmetric passive systems in which we prove new results about spectral zeros and balancing of state systems using extremal Algebraic Riccati Equation (ARE) solutions. We first show that for symmetric state space systems, not just are the poles and system zeros interlaced, but in fact, the poles, spectral zeros and system zeros are interlaced too. In the context of positive real balanced state-space realization, we introduce a notion of 'quasi-balanced', which turns out to inter-relate various extremal Riccati equation solutions of the literature. Finally, using the interpretation that the extremal ARE solutions indicate the minimum/maximum energy considerations during charging/discharging processes, we propose methods to choose a suitable subset of spectral zeros at which the reduced order system should interpolate the original transfer function in order to have lower error with respect to the H-infinity, H-2 and Hankel norms.
Year
DOI
Venue
2018
10.23919/ECC.2018.8550159
2018 EUROPEAN CONTROL CONFERENCE (ECC)
Keywords
Field
DocType
spectral zeros, maximum and minimum Riccati equation solutions, error in approximation, balancing methods in MOR
Applied mathematics,Read-only memory,Model order reduction,Interpolation,Transfer function,Riccati equation,Algebraic Riccati equation,Passive systems,State space,Mathematics
Conference
Citations 
PageRank 
References 
0
0.34
0
Authors
3
Name
Order
Citations
PageRank
Sandeep Kumar100.68
Madhu N. Belur23713.87
Debasattam Pal32812.84