Parametric approximation of electrical circuit responses
Katarzyna Opalska , Leszek J. Opalski
AbstractThe paper presents results of numerical investigations of two algorithms for parametric approximation of such responses of electric circuits which can be accurately approximated by rational functions of frequency, time etc. for fixed values of parameters. A piecewise multilinear algorithm and a two-stage algorithm with frequency and magnitude scaling of root rational models are compared. Both techniqus use Vector Fitting algorithm for preliminary rational approximations of the responses for parameter values from a fixed grid. Numerical investigation uses three circuits which differ in dependence of rational model order on parameter values (fixed or variable) and also on parameterization of circuit elements (linear or non-linear). Dependence of RMS approximation error on grid density is found to be approximately quadratic w.r.t. each variable, admitting pretty large size of grid cells. It is shown, that the two-stage parametric approximation is not only (typically much) more accurate, but does not show spurious oscillations of the predicted responses. The price for the accuracy is the exponential complexity of the more accurate of the two algorithms.
|Pages||108082S-1 - 108082S-11|
|Publication size in sheets||0.55|
|Book||Romaniuk Ryszard, Linczuk Maciej Grzegorz (eds.): Proceedings of SPIE: Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2018, Proceedings of SPIE: The International Society for Optical Engineering, vol. 10808, 2018, SPIE - the International Society for Optics and Photonics, ISBN 9781510622036, 2086 p., DOI:10.1117/12.2504983|
|Keywords in Polish||parametric approximation, rational functions, vector fitting, frequency response, simulation, parametric modeling, optimization|
|Score|| = 0.0, 16-10-2018, ArticleFromJournal|
= 5.0, 16-10-2018, ArticleFromJournal - czasopismo zagraniczne spoza list
* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.