System Identification of Electromagnetic Devices Based on Full-Wave Computations
Paper i proceeding, 2021

We present a framework that allows for the estimation of port-to-port characteristics of electromagnetic devices, which are linear and passive. Our approach is based on system identification (SI) techniques applied to the numerically computed admittance or impedance matrices of the electromagnetic device. The SI procedure yields a low-order model expressed in terms of a Padé approximant, which is represented as the ratio of two polynomials with respect to frequency. In this article, we demonstrate that the admittance and impedance matrices computed at a rather small number of frequency points can yield a highly accurate low-order model that describes the system response as a continuous function throughout the frequency band used for estimation of the model. The computational cost to store and evaluate the low-order model is basically negligible in comparison to the computational cost required by the numerical full-wave solver. The derived port-to-port model allows for the individual and independent analysis of subsystems that do not couple electromagnetically, where such subsystems can be combined later to form a complete system and this allows for great flexibility in a virtual design-process. We test our approach on two different crosstalk problems.

Method of Moments

EMC

System Identification

MIMO

Crosstalk

Författare

Carl Holmberg

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

Thomas Rylander

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

Jan Carlsson

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

Tomas McKelvey

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

2021 Joint IEEE International Symposium on Electromagnetic Compatibility Signal and Power Integrity, and EMC Europe, EMC/SI/PI/EMC Europe 2021

952-957

2021 Joint IEEE International Symposium on Electromagnetic Compatibility Signal and Power Integrity, and EMC Europe, EMC/SI/PI/EMC Europe 2021
Raleigh, USA,

Virtuell elektrisk drivlina

VINNOVA (2017-05505), 2018-02-26 -- 2021-12-31.

Ämneskategorier

Beräkningsmatematik

Sannolikhetsteori och statistik

Reglerteknik

DOI

10.1109/EMC/SI/PI/EMCEurope52599.2021.9559211

Mer information

Senast uppdaterat

2021-11-15