Back to Search Start Over

Rapid Optimization of Compact Microwave Passives Using Kriging Surrogates and Iterative Correction

Authors :
Slawomir Koziel
ALBERTO ROJAS BUENO
Anna Pietrenko-Dabrowska
Verkfræðideild (HR)
Department of Engineering (RU)
Engineering Optimization & Modeling Center (EOMC) (RU)
Tæknisvið (HR)
School of Technology (RU)
Háskólinn í Reykjavík
Reykjavik University
Source :
IEEE Access, Vol 8, Pp 53587-53594 (2020)
Publication Year :
2020
Publisher :
IEEE, 2020.

Abstract

Publisher's version (útgefin grein)<br />Design of contemporary microwave components is & x2014;in a large part & x2014;based on full-wave electromagnetic (EM) simulation tools. The primary reasons for this include reliability and versatility of EM analysis. In fact, for many microwave structures, notably compact components, EM-driven parameter tuning is virtually imperative because traditional models (analytical or network equivalents) are unable to account for the cross-coupling effects, strongly present in miniaturized layouts. At the same time, the cost of simulation-based design procedures may be significant due to a typically large number of evaluations of the circuit at hand involved. In this paper, a novel approach to expedited design closure of compact microwave passives is presented. The proposed procedure incorporates available designs (e.g., existing from the previous design work on the same structure) in the form of the kriging interpolation models, utilized to yield a reasonable initial design and to accelerate its further refinement. An important component of the framework is an iterative correction procedure that feeds the accumulated discrepancies between the target and the actual design objective values back to the kriging surrogate to produce improved predictions. The efficacy of our methodology is demonstrated using two miniaturized impedance matching transformers with the optimized designs obtained at the cost of a few EM simulations of the respective circuits. The relevance of the iterative correction is corroborated through the comparative studies showing its superiority over rudimentary gradient-based refinement.<br />This work was supported in part by the Icelandic Centre for Research (RANNIS) under Grant 174114051, and in part by the National Science Centre of Poland under Grant 2015/17/B/ST6/01857.<br />"Peer Reviewed"

Details

Language :
English
ISSN :
21693536
Volume :
8
Database :
OpenAIRE
Journal :
IEEE Access
Accession number :
edsair.doi.dedup.....18fbf6072f5d320a0b4e3c98d6d2fdcd