Space Mapping and Beyond: Knowledge-Driven Microwave Design Optimization

Slawomir Koziel, Kaj Madsen

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    8 Downloads (Pure)

    Abstract

    Design closure exploiting electromagnetic (EM) solvers has become one of the fundamental design tools in contemporary microwave engineering. For many structures, adjustment of geometry and/or material parameters can only be done through repetitive EM simulations because analytical design formulas either do not exist or can only provide initial designs that need to be further refined. Unfortunately, EM-driven optimization is a challenging problem with the major bottleneck being a high computational cost of accurate simulation. This problem can be alleviated by using fast and yet reliable surrogate models that can replace the CPU-intensive EM-simulated structure of interest in the search for optimum design. The surrogate models exploiting physically-based low-fidelity models (e.g., circuit equivalents) can be particularly efficient: the knowledge about the structure under design embedded in such a low-fidelity model allows us to dramatically reduce the number of EM simulations necessary to find a satisfactory design. Here, we review the concept of knowledge-driven design as well as specific design techniques, including space mapping, simulation-based tuning, and various response correction methods. Discussion on open problems and perspectives of these methodologies is also included.
    Original languageEnglish
    Title of host publication2012 IEEE MTT-S International Microwave Symposium Digest (MTT)
    Number of pages3
    PublisherIEEE
    Publication date2012
    ISBN (Print)978-1-4673-1088-8, 978-1-4673-1085-7
    ISBN (Electronic)978-1-4673-1086-4
    DOIs
    Publication statusPublished - 2012
    EventIEEE International Microwave Symposium (IMS) - Montreal, Canada
    Duration: 17 Jun 201222 Jun 2012
    http://ims2012.mtt.org/

    Conference

    ConferenceIEEE International Microwave Symposium (IMS)
    CountryCanada
    CityMontreal
    Period17/06/201222/06/2012
    Internet address
    SeriesI E E E - M T T S International Microwave Symposium. Digest
    ISSN0149-645X

    Keywords

    • Computer-aided design (CAD)
    • knowledge-driven design
    • Space Mapping
    • Simulation-based tuning
    • Surrogate modeling

    Fingerprint Dive into the research topics of 'Space Mapping and Beyond: Knowledge-Driven Microwave Design Optimization'. Together they form a unique fingerprint.

    Cite this