Design automation combining analytical models and numerical simulations for three-level neutral-point-clamped dc-dc Converter

Yudi Xiao, Martijn Duraij, Zhe Zhang, Gabriel Zsurzsan, Mads Hansen, Brian Thomsen

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

    Abstract

    Design automation in power electronics has been highlighted in recent years. Design flow, multi-physics co-design, data standardization have been identified as the major gaps for engineers and researchers to fill. As an exploratory effort, this paper presents an automated design for the three-level neutral-point-clamped dc-dc converter. The design firstly uses analytical models to determine the circuit level parameters. The problem size can thereby be reduced. The parameter candidates left are fed to automated FEMM and COMSOL simulations for precise modeling of magnetic components. Database published by ferrite and powder core manufacturers is incorporated in the design. The code of the entire design flow has been published on GitHub.
    Original languageEnglish
    Title of host publicationProceedings of 2021 IEEE Design Methodologies Conference
    Number of pages6
    PublisherIEEE
    Publication date2021
    ISBN (Print)978-1-6654-3140-8
    DOIs
    Publication statusPublished - 2021
    Event2021 IEEE Design Methodologies Conference - Virtual event, Bath, United Kingdom
    Duration: 14 Jul 202115 Jul 2021
    https://ieeexplore.ieee.org/xpl/conhome/9529348/proceeding

    Conference

    Conference2021 IEEE Design Methodologies Conference
    LocationVirtual event
    Country/TerritoryUnited Kingdom
    CityBath
    Period14/07/202115/07/2021
    Internet address

    Keywords

    • Design automation
    • Finite element simulation
    • Multi-level dc dc converter
    • Silicon Carbide (SiC)

    Fingerprint

    Dive into the research topics of 'Design automation combining analytical models and numerical simulations for three-level neutral-point-clamped dc-dc Converter'. Together they form a unique fingerprint.

    Cite this