A Power Electronic Converter Control Framework Based on Graph Neural Networks — An Early Proof-of-Concept

Darius Jakobeit, Oliver Wallscheid · 2026

Read A Power Electronic Converter Control Framework Based on Graph Neural Networks — An Early Proof-of-Concept by Darius Jakobeit, Oliver Wallscheid (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-06686.txt.

Description

Darius Jakobeit and Oliver Wallscheid teach one brain-like net to run many power boxes in 2026. Control of such boxes is now tuned box by box, they point out. That blocks reuse when the shape of the box changes from job to job.

Their plan does not care what shape the box has, by design. The box's wire list is writ as a graph of two node kinds. A graph net reads that graph, shaped by the task at hand. Small heads spread through the box then act on what the net finds. One frame could run a whole line of box shapes with no fresh tuning. They show this as an early proof that the thought can work. If it holds, design time drops for whole fleets of such boxes.

Catalog record

CREATORDarius Jakobeit, Oliver Wallscheid
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-06686

Browse the full library · About the Tower of Babel