Articles | Volume 15, issue 2
https://doi.org/10.5194/jsss-15-133-2026
https://doi.org/10.5194/jsss-15-133-2026
Regular research article
 | 
13 Jul 2026
Regular research article |  | 13 Jul 2026

Application of artificial intelligence to determine the phase fraction of welded duplex steels using neural networks with partial use of augmented data

Leon Kaufhold, Julia Wichmann, Galina Polte, and Thomas Schroeter

Cited articles

Anaconda Software Distribution: Anaconda Documentation (Version 2.5.2), Anaconda Inc., https://docs.anaconda.com/ (last access: 18 September 2024), 2024. 
ASTM International: Standard Test Method for Determining Volume Fraction by Systematic ManuPoint Count (ASTM E562-19), ASTM International, https://doi.org/10.1520/E0562-19, 2019. 
Basler, D.: Neuronale Netze mit C# programmieren – Mit praktischen Beispielen für Machine Leanring im Unternehmenseinsatz, Carl Hanser Verlag GmbH Co,. KG, ISBN 978-3-446-46426-1, 2021. 
Beraha, E.: Farbätzung für die Erkennung von Phosphiden, Karbiden und Nitriden in Eisen, Stahl, Werkzeugstahl, rostfreiem Stahl und hitzebeständigen Legierungen, in: Prakt. Metallogr., vol. 5, Walter de Gruyter, Berlin, Boston, 501–508, https://doi.org/10.1515/pm-1968-050904, 1968. 
Download
Short summary
This article describes the determination of ferrite and austenite proportions in the weld zone of duplex steels using image processing and neural networks. These proportions allow conclusions to be drawn about the quality of the weld. For this purpose, neural networks of different architectures were applied to various image data configurations of metallography images. Depending on the architecture and application of the networks, error rates of 2 % to 3 % were achieved, which met expectations.
Share