Skip to main content
main-content

06.03.2018 | Materials Technology | News | Onlineartikel

Optimising the Classification of Materials with Deep Learning

Autor:
Nadine Winkelmann

Materials can be classified quickly and without errors using machine learning techniques. This has allowed computer scientists at the Saarbrücken campus and materials researchers to accurately determine the microstructures of low-carbon steel.

"Manufacturing special steels is an extremely complex process that depends on many individual factors including the chemical composition of the material, the rolling process used and the types of heat treatment that the material is subjected to. Every stage of the production process influences the internal structure of the steel," explains Dominik Britz, PhD student in the Department of Functional Materials at Saarland University, Germany. In order to classify a material, the microscope images are compared with sample images showing a typical geometric microstructure. Since the deviations in the images are sometimes barely visible to the naked eye, classifying materials in this way is prone to errors. Materials scientists, together with computer scientists in Saarbrücken, have therefore developed a method that is much more accurate and objective than conventional quality control procedures.

Classification accuracy of 93 percent

Machine learning methods (deep learning) allow computers to recognise complex patterns very rapidly and to assign the geometry of the microstructures in microscope images. They can learn the features of previously classified microstructures and compare these with recognised patterns. Using this approach, Saarbrücken's researchers were able to determine the microstructures of low-carbon steel at a level of accuracy that was previously impossible. "When using our system for microstructural classification, we achieved a level of accuracy of around 93 percent. With conventional methods, only about 50 percent of the material samples are correctly classified," explains Professor Frank Mücklich, who supervised the study.

"We see this as just the beginning of a close cooperative partnership with Saarbrücken’s highly respected computer science research teams. The new deep learning methods will not only help us assess the quality of steel more objectively and more accurately, we also anticipate that our results will be transferable to many other production processes and materials," Mücklich explains.

The scientific study "Advanced Steel Microstructural Classification by Deep Learning Methods" has been published in Scientific Reports from the journal "Nature".

Weiterführende Themen

Die Hintergründe zu diesem Inhalt

Das könnte Sie auch interessieren

Premium Partner

in-adhesivesMKVSNeuer Inhalt

BranchenIndex Online

Die B2B-Firmensuche für Industrie und Wirtschaft: Kostenfrei in Firmenprofilen nach Lieferanten, Herstellern, Dienstleistern und Händlern recherchieren.

Whitepaper

- ANZEIGE -

Technisches Interface Design - Beispiele aus der Praxis

Eine gute Theorie besticht nur darin, dass am Ende einer Entwicklung sinnvolle und nutzergerechte Produkte herauskommen. Das Forschungs- und Lehrgebiet Technisches Design legt einen starken Wert auf die direkte Anwendung der am Institut generierten wissenschaftlichen Erkenntnisse. Die grundlegenden und trendunabhängigen Erkenntnisse sind Grundlage und werden in der Produktentwicklung angewendet. Nutzen Sie die Erkenntnisse aus den hier ausführlich dargestellten Praxisbespielen jetzt auch für Ihr Unternehmen.
Jetzt gratis downloaden!

Marktübersichten

Die im Laufe eines Jahres in der „adhäsion“ veröffentlichten Marktübersichten helfen Anwendern verschiedenster Branchen, sich einen gezielten Überblick über Lieferantenangebote zu verschaffen. 

Bildnachweise