Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Evaluation of AI-solution for Microstructural Analysis of Steel
KTH, School of Industrial Engineering and Management (ITM), Materials Science and Engineering.
KTH, School of Industrial Engineering and Management (ITM), Materials Science and Engineering.
2026 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Manual microstructural analysis of high-strength steel wire is both time-consuming and prone to subjectivity, creating challenges for consistent quality control in industrial production. This study evaluates the feasibility of integrating AI-based image analysis into the metallographic workflow at Suzuki Garphyttan AB, using the Olympus PRECiV software. Both pre-trained integrated AI models and custom-trained deep learning models were assessed against manual baseline measurements for grain size (ASTM E112) and oxide layer thickness (ASTM B487) across five specimens each. Additionally, a standardised image acquisition protocol was developed for the Olympus DSX 1000 3D-microscope. Results demonstrate that all three methods produced grain size numbers in close agreement across all specimens, indicating promising accuracy for industrial use. Oxide layer thickness results showed good overall agreement, though specimens with irregular oxide morphology presented greater divergence between methods. The integrated AI method is the most immediately deployable solution, offering denser sampling and grain size distributions beyond what manual methods provide. The findings support a semi-automated workflow as a viable complement to manual metallographic analysis.

Abstract [sv]

Manuell mikrostrukturell analys av höghållfast ståltråd är både tidskrävande och utsatt för subjektivitet, vilket medför utmaningar för en konsekvent kvalitetskontroll inom industriell produktion. Denna studie utvärderar möjligheten att integrera AI-baserad bildanalys i det metallografiska arbetsflödet hos Suzuki Garphyttan AB, med hjälp av mjukvaran Olympus PRECiV. Både förtränade integrerade AI-modeller och egenutvecklade djupinlärningsmodeller jämfördes mot manuella referensmätningar av kornstorlek (ASTM E112) och oxidskiktstjocklek (ASTM B487) på fem prover vardera. Dessutom utvecklades ett standardiserat protokoll för bildinsamling till Olympus DSX 1000 3D-mikroskop. Resultatet visar att alla tre metoder producerade kornstorlekar i god överensstämmelse för samtliga prover, vilket indikerar lovande noggrannhet för industriell användning. Oxidskiktstjockleken visade generellt god överensstämmelse, men prover med oregelbunden oxidmorfologi uppvisade större avvikelser mellan metoderna. Den integrerade AI-metoden är den mest omedelbart implementerbara lösningen som samtidigt erbjuder fler mätpunkter och en mer detaljerad bild av kornstorleksfördelningen jämfört med manuella metoder. Resultaten stöder ett halvautomatiserat arbetsflöde som ett lämpligt komplement till manuell metallografisk analys.

Place, publisher, year, edition, pages
2026. , p. 46
Series
TRITA-ITM-EX ; 2026:117
National Category
Engineering and Technology
Identifiers
URN: urn:nbn:se:kth:diva-384435OAI: oai:DiVA.org:kth-384435DiVA, id: diva2:2082233
External cooperation
Suzuki Garphyttan AB
Subject / course
Materials and Process Design
Educational program
Master of Science in Engineering - Materials Design and Engineering
Supervisors
Examiners
Available from: 2026-06-30 Created: 2026-06-30

Open Access in DiVA

fulltext(8290 kB)77 downloads
File information
File name FULLTEXT01.pdfFile size 8290 kBChecksum SHA-512
f11dc17eef5857952ec44cd426605fd3b8d126ad5d0822f577b6a504994b3fffbcbb5df886630b633b33a4062c6c6b43c7583934e0ff06bbb3349dd18cb9d8f9
Type fulltextMimetype application/pdf

By organisation
Materials Science and Engineering
Engineering and Technology

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 1313 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf