Determination of material deformation rate based on artificial intelligence using surface microstructure images


Creative Commons License

Özdem S., Seyyarer E., Orak İ. M.

9th (Online) International Conference on Applied Analysis and Mathematical Modeling (ICAAMM21), İstanbul, Turkey, 11 - 13 June 2021, pp.110-117

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.20852/ntmsci.2021.437
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.110-117
  • Van Yüzüncü Yıl University Affiliated: Yes

Abstract

Making evaluations on the ever-increasing data and thus obtaining meaningful outcome becomes more and more important in the digital age. Thanks to the recent advances in artificial intelligence technologies, processing this data and making predictions is very popular. Within the scope of this study, S235 JR steel, which is used in many areas such as bridges, railways, industrial buildings, vehicle manufacturing and oil-gas exploration stations, has been deformed at different rates in the laboratory environment. Microstructure images of the deformed materials were obtained with the help of a microscope after some metallographic processes. As a result of the study, it is aimed to contribute to the literature by creating a data set containing microstructure images of S235 JR steel, which has been deformed at different rates. In addition, Convolutional Neural Network (CNN) and Multilayer Artificial Neural Network (ANN) models, which are among the deep learning methods of artificial intelligence technology, were used to classify the deformation rates. In order to minimize the error in both deep learning models, Adam Optimization Algorithm has been preferred. The results obtained with the Adam Optimization Algorithm in CNN and Multilayer ANN models were compared and the highest success in the classification process was obtained in the CNN model.