Image-assimilation of deformed dual-phase steel microstructure via U-net deep learning
Takashi Matsuno, Yota Fukuda, Kazuyuki Shimizu, Hiroto Shoji, Mitsuru Ohata, Norio Yamashita, Hideo Yokota, Tetsuro Murai
Abstract. Ferrite-martensite dual-phase (DP) steels are widely utilized in automotive applications as high-strength materials owing to their optimal combination of tensile strength and ductility. However, compared with conventional mild steels, their high strength often limits ductility, which frequently results in material failure during press forming. Although various countermeasures have been employed in press-forming methodologies and material development, manufacturers continue to encounter sporadic fracture defects, occurring at a probability of one in tens or hundreds of instances. Such probabilistic defects lack reproducibility, making it challenging to identify the underlying cause associated with the material microstructure. To address this issue, elucidating the mechanical state of a microstructure that has undergone probabilistic ductile fracture is crucial. This study proposes a novel method for determining the stress and strain distributions within the deformed microstructure of DP steel by assimilating finite element (FE) simulations. This method involves a machine-learning process using multiple representative volume elements (RVEs) of FE simulations, which incorporate virtual microstructures mimicking actual DP steels. The deep learning model, based on a U-net architecture, was trained using the deformed microstructural meshes of the RVE-FE simulations to enable assimilation of the actual deformed microstructure. By inputting the deformed DP microstructures obtained via 3D serial sectioning into the U-Net model, the stress–strain distribution within the actual DP microstructure was successfully visualized.
Keywords
Image Assimilation, Deep Learning, Surrogate Analysis, Dual-Phase Steel
Published online 5/7/2025, 8 pages
Copyright © 2025 by the author(s)
Published under license by Materials Research Forum LLC., Millersville PA, USA
Citation: Takashi Matsuno, Yota Fukuda, Kazuyuki Shimizu, Hiroto Shoji, Mitsuru Ohata, Norio Yamashita, Hideo Yokota, Tetsuro Murai, Image-assimilation of deformed dual-phase steel microstructure via U-net deep learning, Materials Research Proceedings, Vol. 54, pp 963-970, 2025
DOI: https://doi.org/10.21741/9781644903599-103
The article was published as article 103 of the book Material Forming
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
References
[1] T. Matsuno, D. Kondo, T. Hama, T. Naito, Y. Okitsu, S. Hayashi, K. Takada, Flow stress curves for 980MPa- and 1.5GPa-class ultra-high-strength steel sheets weakened under high-stress triaxiality, Int. J. Mech. Sci., 261 (2024) 108671. https://doi.org//10.1016/j.ijmecsci.2023.108671
[2] T. Matsuno, T. Fujita, T. Matsuda, Y. Shibayama, T. Hojo, I. Watanabe, Unstable stress-triaxiality development and contrasting weakening in two types of high-strength transformation-induced plasticity (TRIP) steels: Insights from a new compact tensile testing method, J. Mater. Process. Tech., 322 (2023) 118174. https://doi.org/10.1016/j.jmatprotec.2023.118174
[3] Y. Matsuki, S. Tobita, K. Nakagawa, T. Shinmiya, Y. Yamasaki, Y. Tamai, Effect of Microstructural Transformation upon Edge Heat Treatment on Stretch Flangeability of Ultrahigh-Strength Steel with Sheared Edge, Mater. Trans., 64 (2023) 2278-2285. https://doi.org//10.2320/matertrans.MT-P2023001
[4] T. Matsuno, K. Sato, R. Okamoto, M. Mizumura, M. Suehiro, Synergy Effect of shear angle and anisotropic material ductility on hole-expansion ratio of high-strength steels, J. Mater. Process. Tech., 230 (2016) 167-176. https://doi.org/10.1016/j.jmatprotec.2015.10.018
[5] T. Matsuno, Y. Sekito, K. Kawasaki, Microstructure characterization of fine grains near hot-sheared surface formed during hot-stamping process, J. Mater. Process. Tech., 229 (2016) 570-581. https://doi.org/10.1016/j.jmatprotec.2015.10.012
[6] T. Matsuno, Y. Yamazaki, S. Matsubara, J. Eguchi, K. Shimizu, M. Koyama, Y. Okitsu, T. Tsuchiyama, Local Work-Hardening and Lüders Deformation Invariance in TRIP-Aided Medium-Mn High-Strength Steel: Significant TRIP Effect Changes in the Warm Temperature Range up to 200 ℃, J. Jpn. Soc. Technol. Plast., 66 (2025) 2400702. https://doi.org//10.9773/sosei.240702
[7] T. Matsuno, K. Furukawa, Y. Okitsu, M. Koyama, T. Tsuchiyama, Local yield stress and its unusual independence on multi-axial stress states during Lüders deformation of medium-Mn, high-strength steel, ISIJ Int., 64 (2024) 859-867. https://doi.org/10.2355/isijinternational.ISIJINT-2023-326
[8] T. Matsuno, C. Teodosiu, D. Maeda, A. Uenishi, Mesoscale simulation of the early evolution of ductile fracture in dual phase steels, Int. J. Plast., 74 (2015) 17-34. https://doi.org/10.1016/j.ijplas.2015.06.004
[9] T. Matsuno, N. Kinoshita, T. Matsuda, Y. Honda, T. Yasutomi, Microvoid Formation of Ferrite-martensite Dual-phase Steel via Tensile Deformation after Severe Plastic Shear-deformation, ISIJ Int., 63 (2023) 941-949. https://doi.org//10.2355/isijinternational.ISIJINT-2023-012
[10] T. Matsuno, Y. Ueda, T. Takahashi, T. Hama, T. Hojo, Y. Shibayama, M. Ridha, Y. Okitsu, M. Takamura, Delayed fracture stress thresholds degraded by shear punching process in ultra-high-strength steel sheets: Analysis using an in-plane bending test with numerical assimilation, J. Manuf. Process., 119 (2024) 1005-1021. https://doi.org/10.1016/j.jmapro.2024.04.013
[11] N. Yamashita, T. Matsuno, D. Maeda, M. Kikuzuki, H. Yokota, Efficient 3D observation of steel microstructure using serial sectioning with precision cutting and on-site etching, Precis. Eng., 75 (2022) 37-45. https://doi.org//10.1016/j.precisioneng.2022.01.004
[12] T. Matsuno, T. Yoshioka, I. Watanabe, L. Alves, Three-dimensional finite element analysis of representative volume elements for characterizing the effects of martensite elongation and banding on tensile strength of ferrite-martensite dual-phase steels, Int. J. Mech. Sci. 163(2019) 105133. https://doi.org/10.1016/j.ijmecsci.2019.105133
[13] Ronneberger, P. Fischer, T. Brox, U-Net: Convolutional Networks for Biomedical Image Segmentation, in: N. Navab, J. Hornegger, W.M. Wells, A.F. Frangi (Eds.) Medical image computing and computer-assisted intervention – MICCAI 2015, Springer Link, Munich, Germany, (2015), 234-241. https://doi.org/10.1007/978-3-319-24574-4_28
[14] J. Kang, Y. Ososkov, J. D. Embury, D. S. Wilkinson, Digital image correlation studies for microscopic strain distribution and damage in dual phase steels, Scripta Mater. 56(2007), 999-1002. https://doi.org/10.1016/j.scriptamat.2007.01.031
