Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Investigation on optimal microstructure of dual-phase steel with high strength and ductility by machine learning

Misato Suzuki, Kazuyuki Shizawa, Mayu Muramatsu

TL;DR

An inverse analysis framework that proposes a microstructure for dual-phase (DP) steel that exhibits high strength and ductility and a microstructure with a fine grain size was proposed by using the developed framework.

Abstract

In this study, we developed an inverse analysis framework that proposes a microstructure for dual-phase (DP) steel that exhibits high strength and ductility. The inverse analysis method proposed in this study involves repeated random searches on a model that combines a generative adversarial network (GAN), which generates microstructures, and a convolutional neural network (CNN), which predicts the maximum stress and working limit strain from DP steel microstructures. GAN was trained using images of DP steel microstructures generated by the phase-field method. CNN was trained using images of DP steel microstructures, the maximum stress and the working limit strain calculated by the dislocation-crystal plasticity finite element method. The constructed framework made an efficient search for microstructures possible because of a low-dimensional search space by a latent variable of GAN. The multiple deformation modes were considered in this framework, which allowed the required microstructures to be explored under complex deformation modes. A microstructure with a fine grain size was proposed by using the developed framework.

Investigation on optimal microstructure of dual-phase steel with high strength and ductility by machine learning

TL;DR

An inverse analysis framework that proposes a microstructure for dual-phase (DP) steel that exhibits high strength and ductility and a microstructure with a fine grain size was proposed by using the developed framework.

Abstract

In this study, we developed an inverse analysis framework that proposes a microstructure for dual-phase (DP) steel that exhibits high strength and ductility. The inverse analysis method proposed in this study involves repeated random searches on a model that combines a generative adversarial network (GAN), which generates microstructures, and a convolutional neural network (CNN), which predicts the maximum stress and working limit strain from DP steel microstructures. GAN was trained using images of DP steel microstructures generated by the phase-field method. CNN was trained using images of DP steel microstructures, the maximum stress and the working limit strain calculated by the dislocation-crystal plasticity finite element method. The constructed framework made an efficient search for microstructures possible because of a low-dimensional search space by a latent variable of GAN. The multiple deformation modes were considered in this framework, which allowed the required microstructures to be explored under complex deformation modes. A microstructure with a fine grain size was proposed by using the developed framework.
Paper Structure (14 sections, 18 equations, 23 figures, 6 tables)

This paper contains 14 sections, 18 equations, 23 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Overview of inverse analysis framework
  3. Construction of GAN
  4. Phase-field method
  5. Generative adversarial network
  6. Construction of CNN
  7. Dislocation-crystal plasticity finite element method
  8. Convolutional neural network
  9. Results and Discussion
  10. Output of DP steel microstructures based on latent variables
  11. Output of maximum stress and working limit strain based on microstructures
  12. Investigation of DP steel microstructure with high strength and ductility
  13. Discussion of employing random search for inverse analysis
  14. Conclusion

Figures (23)

  • Figure 1: Conceptual diagram of the inverse analysis framework applied to DP steel
  • Figure 2: An example of the initial conditions and generated microstructures in the phase-field analysis. (a) DP steel microstructure obtained by the phase-field method. (b) DP steel microstructure obtained in the experiment myeong2017effect.
  • Figure 3: Example of converting the results of phase-field analysis to pixel data. (a) Point data. The upper image means the probability that the point is variant$1$ of martensite. The lower image means the probability that the point is variant$2$ of martensite. (b) Pixel data.
  • Figure 4: Architecture of GAN. (a) Architecture of generator network. (b) Architecture of discriminator network.
  • Figure 5: Boundary conditions for FEM. (a) Tensile toward $x$ direction. (b) Tensile toward $y$ direction. (c) Shear toward $x$ direction. (d) Shear toward $y$ direction
  • ...and 18 more figures