Cart Summary
0 item(s) ($0.00)
Product Search
All Products
AIST Transactions
Books
Conference Proceedings
Industry Data
Iron & Steel Technology
About the AIST Resource Center
Product Details:

Machine Learning Surrogate Model for Sensitivity Analysis in Hot Stamping

The role of simulation in the sensitivity analysis studies is crucial due to the need of exploring the parameter space without perturbing the real manufacturing system. In hot stamping, the initial conditions, the setup parameters and the materials properties form a wide domain. The configurations within the domain have a direct impact in the thermomechanical phenomena and the material phase transformations occurring during hot stamping. Therefore, many simulations are required for extensive sensitivity analysis and optimization studies during the design stage to ensure the desired mechanical properties in the final parts. However, the elevated cost in time and computational resources of these simulations and the high dimensionality of the domain is an important limitation. The aim of this work is to present a pipeline to overcome this drawback, with a Machine Learning based surrogate model of the simulation of the hot stamping of a hat-shaped part of boron steel. Moreover, a comparison between a Latin Hypercube Sampling and a Forward Selection method is implemented to show the sampling importance in surrogate modeling. The introduced methodology is an enabler to boost the sensitivity analysis and optimization procedures, due to the fast response of the surrogate model estimations. The proof-of-concept results show high potential in the soft-real time prediction of unseen configurations within the domain, focusing on important variables regarding the mechanical properties and the quality of the final part, such as the temperature and the martensite content.

DOI Number: 10.33313/512/A0201
Product:
2024 CHS2 Conference Proceedings
PR-512-A0201
Albert Abio, Francesc Bonada, Jorgen Kajberg, Fredrik Larsson, Daniel Casellas, Jaume Pujante, et al
May 27, 2024
PDF
$15.00
$25.00
$25.00
Available for Immediate Download