Deep learning in shock response of metastable high entropy alloy: from prediction to inverse microstructure design
Year: 2025
Type: article
Abstract: Metastable high-entropy alloys (HEA) have attracted great interest due to their excellent mechanical performance during shock environments. However, the complex non-linear relationship between microstructural characteristics and shock response remains to be revealed. In this study, a series of dynamic compression experiments of V 10 Cr 10 Fe 45 Co 30 Ni 5 metastable HEA have been performed to disclose the role of martensitic transformation in shock response. Corresponding to the martensitic transformation mechanism and the temperature effect, the dislocation-based crystal plasticity constitutive model considering martensitic transformation and adiabatic temperature rise has been established to investigate the shock response. Then, a novel bidirectional deep learning framework with convolutional neural networks (CNN) and generative adversarial networks (GAN) models is proposed to establish the non-linear relationship between microstructure and shock response, progressively achieve the rapid prediction of shock responses, and the inverse design of microstructure. Based on the dataset generated within the crystal plasticity hypothesis space, the deep learning framework is trained to well map the non-linear relationship between microstructure and shock response. In addition, the GAN model can accurately determine the texture and temperature based on the targeted stress–strain curve. The present study can help understand the connection between shock response and microstructure design of the metastable HEA.
Source: Materials & Design
Authors: Chuan-zhi Liu, Wen An, Qiang Zhou, Qi-lin Xiong
Institution: Huazhong University of Science and Technology
Language: English
Topic: High Entropy Alloys Studies
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,650
OpenAlex ID: W4416409327