About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithm Development in Materials Science and Engineering
|
Presentation Title |
Parameter Prediction of Anisotropic Yield Function from Neural Network-based Indentation Plastometry |
Author(s) |
Minwoo Park, Kyeongjae Jeong, Kyungyul Lee, Dongil Kwon, Myoung-Gyu Lee, Heung Nam Han |
On-Site Speaker (Planned) |
Minwoo Park |
Abstract Scope |
Indentation plastometry provides an efficient and non-destructive approach to predict the plastic characteristics of materials, replacing traditional destructive test. Plasticity anisotropy is an important factor that affects the formability of materials. Recently, attempts have been made to accurately predict it through indentation plastometry, but sufficient research has not been conducted.
In this study, we propose a framework that utilizes finite element(FE) analysis and artificial neural networks(ANNs) to infer plastic anisotropy from residual indentation data. By comparing actual indentation data with FE simulations, we investigate the influence of mechanical parameters on indentation response and an accurate representation of the indentation process was achieved. An FE-NN model was developed by training it on a database constructed from FE simulations implementing a digital twin. The predictability of the FE-NN model was evaluated by comparing and validating the predicted plastic anisotropy, obtained through inverse analysis, with uniaxial tensile test data measured in different directions. |
Proceedings Inclusion? |
Planned: |
Keywords |
Mechanical Properties, Machine Learning, Modeling and Simulation |