| About this Abstract |
| Meeting |
2023 TMS Annual Meeting & Exhibition
|
| Symposium
|
Accelerated Discovery and Insertion of Next Generation Structural Materials
|
| Presentation Title |
Using Machine Intuitive Learning to Predict Advanced Steel Properties |
| Author(s) |
Krista R. Limmer, Andrew Garza, Heather Murdoch, Benjamin Szajewski, Daniel Field, Christopher Rinderspacher, Levi McClenny, Mulugeta Haile |
| On-Site Speaker (Planned) |
Krista R. Limmer |
| Abstract Scope |
Recent advances in data science and high-throughput materials simulations are being evaluated to accelerate advanced steel alloy development. Here we use machine learning (ML) to take advantage of the large amounts of historic data available for martensitic steels. A series of models and diagrams using varying amounts of data are used to develop predictive ML models. Multiple approaches are used to assess the degree of information required to predict toughness as a function of composition and processing parameters. The first approach directly minimizes the composition and processing variables using Gaussian process regression. The latter approaches incorporate various differing neural networks, such as multilayer perceptron and recurrent neural networks, to predict toughness based on intermediate variables. These intermediate variables are synthetic microstructures and thermodynamic properties generated using high-throughput CALPHAD simulations. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Iron and Steel, Machine Learning |