About this Abstract |
Meeting |
2nd World Congress on High Entropy Alloys (HEA 2021)
|
Symposium
|
2nd World Congress on High Entropy Alloys (HEA 2021)
|
Presentation Title |
Machine Learning-assisted Phase Prediction for High Entropy Alloys and Feature Analysis |
Author(s) |
Kyungtae Lee, Timothy Q. Hartnett, Mukil V. Ayyasamy, Prasanna V. Balachandran |
On-Site Speaker (Planned) |
Kyungtae Lee |
Abstract Scope |
High entropy alloys (HEAs) are typically solid solutions of face centered cubic, body centered cubic, or hexagonally closed packed phases. The goal of this work is to develop a machine learning approach to optimally guide the HEA design with the targeted phase(s) in the microstructure. Our input dataset for ML models is prepared by merging several datasets from previous reports about the HEA phase prediction. The dataset ranges from binary to multi-component alloys, categorized by seven different phases. We represent each alloy using 125 features that were mainly generated based on the physical and chemical properties of elements that make up the alloy. Redundant features were removed using correlation analysis. Two machine learning algorithms such as random forest and ensemble of support vector machines are considered for the classification learning task, showing predictability close to 90%. Global and local feature importance analysis uncovers insights into the HEA phase formation problem. |
Proceedings Inclusion? |
Undecided |