About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Thermodynamics and Kinetics of Alloys
|
Presentation Title |
CALPHAD and Data-driven Approach for Phase Prediction Model in Refractory High-entropy Alloys |
Author(s) |
Jiwon Park, Chang-Seok Oh |
On-Site Speaker (Planned) |
Jiwon Park |
Abstract Scope |
In this work, a data-driven approach is proposed for the phase prediction models and feature selection strategies in refractory high-entropy alloys (RHEA) based on the data collected from the literature, CALPHAD computations, and data analyses. Themodynamic properties such as liquidus and solidus temperatures, and the enthalpy of mixing were added to extend relevant features. As the dataset consists of many features compared to the number of composition sets, dimensionality reduction is essential to improve the model efficiency. Several dimensionality reduction methods including PCA and GA in feature selection are compared, and the machine learning models are explained by Sharpley values in the phase classification models when the phases are categorized in 4 classes; single BCC, phase-separated BCCs, BCC, FCC and/or HCP, and mixture of BCC and intermetallic. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, High-Entropy Alloys, Machine Learning |