Efficient Phase Classification in Aluminum-Containing High Entropy Alloys Using Machine Learning

High entropy alloys, especially aluminum-containing high entropy alloys (HEAl), have attracted extensive attention due to their excellent mechanical properties and wide application potential. However, HEAl has diverse components and complex phase structures. Traditional phase prediction methods such as CALPHAD are limited in efficiency and accuracy when dealing with complex multi-component systems. In this paper, a machine learning model is used to predict the phase structure of HEAl. Four machine learning algorithms, including multi-layer perceptron (MLP), support vector machine (SVM), gradient boosted decision tree (GBDT) and random forest (RF), are used. Feature selection is performed on a data set (17V) of 17 feature descriptors, and an optimized data set (8V) of 8 feature descriptors is formed after redundant features are eliminated. The predicted results show that the optimized 8V can improve the generalization ability of the model, especially the SVM and RF models show stronger prediction stability when dealing with data imbalance and random sampling of the test set. The F1-score results show that the prediction performance of FCC and BCC phases is improved after data dimensionality reduction and feature optimization. The model proposed in this paper has strong adaptability in dealing with the diversity and complexity of HEAl phase prediction, providing an effective way for the theoretical design of HEAl.