Prediction of phase via machine learning in high entropy alloys

High Entropy Alloys (HEAs) are a new group of materials with numerous applications. Phase formation in HEAs is necessary for their qualities, but it is difficult to anticipate them effectively. In this study, we are implementing a support vector machine (SVM) model, we separate the model into stable body centre cubic (BCC), face centre cubic (FCC) and a few HEAs with help of cross-validation. From 381 as-cast data the model was used to predict the same atom HEAs of 81 FCC and 155 BCC, with the remaining being FCC + BCC and intermetallic from the compositional space of elements in 14 metals, with accuracy above 86% in training and testing. With much higher prediction accuracy, machine learning methods have been utilised in this study to categorise and identify the phases in HEAs with comparability-predicted precision. Five Thermodynamic features are estimated based on prior research. Various selection strategies were applied to quantify the feature relevance. Entropy variation is discovered to be the least prominent aspect using feature importance.