| Abstract Scope |
High-entropy alloys (HEAs) have roughly 10⁸ potential compositions from 64 elements in the periodic table. These large design combinations present a considerable challenge to materials scientists, in forecasting compositions suitable for thermal and mechanical behavior. Prediction through experiments is time-consuming and expensive. AlCrFeCoNi-based HEAs are particularly appealing for high-temperature structural applications due to their solid-solution stability and enhanced performance at elevated temperatures. In this study, machine learning algorithms are employed to optimize the thermomechanical properties of these alloys at high temperatures. Four predictive models, an artificial neural network (ANN), a random forest (RF), extreme gradient boosting (XGBoost), and a support vector machine (SVM), are employed using experimental data from open literature. XGBoost has the best prediction accuracy, allowing for the precise calculation of important parameters such as hardness, yield strength, and thermal stability. This work establishes a method for accelerating the design of next-generation high-temperature HEAs. |