Prediction of the optimal hydrogen storage in TiVCrMnAlFe alloys using machine learning models

The quest for efficient hydrogen storage materials is vital for the advancement of clean energy technologies. Among various candidates, high entropy alloys (HEAs) such as TiVCrFeAl have received significant attention due to their tunable structures and favorable thermodynamic properties. This study explores the application of machine learning (ML) techniques to predict the hydrogen storage capacity of the TiVCrFeAl alloy system. Using a dataset compiled HEAPS and calculated properties, multiple regression models, such as Random Forest, Gradient Boosting, Decision Tree, XGBoost, Linear Regression and Support Vector Regression, were trained to capture complex relationships between alloy composition, processing parameters, and hydrogen weight percent (wt%). Data preprocessing steps included feature selection, imputation of missing values, and standardization to ensure model robustness. The performance of the models was evaluated using cross-validation and test set metrics such as R² and mean squared error. Results show that ensemble-based models, particularly Random Forest and XGBoost, achieved high predictive accuracy, demonstrating the effectiveness of ML in modeling nonlinear property trends in HEAs. This approach offers a powerful tool for screening and optimizing hydrogen storage materials, accelerating the discovery process through computational insight.