| Abstract Scope |
In this study, several machine-learning (ML) models, namely linear regression model, decision trees-based models (i.e., random forest regressor, extra trees regressor, and extreme gradient boosted) and deep learning-based model (i.e., artificial neural network) have been utilized to predict solute segregation energy of Nb solute in Ni grain boundaries (GBs). Towards this, the ML models have been trained using the data extracted from pre-segregation GBs. All ML models performed well, yielding remarkably low root mean square error values for both bicrystal and nanocrystalline specimens. Finally, ML models have been employed to predict solute segregation energy for nine distinct solutes (Co, Fe, Ag, Nb, Zr, Cu, Cr, Mn, Al). This assessment aims to evaluate the performance and effectiveness of our model for bicrystal Ni GBs. The ML model found to be insensitive to the types of solutes and their structures as it demonstrated excellent performance across a wide range of solutes. |