Interpretable machine learning-assisted design of Fe-based nanocrystalline alloys with high saturation magnetic induction and low coercivity

Abstract

Overcoming the trade-off between saturation magnetic induction ( B s ) and coercivity ( H c ) of Fe-based nanocrystalline alloys (FNAs) remains a great challenge due to the traditional design relying on trial-and-error methods, which are time-consuming and inefficient. Herein, we present an interpretable machine learning (ML) algorithm for the effective design of advanced FNAs with improved B s and low H c . Firstly, the FNAs datasets were established, consisting of 20 features including chemical composition, process parameters, and theoretically calculated parameters. Subsequently, a three-step feature selection was used to screen the key features that affect the B s and H c of FNAs. Among six different ML algorithms, extreme gradient boosting (XGBoost) performed the best in predicting B s and H c . We further revealed the association of key features with B s and H c through linear regression and SHAP analysis. The valence electron concentration without Fe, Ni, and Co elements (VEC1) and valence electron concentration (VEC) ranked as the most important features for predicting B s and H c , respectively. VEC1 had a positive impact on B s when VEC1 < 0.78, while VEC had a negative effect on H c when VEC < 7.12. Optimized designed FNAs were successfully prepared, and the prediction errors for B s and H c are lower than 2.3 % and 18 %, respectively, when comparing the predicted and experimental results. These results demonstrate that this ML approach is interpretable and feasible for the design of advanced FNAs with high B s and low H c .