摘要
Overcoming the trade-offbetween 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-anderror 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).
基金
supported by the National Key R&D Program of China(Grant No.2022YFB2404101)
the“Pioneer”R&D Programof Zhejiang Province(No.2023C01075)
the Youth Innovation Promotion Association CAS(Grant No.2021294)
the Ningbo Natural Science Foundation(No.2021J197).
