The intrinsic strength prediction by machine learning for refractory high entropy alloys

Abstract

Herein, we trained machine learning (ML) model to quickly and accurately conduct the strength prediction of refractory high entropy alloys (RHEAs) matrix. Gradient Boosting (GB) regression model shows an outstanding performance against other ML models. In addition, the heat of fusion and atomic size difference is shown to be paramount to the strength of the high entropy alloys (HEAs) matrix. In addition, we discussed the contribution of each feature to the solid solution strengthening (SSS) of HEAs. The excellent predictive accuracy shows that the GB model can be efficient and reliable for the design of RHEAs with desired strength.