Study on the key features of dynamic compressive fracture strain of Ti-Zr-Nb solid solution alloys through random forest regressor

In some specific application fields, dynamic fracture strain regarding as evaluating dynamic properties of Ti-Zr-Nb solid solution alloy have attracted extensive attention. However, the main influence factors of the dynamic strain of alloys were unclear. For the purpose of regulating Ti-Zr-Nb alloys' dynamic plasticity and clarify main influence factors of the dynamic plasticity of the materials, powder metallurgy, dynamic properties test combined with machine learning were performed. 56 Ti-Zr-Nb alloys were prepared through powder metallurgy and their dynamic compressive fracture strain was tested. Furthermore, optimization of machine learning model and selection of key features for the prediction of dynamic compressive fracture strain were carried out. The prediction accuracy of optimized model was more than 80%, and three key features that significantly influence the dynamic fracture strain were selected and ordered as: VEC>λ>ΔG.