Enhanced processing map of Ti–6Al–2Sn–2Zr–2Mo–2Cr–0.15Si aided by extreme gradient boosting

A processing map is required for Ti alloys to find processing parameters securing a high formability. This study adopted the extreme gradient boosting (XGB) approach of machine learning to predict a flow curve and plot a processing map with less experiments for the first time. The optimum XGB model predicted flow curves of Ti–6Al–2Sn–2Zr–2Mo–2Cr–0.15Si at 1073–1273 K and 10 s−1. The predicted data were used to plot a processing map, which showed a higher accuracy in the instability map as compared with the map without XGB. The XGB model also anticipated the power dissipation map at low strain rates. The low accuracy at high strain rates would be improved by alleviating the bias towards a flow hardening. This work has successfully proven the potential usefulness of XGB for plotting an enhanced processing map in light of a higher accuracy with less experiments. Metals forming and shaping; Metals and alloys; Deformation and fracture; Processing map; Machine learning.