Design of high strength medium-Mn steel using machine learning

Abstract

The objective of present study was to develop a medium-Mn steel with superior tensile properties using machine learning. For this purpose, 1075 datasets on tensile properties of medium-Mn steels were collected from the literature. Based on the datasets, boosted decision tree regression (BD) models were constructed to predict ultimate tensile strength (UTS) and total elongation (TE) of medium-Mn steels. The BD models showed low mean absolute errors of ∼65 MPa for UTS prediction and ∼4.9% for TE prediction. The trained BD models predicted that Fe-5.5Mn-0.2C-0.3Si (wt%) steel would have high UTS of 1957 MPa and TE of 10.7%, when austenitized at 780 °C for 4 min and air-cooled. The predicted UTS and TE matched well with experimentally measured values of UTS of 1952 MPa and TE of 9.9%, indicating the outstanding predictability of the BD models. In addition, the measured UTS (1952 MPa) was ∼100 MPa higher, without a great loss of TE, than the highest UTS (1863 MPa) of Fe-(3–12)Mn-(<0.3)C-(<0.5)Si-(<1)Al (wt%) steels reported to date.