PREDICTION OF MECHANICAL PROPERTIES OF LEAN DUPLEX STAINLESS STEEL BY USING RANDOM FOREST

抄録

Lean duplex stainless steel (LDSS), which is expected to be applied in bridges, exhibits a rounded stress-strain curve. For this reason, a constitutive equation that is able to express the curve accurately is required for ultimate strength analysis of LDSS structures. The authors have already proposed the modified Ramberg-Osgood (MRO) curve as a suitable equation. However, to describe the equation, not only are 0.2% proof stress and tensile strength required, as specified in the common material standard and mill certificates, but also mechanical properties such as proportion limit, etc. The present study collected tension coupon test results for LDSS, and created a simple prediction equation by means of linear regression analysis. It also predicted the mechanical properties by using Random Forest (RF), a machine learning method. When comparison was made, it was revealed that RF predicts mechanical properties as accurately as a prediction equation.