Automatic Identification of Ductile Fracture Parameters and Optimization of Shearing Process Conditions through Autonomously-driven Finite Element Analysis with Machine Learning-based Optimization

Abstract

We conducted an in-situ observation experiment and an autonomously-driven finite element analysis (ADFEA) of the shearing process for a low-carbon cold-rolled steel. The ADFEA consists of finite element analysis and an optimization method based on machine learning. A critical damage value of Cockcroft and Latham’s ductile fracture criterion was identified to minimize the errors with ADFEA. Two errors were defined as the differences between the experimental and analytical results of the sheared surface lengths of the blank and the scrap. The sheared surface length was maximized by ADFEA with respect to shearing process parameters: the punch tip radius, the die tip radius, and the clearance between the punch and the die. We established a digital knowledge archive (DnA), which visualizes the ADFEA results to identify the critical relationships between process parameters and product properties to facilitate automatic technology transfer.