Identifying productivity-limiting factors in progressive die stamping: data-driven methodology for process optimization

Abstract

Manufacturing products in multi-stage forming processes through blanking, deep-drawing and bending operations with progressive dies is one of the most economically relevant processes in the sheet metal forming industry and allows for complex geometries. In order to be able to ensure a reliable operation of the tools, manufacturers choose stroke speed conservatively, which limits the productivity and profitability. For this reason, this paper describes which productivity-limiting factors affect multi-stage forming processes and how machine learning in combination with explainable artificial intelligence methods can be used to identify and counteract productivity-limiting factors. By equipping the processes with multiple sensors, stroke rate-dependent anomalies can be detected at an early stage and countermeasures can be derived proactively, thus making the process more reliable and ensuring product quality. The methodology developed is demonstrated using two use cases which show that vibrations and friction in a progressive die are significantly reduced.