Thermography-based methods for monitoring of hardness and workpiece placement in press hardening

In this approach, we present a concept for monitoring hardness and workpiece position in press hardening based solely on thermographic images. Thus, independent of press type and control, the concept can be utilized. We present several ways to process the data, including classical image processing, basic machine learning, and deep learning, which differ in the required supervision during the implementation and in the achieved accuracy. We demonstrate that all methods are able to handle their respective use cases, whereas there are differences in accuracy and necessary supervision, which are addressed in the paper.