Models for emulating the human quality assessment of laser cut edges

Abstract

With the increasing degree of automation in production, the ability to automatically assess and monitor the process quality is gaining in importance. In laser cutting, however, the quality of the cut edge is usually evaluated by humans. This paper aims to emulate their quality assessment based on objective criteria and to expand the understanding of which properties of the edge are relevant.

For this purpose, we carried out an expert survey: five experts rated the quality of 100 laser cut edges. Additionally, numerous objective features were measured for each edge. Based on this data, two models were developed that mimic the human quality assessment very well.