Innovative AI-based Quality Control for More Stable Laser Cutting in Electromobility
Laser cutting is an established cutting process without tool wear for metallic and non-metallic workpieces of different material thicknesses and any cutting contour. This makes it predestined for the production of electrified powertrain components. Copper, a particularly important material for electrification, poses special challenges for the cutting process due to its low absorptivity at common laser wavelengths. Only process control makes it possible to reliably manufacture important components such as copper pins made of 0.5 mm thin copper sheet or current conductors (so-called busbars) in the sheet thickness range of 6 to 8 mm.
The aim is to develop a process control system to enable economical and reliable laser cutting processes of copper materials.
In the project modern algorithms from the field of machine learning are used to extract information from multivariate sensor data. From this information, parameters for evaluating the cutting quality are derived, taking into account physical process models. By feedback to the process, a control is thus made possible. IC9 - CutAIye uses control concepts from the field of artificial intelligence such as reinforcement learning.
This approach goes far beyond the state of the art in laser cutting, since it allows a control mapping from the high-dimensional space of the measured variables to the equally high-dimensional space of the manipulated variables. This makes it possible to control cutting parameters such as laser power, cutting speed and gas pressure, which mutually influence each other and whose interaction is highly complex.