Project Subtitle
Industrial Reinforcement Learning for the Quality Control of Metal Forming Processes
Companies & Partners
Mubea, Tailor Rolled Blanks GmbH, Eichsfelder Schraubenwerke GmbH, Schomäcker Federnwerk GmbH, Quality Automation GmbH and MAWI GmbH and Schuler Pressen GmbH, WZL of RWTH Aachen University.
Problem
Instabilities due to external influencing variables and unknown interdependencies between process parameters and product quality characteristics often lead to scrap in bulk metal forming processes despite existing process controls.
Solution
A method based on reinforcement and transfer learning is implemented for the implementation of novel controllers in the quality control loops of bulk metal forming processes. In order to reduce the learning time of the reinforcement learning algorithm and to save resources, it is not only trained directly on the real process, but also on a stochastic process simulation. The knowledge gained from the simulation is then transferred to the quality control loop of the control loop by means of transfer learning. The result is a quality control of bulk metal forming processes, which controls the processes automatically, comprehensively and in real time and optimizes the quality of the processes.
Outcome
The improved process quality in turn increases the quality of the automotive products and reduces scrap.