Running in forming machines is often a lengthy and cost-intensive process. The machines are then often blocked for hours or even days by "experimenting" and "trying out" different process parameters. The rejects generated in the process, the so-called faulty parts, increase the costs incurred in addition to the machine hours.

Basically, run-in processes block machines, materials and people, leaving them unavailable for actual, productive manufacturing. To remedy this situation, a team of companies has come together around Aalen University and contributed their expertise and software tools to find a solution to this very problem.

The IoT platform ThingWorx (PTC), provided by the company Inneo Solutions in Ellwangen, as well as the corresponding simulation software LS-DYNA and AI software from Mathlab, which are integrated into the process model by Aalen University under the direction of Dr. Wolfgang Rimkus, Julius Schlosser and Dr. Sebastian Feldmann, will be used.

The aim of the project is to use machine learning to evaluate data from real production, in this case forming, by means of image recognition and the correlation of the production parameters used. The data is collected using a smart tool with integrated sensors. Algorithms are then generated from this using machine learning. This allows corresponding patterns that lead to defective parts to be recognized and stored as a cause-effect relationship.

This process knowledge is then made available to the virtual world for simulation purposes. The LS-DYNA simulation software uses the CAD data of the tool and blank and the process knowledge gained to carry out forming simulations every minute. The reference data obtained from the forming process is used to train the AI. These results are then continuously evaluated by connected AI software and recognized patterns are recorded. On this basis, the AI software optimizes the production parameters provided and suggests process parameters for real production, which then reduces the running-in of the machine to a minimum.

The potential savings that can be achieved in this way can give industry a head start in terms of production times and production costs. The project "Real-time data acquisition and parameter correction using an AI trained with simulation data ("SimKI")" is funded by the Baden-Württemberg Ministry of Economic Affairs, Labor and Housing as part of the "AI for SMEs" innovation competition. as