How BayernKI Supports Engineering

A research team from Munich University of Applied Sciences is using the BayernKI AI infrastructure to advance quality assurance and product development in engineering. The researchers trained graph neural networks with simulation data to predict stresses in welded components, helping to assess the reliability and safety of welded joints more efficiently.

Prof. Marcin Hinz from the Chair of Mechanical, Automotive and Aviation Technology at the Munich University of Applied Sciences notes: “With AI, we need less time to evaluate stresses during welding, whilst achieving an accuracy of between 97 and 99 per cent.”

The project also lays the groundwork for future industrial applications, such as evaluating complex structures, testing new materials, and inspecting critical infrastructure, including bridges and railway systems.

Read the article in full length at https://www.lrz.de/en/news/detail/how-ai-is-improving-welding