Industrial production uses various manufacturing processes. A considerable part of the added value is directly attributable to the use of joining techniques such as welding.
There are numerous methods of destructive as well as non-destructive testing for process monitoring of different joining processes. All testing methods require a high level of technical expertise of the personnel and are usually carried out after the joining process. However, testing the components after the joining process increases production times to a not insignificant extend. Added to this, especially in the case of destructive testing, are tons of test scrap, which is neither resource saving nor economical. Non-destructive testing methods would therefore be desirable, when they can already be used for quality assurance during the joining process.
One possibility for the realization of a non-destructive testing method – the detection of faults in the process based on its noise (airborne sound analysis) – is being investigated within the AKoS project by a broadly diversified consortium. The work here concentrates on non-destructive and automated quality assurance of safety-critical components in the joining process, specifically during welding.
The aim of this project is therefore the development of a universally applicable adaptive learning algorithm for in-situ weld seam inspection. By adapting its parameters, the algorithm should be transferable to as many joining processes as possible in order to allow statements to be made about the occurrence of selected irregularities. It is important that the algorithm can be operated by non-experts and adapted to constantly changing boundary conditions of joining processes.
July 2020 – June 2022