Joining processes in general and beam welding processes in particular, are a highly important part of the value-added chain. They are directly dependent on upstream production (procured material, deviations from the nominal geometry) and have a direct influence on downstream processes (cleaning, surface properties of the joining points). The components are often delivered to the machine without (documented) knowledge of the actual state of the machine, and often it can be seen only then whether the parts delivered from the worldwide network meet the requirements of the locally available plant technology and can thus be joined faultlessly.
The actual state is of decisive importance for many joining processes, as a joint can be created by adapting the process parameters despite different preparation of good parts. However, it is still unclear which data are all of relevance for quality-compliant joining. Furthermore, today’s joining processes are generally not “adaptive”. An optimization of process parameters due to deviations in the preceding process steps only takes place via the know-how of the responsible plant operators.
The general goal of this project is therefore to develop a networked (internet-based), location-independent, self-learning system for process control for beam welding processes on the basis of a digital shadow. First of all, the process knowledge must be made available in digitized form (information technology, data processing) to allow sufficient speed for process control. Furthermore, correlations must be established between process information (here: beam welding) and process results or process behaviour (sensor technology, engineering technology).