Roll forming is a continuous cold forming process that combines high process speeds with almost 100% material utilisation. Straightening processes are used to compensate for profile errors during production.

As part of the development of an intelligent straightening process for roll forming, forces and positions are linked to correct adjustment trajectories. This enables a digitalisation of the traditionally experience-based process of manual straightening and increases its process reliability in the light of an increasing shortage of skilled workers in Germany.

In order to reduce the time and money required to generate extensive, labelled data sets, this thesis aims to transfer the model-based knowledge of one use case to another. For this purpose, data sets with different materials are recorded and models based on them are transferred from one use case to another using domain adaptation methods.

Existing experience in processing data using AI models is beneficial.

Research method

Theoretical, experimental