The TWIN research project at ScaDS.AI and the University of Leipzig focused on the development process for manufacturing prototypes (product development phase) using selective laser melting. To this end, TWIN developed a digital product service system
involving the entire additive manufacturing value chain. The aim was to optimise and monitor complex industrial systems with the help of machine learning methods in order to be able to provide meaningful recommendations for action during the production process. The thesis examined the use of neural networks for the classification of quality characteristics within 3D
printing processes. Two data sets with greyscale images documenting the printing process served as the basis. The focus is on the investigation of neural networks inspired by numerical methods for solving differential equations. Various approaches of this type are compared and the results evaluated. Another focus is the stability analysis of such methods.