Victor, Viny SaajanViny SaajanVictorEttmüller, ManuelManuelEttmüllerSchmeißer, AndreAndreSchmeißerLeitte, HeikeHeikeLeitteGramsch, SimoneSimoneGramsch2025-08-182025-08-182024-06-25https://publica.fraunhofer.de/handle/publica/49066710.1109/CAI59869.2024.00138Industrial textiles have increasingly become a major part of our day-to-day lives owing to their various desirable properties. Melt spinning processes are a primary and integral part of the production of these textiles. Optimizing the spinning process while maintaining desirable quality is one of the key challenges for the textile industry. Although numerical models, which are digital twins of physical processes, are often used in optimization, they tend to be computationally expensive for complex scenarios. Hence, in this paper, we utilize machine learning to facilitate the optimization of melt spinning processes. We present a novel, reliable, and informed machine-learning model that is both data-and physics-driven. We further demonstrate the capability of this model to accelerate the optimization and analysis of melt spinning processes.enindustrial textiles500 Naturwissenschaften und Mathematikconference paper