Advancing Polymeric Yarn Manufacturing: A Hybrid Modelling Approach Combining AI and Molecular Dynamics

This study introduces a novel hybrid modelling framework for optimizing polymeric yarn extrusion process. The synthetic textile industry faces significant challenges, including high costs and inefficiencies due to the complex interplay between material properties and manufacturing conditions. Traditional trial-and-error methods are costly and inefficient. The proposed hybrid model combines atomistic simulations, surrogate modelling, and machine learning techniques, specifically neural networks, to predict and control the microstructural and mechanical properties of polymeric yarns. The atomistic model, implemented using coarse-grained molecular dynamics, captures essential molecular interactions and links crystallinity and orientation to mechanical properties. A surrogate model, trained on molecular dynamics data, provides real-time predictions of mechanical properties such as tenacity and stress-strain behaviour. The final hybrid model integrates these approaches, using neural networks to predict microstructural properties from process variables and mechanical properties from microstructural data. Validation through experimental comparisons demonstrates the model ability to enhance production efficiency.