Machine learning modeling of melt-spinning for yarn property prediction

Mehran Dadgar, Kamyar Noshad,
Machine learning modeling of melt-spinning for yarn property prediction,
Results in Materials,
Volume 28,
2025,
100776,
ISSN 2590-048X,
https://doi.org/10.1016/j.rinma.2025.100776.
(https://www.sciencedirect.com/science/article/pii/S2590048X25001219)
Abstract: The quality and performance of yarn in textile production are critically dependent on a range of machine operating parameters. A precise understanding of how these parameters influence yarn properties is essential for both researchers and industrial practitioners aiming to optimize production efficiency and product performance. This study investigates eleven key input parameters—primarily tension, temperature, and speed across various sections of the spinning machine—and their corresponding effects on yarn characteristics, including elongation, tenacity, modulus, and shrinkage. Such an analysis offers a practical framework for industry professionals, enabling them to predict the impact of modifications in each input parameter on specific yarn properties. Acquiring this knowledge facilitates the attainment of target yarn specifications with reduced production time, minimized material waste, and without repeated trial-and-error testing on the machine. These insights provide a systematic approach to process control, supporting improved yarn quality and optimized manufacturing operations.
Keywords: BCF polyester yarn; Machine learning; Property prediction; Fiber engineering