2026, Vol. 11, No. 2. - go to content...

Permanent address of this page - https://kostumologiya.ru/en/28tlkl226.html

Метаданные этой статьи так же доступны на русском языке

Full article in PDF format (file size: 966.9 KB)

For citation:

Orlov A.V., Shustov Yu.S. Prediction of physical and mechanical properties of needle-punched nonwoven materials based on neural network modeling. Journal of Clothing Science. 2026; 11(2). Available at: https://kostumologiya.ru/PDF/28TLKL226.pdf (in Russian).

Prediction of physical and mechanical properties of needle-punched nonwoven materials based on neural network modeling

Orlov Alexander Vasilievitch
Russian State University named A.N. Kosygin (Technologies. Design. Art), Moscow, Russia
E-mail: metroid.prime@yandex.ru
RSCI: https://elibrary.ru/author_profile.asp?id=1157673

Shustov Yuriy Stepanovitch
Russian State University named A.N. Kosygin (Technologies. Design. Art), Moscow, Russia
E-mail: 6145263@mail.ru
RSCI: https://elibrary.ru/author_profile.asp?id=484967

Abstract. The article presents the results of a study aimed at developing a neural network-based system for predicting the physical and mechanical properties of needle-punched thermally bonded nonwoven materials for footwear applications made from regenerated fibrous raw materials. The relevance of the work is associated with the need for preliminary assessment of the breaking load and elongation at break of nonwoven fabrics at the stage of selecting the fiber composition and structural and technological parameters, since carrying out a full range of pilot production runs and laboratory tests requires significant time and material costs. The purpose of the study was to construct and verify neural network models that make it possible to calculate the physical and mechanical properties of the material based on specified parameters of the raw material, fabric structure, and needle-punching process. The practical implementation of prediction was carried out in the MATLAB software environment using the Deep Learning Toolbox module. The authors formed an experimental database based on 16 samples of needle-punched nonwoven materials. Surface density, fiber composition, fabric thickness, needle penetration depth, and punch density were used as input parameters. Breaking load and elongation at break in the longitudinal and transverse directions were considered as output parameters. A separate neural network model was constructed for each output parameter. The authors assessed the prediction accuracy and compared the model error with experimental variability. A high agreement between calculated and experimental data was established. It is shown that the developed neural network-based prediction system can be used as a calculation tool in the preliminary design of needle-punched nonwoven materials and in the selection of rational structural and technological parameters.

Keywords: needle-punched nonwoven materials; regenerated raw materials; method for predicting physical and mechanical properties; neural network modeling; artificial neural networks; property prediction; MATLAB

Download article in PDF format