A.R. Mukhutdinov1, M.G. Efimov2

1 Kazan National Research Technological University (Kazan, Russia)
2 Kazan Innovative University (Kazan, Russia)
1muhutdinov@rambler.ru, 2jero07@bk.ru

The article presents a method for developing a neural network model for assessing the properties of polyvinyl chloride (PVC) composites using a universal express computational method. The universal express computational method for developing a neural network model for assessing the properties of PVC composites was tested with a relative prediction error of no more than 9%. A neural network model was created that can predict the operational parameters of PVC composites, such as density and plasticity, based on specified input characteristics. It is known that the mechanical properties of PVC composites are difficult to predict due to nonlinear relationships between the components, which makes this study particularly relevant. The research methodology included the development of a knowledge base, the creation of a uniform universal neural network model with general regression in the NeuroShell development environment and its subsequent testing. The initial data for the neural network model included six input characteristics, such as the content of virgin and recycled PVC, calcium carbonate, di-2-ethylhexyl phthalate, chlorinated paraffin wax and CaCO3 particle size. The neural network model was tested on twelve training examples and showed a relative prediction error of no more than 7,8%. The results of the study show that the developed ANN is able to accurately predict the properties of PVC composites, which is confirmed by a low relative error. The prospects of using neural network modeling to solve such problems lies in the fact that the proposed express method allows you to effectively create neural network models with minimal time and resource costs. The article is a significant contribution to the field of computer modeling of materials, offering a new approach to assessing the properties of complex composite materials using modern information technologies.

Mukhutdinov A.R., Efimov M.G. Validation of a universal computational express method for developing a neural network model to evaluate the properties of PVC composites. Nonlinear World. 2025. V. 23. № 2. P. 15–21. DOI: https:// doi.org/10.18127/ j20700970-202501-02 (In Russian)

1. Muhutdinov A.R., Marchenko G.N., Vahidova Z.R. Nejrosetevoe modelirovanie i optimizaciya slozhnyh processov i naukoemkogo teploenergeticheskogo oborudovaniya: monografiya. Kazan': Kazanskij gos. energ. un-t. 2011. 296 s. (In Russian).
2. Muhutdinov A.R., Efimov M.G., Vahidova Z.R. Nejrosetevoe modelirovanie processa nagreva bituminoznogo plasta i issledovanie vliyaniya razlichnyh faktorov. Avtomatizaciya i informatizaciya TEK. 2022. № 9 (590). S. 13–17 (In Russian).
3. Muhutdinov A.R., Efimov M.G. Universal'nye vychislitel'nye ekspress-metody dlya sozdaniya iskusstvennoj nejronnoj seti slozhnogo ob"ekta i innovacionnogo programmnogo modulya na ee osnove: monografiya. Minobrnauki Rossii, Kazan. nac. issled. tekhnol. un-t. Kazan': Izd-vo KNITU. 2022. 156 s. (In Russian).
4. Kibrete F., Trzepieciński T., Shimles H., Woldemichael D. Artificial Intelligence in Predicting Mechanical Properties of Composite Materials. Journal of Composites Science. 2023. V. 7(9). P. 364. DOI:10.3390/jcs7090364.
5. Sanli T., Gurses E., Coker D., Kayran A. Development of Artificial Neural Network Based Design Tool for Aircraft Engine Bolted Flange Connection Subject to Combined Axial and Moment Load. ASME 2017 International Mechanical Engineering Congress and Exposition. V001T03A023. DOI:10.1115/IMECE2017-70448.
6. Abueidda D., Almasri M., Ammourah R., Ravaioli U., Jasiuk I., Sobh N. Prediction and optimization of mechanical properties of composites using convolutional neural networks. Composite Structures. 2019. 10.1016/j.compstruct.2019.111264.
7. Osman A., Sapuan S., Mahdi E., Othman M. Prediction and Detection of Failures in Laminated Composite Materials Using Neural Networks. A Review. Polymers and Polymer Composites. 2006. V. 14(4). P. 431–442.
8. Nikahd M., Shokrollahi Y., Gholami K., Azamirad G. Deep Learning Techniques for Predicting Stress Fields in Composite Materials: A Superior Alternative to Finite Element Analysis. Journal of Composites Science. 2023. V. 7(8). DOI:10.3390/jcs7080311.
9. Reddy B.S., Premasudha M., Panigrahi B.B., Cho K.K., Reddy N.G.S. Modeling constituent–property relationship of polyvinylchloride composites by neural networks. Polymer Composites. 2020. V. 41(8). P. 3208–3217.
10. Asrofi M., Arpitha G.R., Rajkumar M., Hemath M., Nagaraja B.K., Neerakallu S., Thiagamani S.M.K. An artificial neural network prediction on physical, mechanical, and thermal characteristics of giant reed fiber reinforced polyethylene terephthalate composite. Journal of Industrial Textiles. 2022. V. 51(1). P. 769–803.
11. Kibrete F., Trzepieciʼnski T., Gebremedhen H.S., Woldemichael D.E. Artificial Intelligence in Predicting Mechanical Properties of Composite Materials. Journal of Composites Science. 2023. V. 7. P. 364. DOI:10.3390/jcs7090364.
12. Altarazia S., Ammouria M., Hijazib A. Artificial neural network modeling to evaluate polyvinylchloride composites’ properties Industrial Engineering Department. Computational Materials Science, Amman, Jordan. 2018. № 153. P. 1–9.