Development of Predictive Models for Elastomeric Materials Based on Machine Learning Methods and Analysis of the Influence of Formulation Components
( Pp. 25-33)

The paper presents an approach to constructing predictive models for the physical and mechanical properties of elastomeric composites using machine learning methods. The relevance of the study is driven by the need to accelerate the development of new materials and reduce the labor intensity of full-scale experiments. An automated machine learning algorithm is proposed, encompassing stages of input data unification, feature space formation, and comparative analysis of regression models (Random Forest, Gradient Boosted Decision Trees, Gradient Tree-Boosting Tweedie, Poisson Regression, Light Gradient Boosting Machine и Stochastic Dual Coordinate Ascent). During experimental validation on datasets containing formulation data with varying content of sulfur, natural rubber (NR), and zinc oxide, predictions were made for theoretical density, Karrer plasticity, brittleness temperature, and curing temperature. It was established that ensemble methods demonstrate the highest predictive capability; however, model accuracy significantly depends on sample representativeness. Intervals of the studied parameters (particularly the 140–150 °C range for curing temperature) characterized by increased prediction uncertainty were identified, requiring additional algorithm calibration. The obtained results confirm the effectiveness of the proposed approach for formulation optimization and identification of hidden dependencies in the “composition-property” system.

Maslova M.A., Kablov V.F., and Rybanov A.A. Development of predictive models for elastomeric materials based on machine learning methods and analysis of the influence of formulation components. Computational Nanotechnology. 13, 2 (2026), 25–33 DOI: 10.33693/2313-223X-2026-13-2-25-33. EDN: ZWDLGN

Abgaryan K.K., Zagordan N.L., Mochalova Yu.D. Application of machine learning methods for analyzing elastic characteristics of multilayer composite materials. In: Mathematical methods of pattern recognition. Abstracts of the 21st All-Russian Conference with International Participation. Moscow, 2023. Pp. 224–226. EDN: KOHJRD.
Kablov V.F., Rybanov A.A., Maslova M.A. Evolution of rubber compound formulation management information systems: From databases to intelligent repositories with integration of artificial intelligence modules. In: Rubber-2025: Traditions and innovations. Proceedings of the XIII All-Russian Conference. Moscow, 2025. Pp. 23–24. EDN: LWBVSC.
Koltsov V.B., Potemkin A.Ya., Konoplin N.A. et al. Physical and chemical modeling of technological processes – a modern way of creating new alternative technologies. Prirodoobustrojstvo. 2010. No. 3. Pp. 98–102. (In Rus.). EDN: MUQAXN.
Maslova M.A. Software and information system for managing the reference database on ingredients and properties of elastomeric material formulations. In: Collection materials of XXIX Regional Conference of Young Scientists and Researchers of the Volgograd Region (Volgograd, September 16 – November 15, 2024). S.V. Kuzmin (ed.). Volgograd, 2024. Pp. 121–122. EDN: DBRQRE.
Eleas A.K., Aboudi E.H. Expectation parameters in the Poisson mixture regression model for latent class by applying genetic algorithm and maximization algorithm. Journal of Economics and Administrative Sciences (JEAS). 2024. Vol. 30. No. 140. Pp. 434–449. DOI: 10.33095/xammnc51.
Guolin Ke, Qi Meng, Finley T. et al. LightGBM: A highly efficient gradient boosting decision tree. In: 31st Conference on Neural Information Processing Systems (NIPS 2017). Long Beach, CA, USA, 2017. Pp. 3149–3157.
Xueheng Qiu, Le Zhang, Suganthan P.N., Amaratunga G.A. Oblique random forest ensemble via Least Square Estimation for time series forecasting. Information Sciences. 2017. Vol. 420. Pp. 249–262. DOI: 10.1016/j.ins.2017.08.060.
Shai Shalev-Shwartz, Tong Zhang. Stochastic dual coordinate ascent methods for regularized loss minimization. Journal of Machine Learning Research. 2013. Vol. 14. No. 1. Pp. 567–599. DOI: 10.5555/2567709.2502598.
Yi Yang, Wei Qian, Hui Zou. Insurance premium prediction via gradient tree-boosted Tweedie compound Poisson models. Journal of Business & Economic Statistics, 2018. Vol. 36. Issue 3. Pp. 456–470. DOI: 10.1080/07350015.2016.1200981.
Feng Zhou, Qunzhi Zhang, Sornette D., Liu Jiang. Cascading logistic regression onto gradient boosted decision trees for forecasting and trading stock indices. Applied Soft Computing. 2019. Vol. 84. Art. 105747. DOI: 10.1016/j.asoc.2019.105747.

machine learning, materials property prediction, elastomeric materials, information model, materials science.