Technical Diagnostics and Non-Destructive Testing, 2026, №03

2026 №03 (03)

2026 №03 (01)

---

"Tekhnichna Diahnostyka ta Neruinivnyi Kontrol" (Technical Diagnostics and Non-Destructive Testing) #3, 2026, pp. 48-53

Neural network identification of control object in the system of automated asphalt concrete laying process

A.G. Protasov, Y.V. Steshenko

National Technical University of Ukraine «Igor Sikorsky Kyiv Polytechnic Institute». 37 Beresteysky Ave., 03056, Kyiv, Ukraine. E-mail: a.g.protasov@gmail.com

The article is devoted to the improvement of a closed-loop automated control system for the asphalt concrete paving process by developing an algorithm for neural network identification of the control object and its integration into the system circuit. Based on a multilayer perceptron, a logical structure has been formed that reproduces the nonlinear complex influence of thermal and mechanical factors on the formation of coating density. The proposed approach overcomes the key problem of traditional analytical models, namely their low adaptability to changing field conditions and the need for constant laboratory recalculation and calibration of empirical coefficients when switching to a different mixture composition or equipment type. It was possible to reduce the average absolute error in determining the density of the asphalt concrete layer on the test sample owing to the procedure for standardizing the input features and the use of regularization mechanisms. This provides increased accuracy and reliability of control, guaranteeing reliable detection of potential areas of under compaction even at the stage of placing the mixture. At the same time, the high computational efficiency of the algorithm has been proven, since the experimentally recorded average inference time of the neural network model is 4.038 ms, which confirms the suitability of the model for realtime systems. A comparative analysis of statistical accuracy metrics was conducted for the basic analytical model, classical linear regression, and the constructed multilayer perceptron to numerically confirm the advantages of the neural network approach. The convergence of the actual and predicted by the neural network model values of the density of asphalt concrete paving is considered. The nature of the distribution of points confirms that the developed MLP (Multilayer Perceptron) structure is capable of reproducing with high accuracy the complex relationships between the input technological factors and the final quality indicator without systematic biases or accumulation of errors at the range edges. 8 Ref., 1 Tabl., 2 Fig.
Keywords: automatic control, asphalt concrete paving, neural network identification, multilayer perceptron, inference time

Received: 12.08.2026
Received in revised form: 28.08.2026
Accepted: 06.09.2026
Posted online: 22.09.2026

References

1. Forgione, M., Piga, D. (2021) Continuous-time system identification with neural networks: Model structures and fitting criteria. European J. of Control, 59, 69–81. DOI: https://doi.org/10.1016/j.ejcon.2021.01.008
2. Dong, A., Starr, A., Zhao, Y. (2023) Neural network-based parametric system identification: A review. Intern. J. of Systems Science, 54(13), 2676–2688. DOI: https://doi.org/10.1080/00207721.2023.2241957
3. Steshenko, Y. V., Protasov, A.G. (2025) Automated control system for the technological process of asphalt concrete coating laying. Technical Diagnostics and Non-Destructive Testing, 2, 30–35 [in Ukrainian]. DOI: https://doi.org/10.37434/tdnk2025.02.05
4. Steshenko, Y.V., Protasov, A.G. (2026) Intelligent system for adaptive control of the technological process of laying asphalt concrete based on neural network models. KPI Science News, 142(1), 32–39 [in Ukrainian]. DOI: https://doi.org/10.20535/kpisn.2026.1.350095
5. Commuri, S., Mai, A.T., Zaman, M. (2011) Neural networkbased intelligent compaction analyzer for estimating compaction quality of hot asphalt mixes. J. of Construction Engineering and Management, 137(9), 634–644. DOI: https://doi.org/10.1061/(asce)co.1943-7862.0000343
6. Xue, Z., Cao, W., Liu, S. et al. (2021) Artificial neural networkbased method for real-time estimation of compaction quality of hot asphalt mixes. Appl. Sci., 11(15), 7136. DOI: https://doi.org/10.3390/app11157136
7. Yu, S., Shen, S., Lu, M. (2023) Data sensing and compaction condition modeling for asphalt pavements. Automation in Construction, 154, 105021. DOI: https://doi.org/10.1016/j.autcon.2023.105021
8. Hunt, K.J., Sbarbaro, D. (1991) Neural networks for nonlinear internal model control. IEE Proc. D: Control Theory and Applications, 138(5), 431–438. DOI: https://doi.org/10.1049/ip-d.1991.0059

---

This article is licensed under a Creative Commons Attribution Non-Commercial 4.0 International (CC-BY-NC).

---

Suggested Citation

A.G. Protasov, Y.V. Steshenko (2026) Neural network identification of control object in the system of automated asphalt concrete laying process. Technical Diagnostics and Non-Destructive Testing, 03, 48-53.

---

Advertising in this issue:

---