I.I. Naumov1, D.Yu. Konshin2, K.A. Moroz3
1–3 Don State Technical University (Rostov-on-Don, Russia)

1 naumov-85@yandex.ru, 2 danill.konshin@yandex.ru, 3 kmoroz@donstu.ru

Modern manufacturing systems transitioning to advanced production standards require a high degree of adaptability, transparency in decision-making, and fault tolerance. Traditional automation methods, based on rigid algorithms and centralized control logic, prove insufficiently flexible. They underutilize the potential of big data, scale poorly, struggle with network latency or connectivity disruptions, and require significant effort to integrate new types of equipment. The limited intelligence of local controllers complicates autonomous operation, while the opacity of algorithms hinders decision-making by personnel and diagnostic processes.

The objective of the article is to develop a conceptual architecture of an intelligent distributed production control system based on embedded devices with edge artificial intelligence capabilities, multi-agent approach, and explainable AI technologies. The system must ensure local loop autonomy, seamless interaction with cloud services, adaptation to changing equipment operating conditions, and transparency of algorithms for engineering and operational personnel.

The architecture has been proposed that combines the computing capabilities of local nodes and cloud components with distributed logic based on a multi-agent system. Edge devices embedded in production equipment gain the ability to perform local data preprocessing, run machine learning models, respond instantaneously to network delays, and maintain an autonomous safety loop.

Implementing the proposed architecture enables enterprises to enhance the resilience of manufacturing processes to internal and external disruptions, improve the quality and speed of decision-making, reduce reliance on centralized control systems, and increase energy and resource efficiency.

Naumov I.I., Konshin D.Yu., Moroz K.A. Concept of adaptive production automation architecture with local and cloud components based on edge artificial intelligence and a multi-agent system. Neurocomputers. 2026. V. 28. № 2. P. 70–77. DOI: https://doi.org/10.18127/ j19998554-202602-06 (in Russian)

1. Gao R., Krüger J., Merklein M. et al. Artificial intelligence in manufacturing: State of the art, perspectives, and future directions. CIRP Annals. 2024. V. 73. DOI: 10.1016/j.cirp.2024.04.101.
2. De Gasperis G., Facchini S.D. A comparative study of rule-based and data-driven approaches in industrial monitoring. arXiv e-prints. 2025. arXiv:2509.15848. DOI: 10.48550/arXiv.2509.15848.
3. Ucar A., Karakose M., Kırımça N. Artificial intelligence for predictive maintenance applications: Key components, trustworthiness, and future trends. Applied Science. 2024. V. 14. № 898. DOI: 10.3390/app14020898.
4. Tziouvaras A., Foukalas F. Edge AI for Industry 4.0: An Internet of Things approach. Proceedings of 24th Pan-Hellenic Conference on Informatics. 2020. P. 121–126. DOI: 10.1145/3437120.3437289.
5. Annapareddy S. Edge AI for real-time fault detection in embedded systems. International Journal of Emerging Trends in Computer Science and Information Technology. 2025. P. 235–238. DOI: 10.56472/ICCSAIML25-129.
6. Li B., Palayew S., Li F. et al. PCBDet: An efficient deep neural network object detection architecture for automatic PCB component detection on the edge. Electronics Letters. 2024. V. 60. № 2. Article e13042. DOI: 10.1049/ell2.13042.
7. Areo G. Edge AI for real-time industrial automation: Architecture, challenges, and case studies. 2025.
8. Grant C., Wallace F., Mackay L. et al. Edge AI for predictive maintenance in industrial IoT environments. 2024.
9. Huang K.Y., Hsu Y.L., Chung C.L. et al. Enhancing healthcare AI stability with edge computing and machine learning for extubation prediction. Science Reports. 2025. V. 15. Article 17858. DOI: 10.1038/s41598-025-02317-z.
10. Montes-Sánchez J.M., Uwate Y., Nishio Y. et al. Predictive maintenance edge artificial intelligence application study using recurrent neural networks for early aging detection in peristaltic pumps. IEEE Transactions on Reliability. 2025. V. 74. № 3. P. 3730–3744. DOI: 10.1109/TR.2024.3488963.
11. Artiushenko V., Lang S., Lerez C. et al. Resource-efficient Edge AI solution for predictive maintenance. Procedia Computer Science. 2024. V. 232. P. 348–357. DOI: 10.1016/j.procs.2024.01.034.
12. Li W., Li T. Comparison of deep learning models for predictive maintenance in industrial manufacturing systems using sensor data. Science Reports. 2025. V. 15. Article 23545. DOI: 10.1038/s41598-025-08515-z.
13. Jiménez-Bravo D., Bajo J., González-Pachón J. et al. Multi-agent system architecture for winter road maintenance: a real Spanish case study. Knowledge and Information Systems. 2024. V. 66. P. 1–19. DOI: 10.1007/s10115-024-02128-0.
14. Cai H., You S., Wu J. Agent-based distributed demand response in district heating systems. Applied Energy. 2020. V. 262. Article 114403. DOI: 10.1016/j.apenergy.2019.114403.
15. Baroud S.Y., Yahaya N.A., Elzamly A. Cutting-edge AI approaches with MAS for PdM in Industry 4.0: Challenges and future directions. Journal of Applied Data Sciences. 2024. V. 5. № 2. P. 455–473. DOI: 10.47738/jads.v5i2.196.
16. Rodríguez M., Kubler S., de Giorgio A. et al. Multi-agent deep reinforcement learning based predictive maintenance on parallel machines. Robotics and Computer-Integrated Manufacturing. 2022. V. 78. Article 102406. DOI: 10.1016/j.rcim.2022.102406.
17. Makris S., Alexopoulos K., Michalos G. et al. An agent-based system for automated configuration and coordination of robotic operations in real time – A case study on a car floor welding process. Journal of Manufacturing and Materials Processing. 2020. V. 4. № 3. Article 95. DOI: 10.3390/jmmp4030095.
18. Arshad A., Ekström J., Lehtonen M. Multi-agent based distributed voltage regulation scheme with grid-tied inverters in active distribution networks.0 Electric Power Systems Research. 2018. V. 160. P. 180–190. DOI: 10.1016/j.epsr.2018.02.015.