Optimizing Energy Consumption at MAZ Using IoT Sensors and Neural Networks for Predictive Analysis
https://doi.org/10.35596/1729-7648-2026-32-1-45-50
Abstract
The energy intensity of the Belarusian mechanical engineering industry remains high – approximately 250 kWh per 1 BYR of output, making further improvements to energy efficiency strategically important. In the first half of 2025, MAZ OJSC achieved an energy savings rate of 6.9 %. The company still lacks an integrated real-time system that would analyze sensor data and predict equipment energy consumption for optimal mode planning and peak load reduction. This article presents an integrated model: IoT sensors collect data on power, vibration, and load, an LSTM neural network accurately forecasts energy consumption for several hours in advance, and an intelligent optimizer automatically redistributes production processes among favorable tariff zones. The system integrates with MAZ’s existing automated metering systems. The model will reduce energy consumption to 98.9 kWh/BYR with a phased implementation of the system in 2026, beginning with a pilot project in the second quarter of this year and achieving full effectiveness by 2027.
About the Authors
E. Polosko
Belarusian State University of Informatics and Radioelectronics
Belarus
Belarus
Polosko Ekaterina, Senior Lecturer at the Department of Economic Informatics
220013, Minsk, Brovki St., 6, Тel.: +375 25 530-89-43
O. Holda
Belarusian State University of Informatics and Radioelectronics
Belarus
Belarus
Cand. Sci. (Econ.), Associate Professor at the Software of Information Systems Department
Minsk
References
1. Galkin Yu. D., Dvornikov O. V., Chekhovsky V. A. (2022) Improved Dual-Gate JFET Model for Analog Integrated Circuits. Doklady BGUIR. 20 (3), 20–25 (in Russian).
2. Kravchenko O. A. (2020) Modification of a Stochastic Model for Optimizing Electricity Supply Costs at an Enterprise. Vestnik BSU. Series 1. (2), 45–52 (in Russian).
3. Vilkina M. V. (2022) Development of Tool Management Within the Framework of Industry 4.0. Rhythm of Mechanical Engineering. (3), 30–36 (in Russian).
4. Karpenko S. M., Karpenko N. V., Bezginnov G. Yu. (2022) Forecasting of Electricity Consumption at Mining Enterprises Using Statistical Methods. Mining Industry. (1), 82–88. DOI: 10.30686/1609-9192-2022-1-82-88 (in Russian).
5. Borovsky A. V., Yumenchuk A. A. (2024) Stochastic Electrical Load Model in the Residential Sector Using the Weibull Probability Density Function. Modeling, Optimization and Information Technologies. 12 (4), 1–18 (in Russian).
6. Hacker P. S., Schrank H. E. (1982) Range Distance Requirement for Measuring Low and Ultralow Sidelobe Antenna Patterns. IEEE Transactions on Antennas and Propagation. AP-30 (5), 956–966.
7. Chu M., Nguyen T., Pandey V., Zhou Y., Pham N H., Bar-Yoseph R., et al. (2019) Respiration Rate and Volume Measurements Using Wearable Strain Sensors. npj Digital Medicine. (2). DOI: 10.1038/s41746-019-0083-3.
8. Wen L., Zhou X., Yang Y. (2022) A Data-Driven Strategy Using Long Short-Term Memory Models and Reinforcement Learning to Predict Building Electricity Consumption. Applied Energy. 309. DOI: 10.1016/j.apenergy.2021.118437.
9. Seljom P., Kvalbein L., Hellemo L., Kaut M., Muсoz Ortiz M. (2021) Stochastic Modelling of Variable Renewables in Long-Term Energy Models: Dataset, Scenario Generation & Quality of Results. Energy. 236. DOI: 10.1016/j.energy.2021.121415.
10. VanderPlas J. (2021) Python Data Science Handbook: Data Science and Machine Learning with Python. Saint Petersburg, Piter Publ.
Review
For citations:
Polosko E., Holda O. Optimizing Energy Consumption at MAZ Using IoT Sensors and Neural Networks for Predictive Analysis. Digital Transformation. 2026;32(1):45-50. (In Russ.) https://doi.org/10.35596/1729-7648-2026-32-1-45-50
Views:
200
JATS XML
Email this article 