Моделювання штучної нейронної мережі для визначення та прогнозування працездатного стану джерел живлення безпілотних систем

Автор(и)
DOI:
https://doi.org/10.34169/2414-0651.2025.3(47).65-72
Ключові слова:
Array, Array, Array, Array, Array, Array, Array, Array
Анотація

This article presents a comparative analysis of two deep learning architectures developed for the simultaneous estimation of three key lithium-ion battery state indicators: state of charge (SoC), state of health (SoH), and remaining useful life (RUL). The research is based on experimental data from an open NASA dataset. The study's methodology encompasses a complete modeling cycle: from data preparation, time series formation, and model construction to their evaluation. Two models are considered: a recurrent neural network with a multi-head attention mechanism (Multi-head Attention RNN) and a hybrid architecture combining convolutional and recurrent neural networks with an attention mechanism (Hybrid CNN-RNN with Attention). Validation results demonstrated that the hybrid model achieved higher accuracy in estimating SoH and RUL, while the recurrent network model provided more stable SoC predictions. On the test dataset, leadership alternated depending on the indicator, highlighting the complementarity of the approaches. The scientific novelty of the work lies in the integrated application of two fundamentally different architectures to a single battery state prediction task using real-world data. This allowed for the identification of each model's advantages and limitations. The practical value of the results lies in their potential use for battery diagnostics and prognostics in power management systems for unmanned platforms, multi-agent decision-making systems, and technical health monitoring complexes. Future research will be directed towards optimizing the architectures, exploring new attention mechanisms, and integrating extended metrics to improve prediction accuracy and reliability under various operating conditions.

Посилання

Resolution of the Cabinet of Ministers of Ukraine No. 256 of March 24, 2023. On the implementation of a pilot project for the procurement of domestically produced unmanned systems for defense purposes. Available at: https://zakon.rada.gov.ua/laws/show/256-2023-%D0%BF#n66Text (accessed on 05.08.2025).

Plett, G.L. (2016). Battery Management Systems. Vol. II. Equivalent-Circuit Methods. Artech House. Pp. 70—73.

BU-105: Battery Definitions and what they mean. Website: https://batteryuniversity.com/article/bu-105-battery-definitions-and-what-they-mean. (Accessed 05.08.2025).

BU-901b: How to Measure the Remaining Useful Life of a Battery. Website: https://batteryuniversity.com/article/bu-901b-how-to-measure-the-remaining-useful-life-of-a-battery.

Hannan, M.A., Lipu, M.S.H., Hussain, A., Saad, M.H. & Ayob, A. (2018). Neural Network Approach for Estimating State of Charge of Lithium-Ion Battery Using Backtracking Search Algorithm. IEEE Access. № 6. Рр. 10069—10079. DOI: https://doi.org/10.1109/ACCESS.2018.2797976

Celik, B., Sandt, R., dos Santos, L.C.P. & Spatschek, R. (2022). Prediction of Battery Cycle Life Using Early-Cycle Data, Machine Learning and Data Management. Batteries. № 8. Р. 266. ttps://doi.org/10.3390(/batteries8120266. DOI: https://doi.org/10.3390/batteries8120266

Li-ion Battery Aging Datasets. NASA Open Data Portal. Website: https://data.nasa.gov/dataset/Li-ion-Battery-Aging-Datasets/uj5r-zjdb/about_data. (accessed: 05.08.2025).

Kaur, K., Garg, A. & Cui, X. Deep learning networks for capacity estimation for monitoring SOH of Li‐ion batteries for electric vehicles. https://doi.org/10.1002/er.6005. DOI: https://doi.org/10.1002/er.6005

Zheng, Y., Wen, H., Han, F., Yang, X. & Zhang, J. (2020). Study on SOC Estimation of Power Battery Based on LSTM Neural Network. Comput. Appl. Softw. № 37. Рр. 78—81.

Zh. Chen, Yue Peng, J. Shen and other. State of health estimation for lithium-ion batteries based on fragmented charging data and improved gated recurrent unit neural network. J. of Energy Storage. Vol. 115. 15 April 2025. P. 115952. https://doi.org/10.1016/j.est.2025.115952. DOI: https://doi.org/10.1016/j.est.2025.115952

Zh. Zhu, Q. Yang, X. & Liu, D. Gao. (2022). Attention-based CNN-BiLSTM for SOH and RUL estimation of lithium-ion batteries. J. of Algorithms & Computational Technology. Vol. 16. Pp. 1—11. https://doi.org/ 10.1177/17483026221130598. DOI: https://doi.org/10.1177/17483026221130598

Cr. Leo. The Math Behind LSTM. Towards data science. Available at: https://towardsdatascience.com/the-math-behind-lstm-9069b835289d.

F.-M. Zhao, De-Xin Gao, Y.-M. Cheng & Q. Yang. (2024). Application of state of health estimation and remaining useful life prediction for lithium-ion batteries based on AT-CNN-BiLSTM. Scientific Reports. № 14. Р. 29026. https://doi.org/10.1038/s41598-024-80421-2. DOI: https://doi.org/10.1038/s41598-024-80421-2

Slyusar, V. & Pochernin, S. (2025). Multi-agent energy management system for unmanned platforms. Prospects for the development of armaments and military equipment of the Ground Forces: Coll. of abstracts of the Intern. Scientific and Technical Conf. May 14–15, 2025. Lviv: NASV. 403 p.

Slyusar, V., Kozlov, V., Pochernin, S. & Nalapko, I. (2025). Conceptual foundations of the swarm employment of unmanned aerial vehicles as intelligent means of electronic warfare. Technology Audit and Production Reserves. № 3 (2 (83)). Рр. 71—80. https://doi.org/10.15587/2706-5448.2025.329989. DOI: https://doi.org/10.15587/2706-5448.2025.329989

Delta Information and Communication System. https://https://delta.mil.gov.ua/ (accessed on 05.08.2025).

Ministry of Defense: Ukraine is successfully integrating into NATO at the logistics level. Available at: Government portal. https://www.kmu.gov.ua/news/minoborony-ukraina-uspishno-intehruietsia-v-nato-na-rivni-lohistyky. (accessed: 05.08.2025).

Cover Image
Завантаження
Опубліковано
30.09.2025
Номер
Том 47 № 3 (2025): Озброєння та військова техніка
Розділ
ІНФОРМАЦІЙНІ СИСТЕМИ
Ліцензія

Авторське право (c) 2026 Сергій Почернін

Creative Commons License

Ця робота ліцензується відповідно до ліцензії Creative Commons Attribution 4.0 International License.

Як цитувати

Моделювання штучної нейронної мережі для визначення та прогнозування працездатного стану джерел живлення безпілотних систем. (2025). Озброєння та військова техніка, 47(3), 65-72. https://doi.org/10.34169/2414-0651.2025.3(47).65-72

Статті цього автора (цих авторів), які найбільше читають