gscienceMining Science and Technology (Russia)Горные науки и технологии2500-0632The National University of Science and Technology MISiIS (NUST MISIS)10.17073/2500-0632-2026-01-1070gscience-1070Research ArticleDIGITAL TECHNOLOGIES AND ARTIFICIAL INTELLIGENCEЦИФРОВЫЕ ТЕХНОЛОГИИ И ИСКУССТВЕННЫЙ ИНТЕЛЛЕКТNeural network models for short-term electricity consumption forecasting at a mining enterprise participating in the wholesale electricity marketНейросетевые модели краткосрочного прогнозирования электропотребления горнопромышленного предприятия для оптового рынка электроэнергииhttps://orcid.org/0009-0008-7117-7029ЕматинЕ. А.EmatinE. A.

Евгений Александрович Ематин – аспирант кафедры энергетики и энергоэффективности горной промышленности

г. Москва

Scopus ID 59012304200

SPIN 2464-1230

Evgenii A. Ematin – PhD-Student, Department of Energy and Energy Efficiency of the Mining Industry

Moscow

Scopus ID 59012304200

SPIN 2464-1230

vjeka@mail.ruhttps://orcid.org/0000-0002-3469-3648КарпенкоС. М.KarpenkoS. M.

Сергей Михайлович Карпенко – кандидат технических наук, доцент кафедры энергетики и энергоэффективности горной промышленности

г. Москва

Scopus ID 57225144783

SPIN 3950-0362

Sergey M. Karpenko – Cand. Sci. (Eng.), Associate Professor of the Department of Energy and Energy Efficiency of the Mining Industry

Moscow

Scopus ID 57225144783

SPIN 3950-0362

ksm_62@mail.ruУниверситет науки и технологий МИСИСРоссияUniversity of Science and Technology MISISRussian Federation20260710202600Online_First1070Copyright © Ematin E.A., Karpenko S.M., 20262026Ематин Е.А., Карпенко С.М.Ematin E.A., Karpenko S.M.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://mst.misis.ru/jour/article/view/1070

Electricity consumption at mining enterprises is markedly nonstationary because of intermittent equipment operation and the rapidly varying loads associated with electric-arc steelmaking. Deviations between actual consumption and day-ahead bids in the Wholesale Electricity and Capacity Market (WECM) entail substantial penalties, making conventional statistical methods insufficiently accurate and creating a need for adaptive neural network approaches. This study developed and evaluated a neural network model for short-term electricity consumption forecasting designed to minimize WECM penalties. Six architectures (RNN, GRU, LSTM, BiLSTM, CNN–LSTM, and attention-based models) were compared under two multi-step forecasting strategies: direct and recursive. Hyperparameters were selected using Hyperband. The study used 2022–2023 industrial data from a large mining and processing enterprise (17,520 hourly measurements). The results showed that the GRU architecture with the direct strategy provided the highest accuracy (MAPE of 5.44%) and the greatest reduction in penalty payments relative to the seasonal naïve model (21.25%), equivalent to RUB 38.6 million in quarterly savings during the test period (Q4 2023). Evidence of excessive model complexity was observed: simpler recurrent networks outperformed attention-based models on a limited dataset. The recursive strategy was inferior to the direct strategy because of cumulative error propagation and the inability to use stochastic exogenous PCA factors. The findings support the transition from expert judgment and conventional statistical methods to neural network approaches at mining enterprises with rapidly varying loads.

Электропотребление горнопромышленных предприятий характеризуется выраженной нестационарностью из-за повторно-кратковременных режимов оборудования и резкопеременной нагрузки электросталеплавильного производства. Отклонение фактического потребления от плановых заявок на оптовом рынке электроэнергии и мощности (ОРЭМ) влечёт значительные штрафные санкции, что делает традиционные статистические методы недостаточно точными и требует адаптивных нейросетевых подходов.  Разработана и обоснована нейросетевая модель краткосрочного прогнозирования электропотребления для минимизации штрафов ОРЭМ. Проведён сравнительный анализ шести архитектур (RNN, GRU, LSTM, BiLSTM, CNN-LSTM, с механизмом внимания) в двух стратегиях многошагового прогнозирования (прямая Direct и рекурсивная Recursive). Гиперпараметры подобраны методом Hyperband. Исследование выполнено на промышленных данных за 2022–2023 гг. (17 520 почасовых измерений) крупного горно-перерабатывающего предприятия. Результаты показали, что архитектура GRU с прямой стратегией обеспечивает наилучшую точность (MAPE 5,44 %) и максимальную экономию штрафных платежей – 21,25 % относительно сезонной наивной модели, что на тестовом периоде (IV квартал 2023 г.) составляет 38,6 млн руб. квартального эффекта. Выявлен эффект избыточной сложности: простые рекуррентные сети превосходят модели с механизмами внимания на данных ограниченного объёма. Рекурсивная стратегия уступает прямой из-за кумулятивного накопления ошибки и невозможности использования стохастических экзогенных PCA-факторов. Выводы подтверждают целесообразность перехода от экспертно-статистических методов к нейросетевым подходам для горнопромышленных предприятий с резкопеременными нагрузками.

горнопромышленное предприятиепрогнозирование электропотреблениярезкопеременные нагрузкимногошаговое прогнозированиеглубокое обучениерекуррентные нейронные сетиGRU (управляемый рекуррентный блок)LSTM (долгая краткосрочная память)снижение штрафных платежейmining enterpriseelectricity consumption forecastingrapidly varying loadsmulti-step forecastingdeep learningrecurrent neural networksgated recurrent unit (GRU)long short-term memory (LSTM)reduction in penalty paymentsReferencesKlyuev R. V., Bosikov I. I., Gavrina O. A. et al. Improving the energy efficiency of technological equipment at mining enterprises. Advances in Intelligent Systems and Computing. 2021;1258:262–271. https://doi.org/10.1007/978-3-030-57450-5_24Klyuev R. V., Bosikov I. I., Gavrina O. A. et al. Improving the energy efficiency of technological equipment at mining enterprises. Advances in Intelligent Systems and Computing. 2021;1258:262–271. https://doi.org/10.1007/978-3-030-57450-5_24Hyndman R. J., Koehler A. B. Another look at measures of forecast accuracy. International Journal of Forecasting. 2006;22(4):679–688. https://doi.org/10.1016/j.ijforecast.2006.03.001Hyndman R. J., Koehler A. B. Another look at measures of forecast accuracy. International Journal of Forecasting. 2006;22(4):679–688. https://doi.org/10.1016/j.ijforecast.2006.03.001Hochreiter S., Schmidhuber J. Long short-term memory. Neural Computation. 1997;9(8):1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735Hochreiter S., Schmidhuber J. Long short-term memory. Neural Computation. 1997;9(8):1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735Cho K., van Merrienboer B., Gulcehre C. et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation. In: Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP). Doha: ACL; 2014. Pp. 1724–1734. https://doi.org/10.3115/v1/D14-1179Cho K., van Merrienboer B., Gulcehre C. et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation. In: Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP). Doha: ACL; 2014. Pp. 1724–1734. https://doi.org/10.3115/v1/D14-1179Bengio Y., Simard P., Frasconi P. Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks. 1994;5(2):157–166. https://doi.org/10.1109/72.279181Bengio Y., Simard P., Frasconi P. Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks. 1994;5(2):157–166. https://doi.org/10.1109/72.279181Vaswani A., Shazeer N., Parmar N. et al. Attention is all you need. In: Advances in Neural Information Processing Systems. 2017;30:5998–6008. https://doi.org/10.48550/arXiv.1706.03762Vaswani A., Shazeer N., Parmar N. et al. Attention is all you need. In: Advances in Neural Information Processing Systems. 2017;30:5998–6008. https://doi.org/10.48550/arXiv.1706.03762Lim B., Arik S. O., Loeff N., Pfister T. Temporal Fusion Transformers for interpretable multi-horizon time series forecasting. International Journal of Forecasting. 2021;37(4):1748–1764. https://doi.org/10.1016/j.ijforecast.2021.03.012Lim B., Arik S. O., Loeff N., Pfister T. Temporal Fusion Transformers for interpretable multi-horizon time series forecasting. International Journal of Forecasting. 2021;37(4):1748–1764. https://doi.org/10.1016/j.ijforecast.2021.03.012Zhou H., Zhang S., Peng J. et al. Informer: Beyond efficient transformer for long sequence time-series forecasting. In: Proceedings of the AAAI Conference on Artificial Intelligence. 2021;35(12):11106–11115. https://doi.org/10.1609/aaai.v35i12.17325Zhou H., Zhang S., Peng J. et al. Informer: Beyond efficient transformer for long sequence time-series forecasting. In: Proceedings of the AAAI Conference on Artificial Intelligence. 2021;35(12):11106–11115. https://doi.org/10.1609/aaai.v35i12.17325Wu H., Xu J., Wang J., Long M. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting. In: Advances in Neural Information Processing Systems. 2021;34:22419–22430. https://doi.org/10.48550/arXiv.2106.13008Wu H., Xu J., Wang J., Long M. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting. In: Advances in Neural Information Processing Systems. 2021;34:22419–22430. https://doi.org/10.48550/arXiv.2106.13008Zeng A., Chen M., Zhang L., Xu Q. Are transformers effective for time series forecasting? In: Proceedings of the AAAI Conference on Artificial Intelligence. 2023;37(9):11121–11128. https://doi.org/10.1609/aaai.v37i9.26317Zeng A., Chen M., Zhang L., Xu Q. Are transformers effective for time series forecasting? In: Proceedings of the AAAI Conference on Artificial Intelligence. 2023;37(9):11121–11128. https://doi.org/10.1609/aaai.v37i9.26317Zhang G. P., Qi M. Neural network forecasting for seasonal and trend time series. European Journal of Operational Research. 2005;160(2):501–514. https://doi.org/10.1016/j.ejor.2003.08.037Zhang G. P., Qi M. Neural network forecasting for seasonal and trend time series. European Journal of Operational Research. 2005;160(2):501–514. https://doi.org/10.1016/j.ejor.2003.08.037Ben Taieb S., Atiya A. F. A bias and variance analysis for multistep-ahead time series forecasting. IEEE Transactions on Neural Networks and Learning Systems. 2016;27(1):62–76. https://doi.org/10.1109/TNNLS.2015.2411629Ben Taieb S., Atiya A. F. A bias and variance analysis for multistep-ahead time series forecasting. IEEE Transactions on Neural Networks and Learning Systems. 2016;27(1):62–76. https://doi.org/10.1109/TNNLS.2015.2411629Карпенко С. М., Карпенко Н. В., Безгинов Г. Ю. Прогнозирование электропотребления на горнопромышленных предприятиях с использованием статистических методов. Горная промышленность. 2022;(1):82–88. https://doi.org/10.30686/1609-9192-2022-1-82-88Karpenko S. M., Karpenko N. V., Bezginov G. Y. Forecasting of power consumption at mining enterprises using statistical methods. Russian Mining Industry. 2022;(1):82–88. (In Russ.) https://doi.org/10.30686/1609-9192-2022-1-82-88Hewamalage H., Bergmeir C., Bandara K. Recurrent neural networks for time series forecasting: current status and future directions. International Journal of Forecasting. 2021;37(1):388–427. https://doi.org/10.1016/j.ijforecast.2020.06.008Hewamalage H., Bergmeir C., Bandara K. Recurrent neural networks for time series forecasting: current status and future directions. International Journal of Forecasting. 2021;37(1):388–427. https://doi.org/10.1016/j.ijforecast.2020.06.008Sangiorgio M., Dercole F. Robustness of LSTM neural networks for multi-step forecasting of chaotic time series. Chaos, Solitons & Fractals. 2020;139:110045. https://doi.org/10.1016/j.chaos.2020.110045Sangiorgio M., Dercole F. Robustness of LSTM neural networks for multi-step forecasting of chaotic time series. Chaos, Solitons & Fractals. 2020;139:110045. https://doi.org/10.1016/j.chaos.2020.110045Zawodnik V., Schwaiger F., Sorger M., Kienberger T. Tackling uncertainty: forecasting the energy consumption and demand of an electric arc furnace with limited knowledge on process parameters. Energies. 2024;17(6):1326. https://doi.org/10.3390/en17061326Zawodnik V., Schwaiger F., Sorger M., Kienberger T. Tackling uncertainty: forecasting the energy consumption and demand of an electric arc furnace with limited knowledge on process parameters. Energies. 2024;17(6):1326. https://doi.org/10.3390/en17061326Ben Taieb S., Bontempi G., Atiya A. F., Sorjamaa A. A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition. Expert Systems with Applications. 2012;39(8):7067–7083. https://doi.org/10.1016/j.eswa.2012.01.039Ben Taieb S., Bontempi G., Atiya A. F., Sorjamaa A. A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition. Expert Systems with Applications. 2012;39(8):7067–7083. https://doi.org/10.1016/j.eswa.2012.01.039Li H., Li S., Wu Y. et al. Short-term power load forecasting for integrated energy system based on a residual and attentive LSTM-TCN hybrid network. Frontiers in Energy Research. 2024;12:1384142. https://doi.org/10.3389/fenrg.2024.1384142Li H., Li S., Wu Y. et al. Short-term power load forecasting for integrated energy system based on a residual and attentive LSTM-TCN hybrid network. Frontiers in Energy Research. 2024;12:1384142. https://doi.org/10.3389/fenrg.2024.1384142Bergstra J., Bengio Y. Random search for hyper-parameter optimization. Journal of Machine Learning Research. 2012;13(1):281–305. URL: https://www.jmlr.org/papers/v13/bergstra12a.htmlBergstra J., Bengio Y. Random search for hyper-parameter optimization. Journal of Machine Learning Research. 2012;13(1):281–305. URL: https://www.jmlr.org/papers/v13/bergstra12a.htmlMarcellino M., Stock J. H., Watson M. W. A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series. Journal of Econometrics. 2006;135(1–2):499–526. https://doi.org/10.1016/j.jeconom.2005.07.020Marcellino M., Stock J. H., Watson M. W. A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series. Journal of Econometrics. 2006;135(1–2):499–526. https://doi.org/10.1016/j.jeconom.2005.07.020Li L., Jamieson K., DeSalvo G., Rostamizadeh A., Talwalkar A. Hyperband: A novel bandit-based approach to hyperparameter optimization. Journal of Machine Learning Research. 2018;18(185):1–52. https://doi.org/10.48550/arXiv.1603.06560Li L., Jamieson K., DeSalvo G., Rostamizadeh A., Talwalkar A. Hyperband: A novel bandit-based approach to hyperparameter optimization. Journal of Machine Learning Research. 2018;18(185):1–52. https://doi.org/10.48550/arXiv.1603.06560Chung J., Gulcehre C., Cho K., Bengio Y. Empirical evaluation of gated recurrent neural networks on sequence modeling. In: NIPS 2014 Workshop on Deep Learning. 2014. https://doi.org/10.48550/arXiv.1412.3555Chung J., Gulcehre C., Cho K., Bengio Y. Empirical evaluation of gated recurrent neural networks on sequence modeling. In: NIPS 2014 Workshop on Deep Learning. 2014. https://doi.org/10.48550/arXiv.1412.3555Bai S., Kolter J. Z., Koltun V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint: 1803.01271. 2018. https://doi.org/10.48550/arXiv.1803.01271Bai S., Kolter J. Z., Koltun V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint: 1803.01271. 2018. https://doi.org/10.48550/arXiv.1803.01271Alhussein M., Aurangzeb K., Haider S. I. Hybrid CNN-LSTM model for short-term individual household load forecasting. IEEE Access. 2020;8:180544–180557. https://doi.org/10.1109/ACCESS.2020.3028281Alhussein M., Aurangzeb K., Haider S. I. Hybrid CNN-LSTM model for short-term individual household load forecasting. IEEE Access. 2020;8:180544–180557. https://doi.org/10.1109/ACCESS.2020.3028281Salinas D., Flunkert V., Gasthaus J., Januschowski T. DeepAR: Probabilistic forecasting with autoregressive recurrent networks. International Journal of Forecasting. 2020;36(3):1181–1191. https://doi.org/10.1016/j.ijforecast.2019.07.001Salinas D., Flunkert V., Gasthaus J., Januschowski T. DeepAR: Probabilistic forecasting with autoregressive recurrent networks. International Journal of Forecasting. 2020;36(3):1181–1191. https://doi.org/10.1016/j.ijforecast.2019.07.001

The authors declare that there are no conflicts of interest present.