A.N. Alyunov1, K.D. Mosolova2

1,2 Financial University under the Government of the Russian Federation (Moscow, Russia)

Problem setting. Forecasting electricity consumption is a complex and important process for industrial enterprises. The accuracy of the forecast for the near future allows not only to avoid additional energy costs, but also to optimize production processes. To achieve this goal, it is necessary to use flexible solutions that provide accurate and interpretable results in a short time. Industrial enterprises (companies) consuming a significant amount of electric energy are required to submit applications every day with information on how much electricity is planned to be purchased for the next day. The deviation of the actual consumption of electricity in a smaller or larger direction leads to additional costs. Consequently, the development of algorithms for forecasting the volume of electric energy consumption is required, while the use of machine learning methods to solve the problem under consideration will improve the accuracy of forecasts based on existing solutions. Machine learning can identify and take into account patterns that are ignored both in manual calculations and when using statistical models.

Target. To develop an algorithm for predicting the consumption of electric energy volumes based on the selected mathematical model, which ensures an increase in the efficiency of an industrial enterprise.

Results. An algorithm for predicting electricity consumption by an industrial enterprise is proposed, which, based on information collected from commercial electricity metering systems, allowed achieving an accuracy of 91%. The "random forest" model can make it possible to optimize production processes, which is an advantage of the developed algorithm.

Practical significance. The research carried out during the development can serve as a starting point for further study of the problem in order to achieve even greater accuracy of forecasts, as well as the creation of software products for forecasting the volume of electricity consumption by industrial enterprises.

Alyunov A.N., Mosolova K.D. Electrical energy consumption prediction using machine learning methods. Neurocomputers. 2023. V. 25. № 5. Р. 58-70. DOI: https://doi.org/10.18127/j19998554-202305-08 (In Russian)

1. Ahire P.G., Patil P.D. Context-Aware Clustering and the Optimized Whale Optimization Algorithm: An Effective Predictive Model for the Smart Grid. International Journal on Recent and Innovation Trends in Computing and Communication. 2023. V. 11. № 1. P. 62–76. DOI 10.17762/ijritcc.v11i1.5987.
2. Matrenin P.V. [Antikeyes] Forecasting and planning of electricity consumption using machine learning (experiment). [Electronic resource] – Access mode: https://habr.com/ru/articles/577732/, date of reference: 20.05.2023. (in Russian)
3. Migranov M.M., Ustinov A.A., Melnikov A.V. Forecasting of electricity consumption. Application practice. Electricity. Transmission and distribution. 2018. № 2(47). P. 44–53. (in Russian)
4. Morgoeva A.D., Morgoev I.D., Klyuev R.V., Gavrina O.A. Forecasting of electric energy consumption by an industrial enterprise using machine learning methods. Proceedings of Tomsk Polytechnic University. 2022. № 7. P. 115–125. DOI 10.18799/24131830/ 2022/7/3527. (in Russian)
5. Kirilychev I.A. Advantages of using artificial neural networks in forecasting energy consumption and electricity prices. Young scientist. 2022. № 413. P. 72–73. (in Russian)
6. Guzhov S.V. Neural networks as a tool for forecasting energy consumption | Expert Bureau EnergiaVita. Energy efficiency, ecology, digitalization. [Electronic resource] – Access mode: https://energiavita.ru/2020/06/28/nejronnye-seti-kak-instrument-prognozirovaniya-ehnergopotrebleniya/, date of reference: 22.05.2023. (in Russian)
7. Ilyinykh M.V. Application of Methods of Artificial Intelligence in Predicting Energy Consumption (Sholom-Aleichem Priamursky State University is Taken as an Example). Bulletin of the Amur State University. Sholom Aleichem. 2020. № 4(41). P. 29–41. DOI 10.24412/2227-1384-2020-4-29-41. (in Russian)
8. Catal C., Ece K., Arslan B., Akbulut A. Benchmarking of Regression Algorithms and Time Series Analysis Techniques for Sales Forecasting. Balkan Journal of Electrical and Computer Engineering. 2019. V. 7. № 1. P. 20–26. DOI 10.17694/bajece.494920.
9. Morgoev I.D., Dzgoev A.E., Klyuev R.V., Morgoeva A.D. Forecasting the consumption of electricity by enterprises of the national economy complex in conditions of incomplete information. News of the KabardinoBalkarian Scientific Center of RAS. 2022. № 3(107).
   P. 9–20. DOI 10.35330/1991-6639-2022-3-107-9-20. (in Russian)
10. Garcia-Martin E., Rodrigues C., Riley G., Grahn H. Estimation of energy consumption in machine learning. Journal of Parallel and Distributed Computing. 2019. V. 134. P. 75–88. DOI 10.1016/j.jpdc.2019.07.007.
11. Banik R., Das P., Ray S., Biswas A. Prediction of electrical energy consumption based on machine learning technique. Electrical Engineering. 2021. V. 103. P. 909–920. DOI 10.1007/s00202-020-01126-z.
12. Tian Y., Yu J., Zhao A. Predictive model of energy consumption for office building by using improved GWO-BP. Energy Reports. 2020. V. 6. P. 620–627. DOI 10.1016/j.egyr.2020.03.003.
13. Chen H., Lee C. Electricity consumption prediction for buildings using multiple adaptive network-based fuzzy inference system models and gray relational analysis. Energy Reports. 2019. V. 5. P. 1509–1524. DOI 10.1016/j.egyr.2019.10.009.
14. Chervyakov N.I., Tikhonov E.E. Application of neural networks for forecasting tasks and problems of identification of forecasting models on neural networks. Neurocomputers. 2003. № 10–11. P. 25–31. (in Russian)
15. Khalyasmaa A.I., Matrenin P.V., Yaroshenko S.A. Error analysis of machine learning algorithms application in electric power engineering problems. Electricity. Transmission and distribution. 2021. № 66. P. 46–53. (in Russian)
16. Kostygov A.M., Kichkin A.V., Barkovets K.A. Forecasting of power consumption of a building based on selective selection of a neural network. Information-measuring and Control Systems. 2017. V. 15. № 9. P. 75–82. (in Russian)
17. Dudek G.A. Comprehensive Study of Random Forest for Short-Term Load Forecasting. Energies. 2022. V. 15. № 20. P. 7547. DOI 10.3390/en15207547.
18. Weather history: reference and information portal. [Electronic resource] – Access mode: https://weatherarchive.ru/, date of reference 10.05.2023. (in Russian)
19. Why you should try Mean Encoding. [Electronic resource] – Access mode: https://towardsdatascience.com/why-you-should-try-mean-encoding-17057262cd0, date of reference: 29.04.2023.