Abstract
In the conditions of sharply variable highland climatic conditions, territorial distribution, generating capacity shortage, high cost of carbon fuel and the absence of large energy storage facilities to cover peak loads, it is necessary to ensure balance reliability in mountain isolated power systems (IES), taking into account the optimal resources allocation. To ensure the required balance reliability, it is necessary to implement reliable forecasting of power consumption in the medium term for planning the generating equipment load, taking into account the necessary and sufficient load coverage, generation costs, environmental friendliness and other criteria. Therefore, increased requirements are placed on the accuracy and robustness of the load forecast. A study of the meteorological factors influence on the medium-term forecasting of electricity consumption at the power plant in the Gorno-Badakhshan Autonomous Oblast (GBAO), located in the Republic of Tajikistan, which is characterized by the above-mentioned specific properties. A neural network model was used to predict power consumption taking into account meteorological factors. In order to increase the effectiveness of model training, an approach based on clustering of meteorological conditions is proposed. Its own neural network model is created for each cluster, in addition, an auxiliary model has been trained, which relates the current conditions to one of the clusters. Thus, instead of a single model that would take into account all possible conditions, a system of much simpler models was created, which increases the interpretability of the forecasting procedure and reduces the risk of retraining.
Keywords
Medium-term forecasting of power consumption, adaptive machine learning models, meteorological conditions, isolated power system, clustering.
1. Medium term load forecasting for jordan electric power system using particle swarm optimization algorithm based on least square regression methods / M. Hattab, M. Ma’itah, T. Sweidan, M. Rifai // J. of Power and Energy Engineering. 2017. Vol. 5. Pp. 75-96. doi: 10.4236/jpee.2017.52005
2. Electricity load forecasting: a systematic review / I.K. Nti, M. Teimeh, O. Nyarko-Boateng, A. F. Adekoya // J. of Electrical Systems and Inf. Technol. 2020. Vol. 7(13). doi: 10.1186/s43067-020-00021-8
3. Abu-Shikhah N., Elkarmi F., Aloquili O. Medium-term electric load forecasting using multivariable linear and non-linear regression // Smart Grid and Ren. Energy. 2011. Vol. 2(2). Pp. 126-135. doi: 10.4236/sgre.2011.22015
4. Hahn H., Meyer-Nieberg S., Pickl S. Electric load forecasting methods: Tools for decision making // European J. of Operational Research. 2009. Vol. 199(3). Pp. 902-907. doi: 10.1016/j.ejor.2009.01.062
5. Borges C.E., Penya Y.K., Fenandez I. Evaluating combined load forecasting in large power system and smart grids // IEEE Transactions on Industrial Informatics. 2013. Vol. 9(3). Pp. 1570-1577. doi: 10.1109/TII.2012.2219063
6. Short-term power load forecasting, profile identification, and customer segmentation: a methodology based on periodic time series / M. Espinoza, C. Joye, R. Belmans, B. De Moor // IEEE Transactions on Power Systems. 2005. Vol. 20(3). Pp. 1622-1630. doi: 10.1109/TPWRS.2005.852123
7. Blind kalman filtering for short-term power load forecasting / S. Sharma, A. Majumdar, V. Elvira, É. Chouzenoux // IEEE Transactions on Power Systems. 2020. Vol. 35(6). Pp. 4916-4919. doi: 10.1109/TPWRS.2020.3018623
8. Arora S., Taylor J.W. Short-term forecasting of anomalous load using rule-based triple seasonal methods // IEEE Transactions on Power Systems. 2013. Vol. 28(3). Pp. 3235-3242. doi: 10.1109/TPWRS.2013.2252929
9. Multi-scale convolutional neural network with time-cognition for multi-step short-term power load forecasting / Z. Deng, B. Wang, Y. Xu, T. Xu // IEEE Access. 2019. Vol. 7. Pp. 88058-88071. doi: 10.1109/ACCESS.2019.2926137
10. Chen B.-J., Chang M.-W., Lin Ch.-J. Load forecasting using support vector machines: a study on EUNITE competition 2001 // IEEE Transactions on Power Systems. 2004. Vol. 19(4). Pp. 1821-1830. doi: 10.1109/TPWRS.2004.835679
11. A short-term power load forecasting method using integrated CNN and LSTM network / S.H. Rafi, Nahid-Al-Masood, S.R. Deeba, E. Hossain // IEEE Access. 2021. Vol. 51. Pp. 32436-32448. doi: 10.1109/ACCESS.2021.3060654
12. The impact of data filtration on the accuracy of multiple time-domain forecasting for photovoltaic power plants generation / S.A. Eroshenko, A.I. Khalyasmaa, D.A. Snegirev, V.V. Dubailova // Applied Sciences. 2020. Vol. 10(22). Art. 8265. doi: 10.3390/app10228265
13. Panapakidis I.P., Christoforidis G.C. Implementation of modified versions of the K-means algorithm in power load curves profiling // Sustainable Cities and Society. 2017. Vol. 35. Pp. 83-93. doi: 10.1016/j.scs.2017.08.002
14. Viola L.G. Clustering electricity usage profiles with K-means. URL: https://towardsdatascience.com/clustering-electricity-profiles-with-k-means-42d6d0644d00 (дата обращения 15.09.2022).
15. Гуломзода А.Х., Сафаралиев М.Х., Люханов Е.А. Модифицированный способ синхронизации Microgrid с внешней изолированной энергосистемой // Электротехнические системы и комплексы. 2021. № 3(52). С. 72-80. doi: 10.18503/2311-8318-2021-3(52)-72-80
16. Разработка моделей среднесрочного прогнозирования электропотребления в изолированно работающих энергосистемах на основе ансамблевых методов машинного обучения / С.М. Асанова, Дж.С. Ахьеев, С.А. Дмитриев, П.В. Матренин // Известия НТЦ единой энергетической системы. 2021. №1 (84). С. 32-39.
17. Адаптивные ансамблевые модели для среднесрочного прогнозирования выработки электроэнергии гидроэлектростанциями в изолированных энергосистемах с учётом изменений температуры / М.Х. Сафаралиев, П.В. Матренин, С.А. Дмитриев, Дж.С. Ахьеев // Электротехнические системы и комплексы. 2022. 1(54). С. 38-45. doi: 10.18503/2311-8318-2022-1(54)-38-45
18. Разработка моделей среднесрочного прогнозирования притока воды для планирования выработки электроэнергии в изолированных энергосистемах / М.Х. Сафаралиев, С.Е. Кокин, С.А. Дмитриев, П.В. Матренин // Электричество. 2022. № 5. С. 58-68. doi: 10.24160/0013-5380-2022-5-58-68
19. Khalyasmaa A.I., Matrenin P.V., Eroshenko S.A. Inappropriate machine learning application in real power industry cases // International Journal of Electrical and Computer Engineering. 2022. Vol. 12(3). Pp. 3023-3032. doi: 10.11591/ijece.v12i3.pp 3023-3032
20. Improving accuracy and generalization performance of small-size recurrent neural networks applied to short-term load forecasting / P.V. Matrenin, V.Z. Manusov, A.I. Khalyasmaa, D.V. Antonenkov // Mathematics. 2020. Vol. 8(12). Art. 2169. doi: 10.3390/math8122169
21. Hartigan J.A., Wong M.A. Algorithm AS 136: A K-Means clustering algorithm // Journal of the Royal Statistical Society. Series C (Applied Statistics). 1979. Vol. 28(1). Pp. 100-108. doi: 10.2307/2346830
22. Principal manifolds for data visualization and dimension reduction / Gorban A.N., Kegl B., Wunsch D.C., Zinovyev A. Berlin: Springer, 2008. 361 p.
23. Kingma D.P., Ba J.L. Adam: A method for stochastic optimization. URL: https://arxiv.org/pdf/1412.6980.pdf (дата обращения 12.09.2022).
Safaraliev M.Kh., Matrenin P.V., Kyrianova N.G., Ghulomzoda A.H. Medium-Term Forecasting of Power Consumption Based on an Artificial Neural Network in Isolated Power Systems. Elektrotekhnicheskie sistemy i kompleksy [Electrotechnical Systems and Complexes], 2022, no. 4(57), pp. 4-11. (In Russian). https://doi.org/10.18503/2311-8318-2022-4(57)-4-11