Forecasting the Power Output Produced by Photovoltaic Power Plants (for the day ahead) Using Machine Learning Techniques
DOI:
https://doi.org/10.24160/1993-6982-2018-4-53-57Keywords:
production forecast, machine learning, solar powerAbstract
Reducing the cost for manufacturing solar panels and increasing the efficiency of the panels themselves and the equipment associated with electricity generation are the main ways of developing solar energy around the world. The development and implementation of systems for forecasting the output generated by renewable energy sources (RES), in particular, solar power plants, is presently one of underestimated lines in the development of renewable energy. The aim of the study is to improve the accuracy of forecasts made for the day, weeks, months, and years ahead, information that will help achieve more economically efficient operation of RES-based facilities, as well as more reliable operation of the power system, which is especially important in view of a growing share of RES in the generation of electricity in Russia. One possible way to solve this problem is to develop models based on the use of machine learning techniques. For developing and using these models, it is necessary to arrange measurements of the quantitative parameters characterizing the state of the atmosphere (ambient air temperature, wind velocity and direction, humidity, and cloudiness) near the photovoltaic power station under consideration, as well as information on the amounts of generation at the given atmospheric parameters. The advantage of the suggested approach lies in its versatility and simplicity of its development. However, since an array of data sampled for a long period of time is necessary for training the model, this approach cannot to be used for new photovoltaic power plants or those that are under construction. In this study, we use the linear regression model, which is the simplest and most reliable model of machine learning that has positively proven itself in problems with a relatively small learning sample, to predict the hourly output produced by the photovoltaic power plant with an installed capacity of 10 MW for the day ahead. The use of the proposed model makes it possible to reduce the average absolute error of the forecast by 19% in comparison with that obtained with the aid of the currently applied model.
References
2. Постановление Правительства Российской Федерации № 1172 от 27.12.2010 «Об утверждении Правил оптового рынка электрической энергии и мощности и о внесении изменений в некоторые акты Правительства Российской Федерации по вопросам организации функционирования оптового рынка электрической энергии и мощности».
3. Максимов Б.К., Молодюк В.В. Теоретические и практические основы рынка электроэнергии. М.: Изд-во МЭИ, 2008.
4. Договор о присоединении к торговой системе оптового рынка. Прил. 12. Регламент определения объемов, инициатив и стоимости отклонений [Электрон. ресурс] http://www.np-sr.ru/regulation/joining/ reglaments/?ssFolderId=58 (дата обращения 05.04.2017).
5. Touati F.A., Al-Hitmi M.A., Bouchech H.J. Study of the Effects of Dust, Relative Humidity, and Temperature on Solar PV Performance in Doha: Comparison between Monocrystalline and Amorphous PVS // Intern. J. Green Energy. 2013. V. 10. No. 7. Pp. 680—689.
6. Научно-прикладной справочник по климату СССР. СПб.: Гидрометеоиздат, 1993. Сер. 3. Вып. 20. С. 96—104.
7. Jäger K. e. a. Solar Energy-fundamentals, Technology, and Systems. Delft: Delft University of Techn., 2014. Pp. 241—242.
8. Reda I., Andreas A. Solar Position Algorithm for Solar Radiation Applications // Solar energy. 2004. V. 76. No. 5. Pp. 577—589.
9. Виссарионов В.И. и др. Солнечная энергетика. М.: Издательский дом МЭИ, 2008. С. 126—128.
10. Глебов Н.И., Кочетов Ю.А., Плясунов А.В. Методы оптимизации. Новосибирск: Изд-во НГУ,2000. С. 66—68.
---
Для цитиров:ания Воротынцев Д.В., Тягунов М.Г. Прогноз выработки электроэнергии фотоэлектрическими электростанциями (на сутки вперед) с использованием машинного обучения // Вестник МЭИ. 2018. № 4. С. 53—57. DOI: 10.24160/1993-6982-2018-4-53-57.
#
1. Renewable Energy Prospects for the Russian Federation (REmap working paper) [Elektron. Resurs] http://irena.org/remap/IRENA_REmap_Russia_findings_2017_RU.pdf (Data Obrashcheniya 05.04.2017).
2. Postanovlenie Pravitel'stva Rossiyskoy Federatsii № 1172 ot 27.12.2010 «Ob Utverzhdenii Pravil Optovogo Rynka Elektricheskoy Energii i Moshchnosti i o Vnesenii Izmeneniy v Nekotorye Akty Pravitel'stva Rossiyskoy Federatsii po Voprosam Organizatsii Funktsionirovaniya Optovogo Rynka Elektricheskoy Energii i Moshchnosti». (in Russian).
3. Maksimov B.K., Molodyuk V.V. Teoreticheskie I Prakticheskie Osnovy Rynka Elektroenergii. M.: Izd-vo MPEI, 2008. (in Russian).
4. Dogovor o Prisoedinenii k Torgovoy Sisteme Optovogo Rynka. Pril. 12. Reglament Opredeleniya Ob′emov, Initsia tiv i Stoimosti Otkloneniy [Elektron. Resurs] http://www.np-sr.ru/regulation/joining/reglaments/?ssFolderId=58 (Data Obrashcheniya 05.04.2017). (in Russian).
5. Touati F.A., Al-Hitmi M.A., Bouchech H.J. Study of the Effects of Dust, Relative Humidity, and Temperature on Solar PV Performance in Doha: Comparison between Monocrystalline and Amorphous PVS. Intern. J. Green Energy. 2013;10;7:680—689.
6. Nauchno-prikladnoy Spravochnik po Klimatu SSSR. SPb.: Gidrometeoizdat, 1993;3;20:96—104. (in Russian).
7. Jäger K. e. a. Solar Energy-fundamentals, Technology, and Systems. Delft: Delft University of Techn., 2014:241—242.
8. Reda I., Andreas A. Solar Position Algorithm for Solar Radiation Applications. Solar energy. 2004;76;5: 577—589.
9. Vissarionov V.I. i dr. Solnechnaya Energetika. M.: Izdatel'skiy Dom MPEI, 2008:126—128. (in Russian).
10. Glebov N.I., Kochetov Yu.A., Plyasunov A.V. Metody Optimizatsii. Novosibirsk: Izd-vo NGU, 2000: 66—68. (in Russian).
---
For citation: Vorotyntsev D.V., Tyagunov M.G. Forecasting the Power Output Produced by Photovoltaic Power Plants (for the day ahead) Using Machine Learning Techniques. MPEI Vestnik. 2018;4:53—57. (in Russian). DOI: 10.24160/1993-6982-2018-4-53-57.
