ENERGY GENERATION, ELECTRIC ENGINEERING AND ELECTROMECHANICS

Improving the Accuracy of the Forecast of Electricity Production by Photovoltaic Power Station Based on the Random Forest Method

Authors

V. V. Kulyk
Vinnytsia National Technical University ROR
M. V. Zathey
Vinnytsia National Technical University ROR
DOI https://doi.org/10.31649/1997-9266-2024-177-6-52-61

Keywords

photovoltaic power station generation forecasting regression analysis machine learning data analysis

Abstract

The work examines the methods used to forecast energy production by Photovoltaic Power Station (PPS), as well as ways to improve the accuracy of the forecast to optimize the structure of the electricity balance in the power system. The research is aimed at identifying effective forecasting approaches and algorithms, assessing their accuracy and reliability. Based on the results of the study, a combination of regression analysis and machine learning methods is proposed, which provides acceptable forecast accuracy for planning power reserves in the power system. The Random Forest method was used as a basis, as it ensures adaptability to the specifics of energy generation by PPS in different regions of Ukraine and in different periods of the year. To increase the efficiency of machine learning, an algorithm for pre-filtering datasets using autoregression and moving average methods was proposed. This allows better preparation of input data for further use in forecasting, ensuring smoothing of time series and elimination of random fluctuations that can negatively affect forecast accuracy. The use of pre-filtering methods allows to identify the main regularities in the data, which, in turn, increases the accuracy of machine learning models. The Random Forest method was not chosen by chance: it is well suited for forecasting tasks where many different factors need to be taken into account, which may affect the results differently depending on time and region. This is especially important in the case of forecasting the energy generation of solar power plants, where production is significantly affected by variables such as cloud cover, temperature, seasonality, etc. Using Random Forest allows you to take into account non-linear dependencies and interactions between factors, which helps to increase the accuracy of forecasts. The use of Random Forest as the main machine learning algorithm is due to its flexibility and ability to adapt to different conditions, which allows you to effectively take into account the features of energy generation in different regions and during different seasons.

389 160

How to Cite

[1]
“Improving the Accuracy of the Forecast of Electricity Production by Photovoltaic Power Station Based on the Random Forest Method”, Вісник ВПІ, no. 6, pp. 52–61, Dec. 2024, doi: 10.31649/1997-9266-2024-177-6-52-61.

Author Biographies

V. V. Kulyk, Vinnytsia National Technical University

Dr. Sc. (Eng.), Associate Professor, Professor of the Chair of Power Plants and Systems

M. V. Zathey, Vinnytsia National Technical University

Post-Graduate Student of the Chair of Power Plants and Systems

References

A. Loureno, et al. “Comparison of forecasting models for photovoltaic power generation,” Energy Conversion and Management, no. 118, pp. 404-418, 2016.
Peter J. Brockwell and Richard A. Davis, Time Series: Theory and Methods, 2016. [Electronic resource]. Available: https://pdfarchived.net/list/time-series-theory-and-methods-peter-j-brockwell-4913414 .
Robert H. Shumway and David S. Stoffer, Time Series Analysis and Its Applications: With R Examples, 2017. [Electronic resource]. Available: http://pzs.dstu.dp.ua/DataMining/times/bibl/TimeSeries.pdf .
G. Zhang, “Time series forecasting using a hybrid ARIMA and neural network model,” Neurocomputing, no. 50, pp. 159-175. 2003. https://doi.org/10.1016/S0925-2312(01)00702-0 .
S. Borchani, et al. “Short-term solar power forecasting using machine learning techniques,” Renewable Energy, no. 116, Part A, pp. 729-743, 2018.
A. C. Cadena, et al. “Weather Forecasting for Photovoltaic Power Prediction Using Machine Learning Techniques,” Energies, no. 11(6), pp. 1362, 2018.
R. J. Hyndman,, and G. Athanasopoulos, Forecasting: principles and practice. OTexts. 2018. [Electronic resource]. Available: https://www.scirp.org/reference/referencespapers?referenceid=2849375 .
C. Zhang, Y. Guo, M. Li “A review of the development and application of artificial neural network models [J],” Computer Engineering and Applications, no. 57(11), pp. 57-69, 2021.
Yi Zhou, et al. “Prediction of photovoltaic power output based on similar day analysis, genetic algorithm and extreme learning machine,” Energy, no. 204, pp. 78-94, 2020.
D. Infield, and M. O’Malley, “A probabilistic forecast methodology for the management of renewable energy: The application of small-scale solar power,” IEEE Transactions on Power Systems, no. 22(3): 1147, pp. 1156, 2007.
Mittal, Amit Kumar, Kirti Mathur, and Shivangi Mittal, “A review on forecasting the photovoltaic power using machine learning,” Journal of Physics: Conference Series, vol. 2286, no. 1. IOP Publishing, 2022.
Ibrahim I. Anwar, M. J. Hossain, and Benjamin C. Duck, “An optimized offline random forests-based model for ultra-short-term prediction of PV characteristics,” IEEE Transactions on Industrial Informatics, no. 16.1, pp. 202-214, 2019.
Andreas C. Müller and Sarah Guido, Introduction to Machine Learning with Python: A Guide for Data Scientists, 2016.
Kumar Abhishek, et al. “A review and analysis of forecasting of photovoltaic power generation using machine learning,” International Conference on Management Science and Engineering Management. Cham: Springer International Publishing, 2022.
Lei Wen, and Xiaoyu Yuan. “Forecasting CO2 emissions in Chinas commercial department, through BP neural network based on random forest and PSO,” Science of the Total Environment, no. 718, pp. 137-194, 2020. ISSN 0048-9697. https://doi.org/10.1016/j.scitotenv.2020.137 .
A. Savitzky, M. J. E. Golay, “Smoothing and Differentiation of Data by Simplified Least Squares Procedures,” Analytical Chemistry, no.36(8), pp. 1627-1639, 1964.
Michel Talagrand, The Generic Chaining: Upper and Lower Bounds of Stochastic Processes, Springer-Verlag, 2005, 222 pp.

Most read articles by the same author(s)

1 2 > >>