Siberian Federal University 
- Issue
- Journal of Siberian Federal University. Engineering & Technologies. 2026 19 (4)
- Authors
- Ignatev, Evgenii V.; Deriugina, Galina V.; Golubkov, Ustin F.
- Contact information
- Ignatev, Evgenii V.: Moscow Power Engineering Institute (Moscow, Russian Federation); ; Deriugina, Galina V. : Moscow Power Engineering Institute (Moscow, Russian Federation); Golubkov, Ustin F. : Moscow Power Engineering Institute (Moscow, Russian Federation)
- Keywords
- wind energy; forecasting; optimization; hybrid methods; forecasting error
- Abstract
The article analyzes the main methodological approaches to forecasting electricity generation by wind farms. Forecasting models are classified based on their time horizon (short-term forecasts and long-term average generation forecasts) and the methodology used for their development (physical, statistical, and hybrid methods). Physical models are based on numerical weather prediction data and characteristics of wind farms, including equipment specifications, site descriptions, and wind turbine placement schemes. These models do not require training on historical data and provide high accuracy in medium- and long-term forecasting. Statistical models use regression methods and machine learning techniques to identify relationships between multiple influencing variables, making them particularly effective for short-term forecasting. However, they require large volumes of high-quality data. Hybrid models combine the strengths of both physical and statistical approaches, incorporating modern techniques such as ensemble learning and neural network algorithms, which enhance forecast accuracy and adaptability. The article provides a review of recent studies on wind power generation forecasting and the key metrics used to assess forecast accuracy. It addresses major challenges related to data processing, the need for standardized model evaluation approaches, and promising future directions for the development of forecasting methods, considering the advancements in computational power and the increasing availability of large datasets
- Pages
- 437–449
- EDN
- FGFHBE
- Paper at repository of SibFU
- https://elib.sfu-kras.ru/handle/2311/158564
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).