Probabilistic Wind Power Forecasting by Using Quantile Regression Analysis

Ozkan, Mehmet Baris
Guvengir, Umut
Kucuk, Dilek
Secen, Ali Unver
Buhan, Serkan
Demirci, Turhan
Bestil, Abdullah
Er, Ceyda
Karagöz, Pınar
Effective use of renewable energy sources, and in particular wind energy, is of paramount importance. Compared to other renewable energy sources, wind is so fluctuating that it must be integrated to the electricity grid in a planned way. Wind power forecast methods have an important role in this integration. These methods can be broadly classified as point wind power forecasting or probabilistic wind power forecasting methods. The point forecasting methods are more deterministic and they are concerned with the exact forecast for a particular time interval. These forecasts are very important especially for the Wind Power Plant (WPP) owners who attend the energy market with these forecasts from day-ahead. Probabilistic wind power forecasting is more crucial for the operational planning of the electricity grid by grid operators. In this methodology, the uncertainty in the wind power forecast for WPPs are presented within some confidence. This paper presents a probabilistic wind power forecasting method based on local quantile regression with Gaussian distribution. The method is applied to obtain probabilistic wind power forecasts, within the course of the Wind Power Monitoring and Forecast Center for Turkey (RİTM) project, which has been realized by TÜBİTAK MAM. Currently, 132 WPPs are included in the project and they are being monitored in real-time. In this paper, the results for 15 of these WPPs, which are selected from different regions of the country, are presented. The corresponding results are calculated for two different confidence intervals, namely 5–95 and 25–75 quantiles.
Citation Formats
M. B. Ozkan et al., “Probabilistic Wind Power Forecasting by Using Quantile Regression Analysis,” presented at the DARE 2017., 2017, Accessed: 00, 2020. [Online]. Available: