Day-Ahead Wind Power Forecasting Considering Value-Oriented Evaluation Metrics

Wind power forecasting is of great significance for the secure operation of the power system with high wind power penetration. This paper proposes a novel day-ahead wind power forecasting method based on a value-oriented comprehensive evaluation metrics. Different from previous forecasting methods which are constructed by only considering overall error mean metrics, the proposed method aims to directly maximize the application value of forecasting results in specific scenarios. Based on the application scenarios of ramp section and peak power section forecasting, comprehensive evaluation metrics is established. Then a LSTM-based wind power forecasting method driven by comprehensive evaluation metrics is constructed. Finally, the numerical analysis based on the actual data of a wind farm in southern China is carried out to verify that the proposed method have a better comprehensive performance.