Short-Term Wind Power Prediction Based on Meteorological Process Division

Abstract

Abstract In the traditional weather classification prediction method based on clustering, the feature quantity is difficult to fully reflect the change information of complex weather, which will weaken the reliability of classification. Therefore, this paper proposes a wind power prediction model based on meteorological process division. Firstly, this method uses the correlation coefficient to analyze the correlation between various meteorological factors and wind power output in numerical weather and then uses the wind speed after wavelet denoising to divide the weather fluctuation process and extract the fluctuation characteristics of each fluctuation process. Secondly, the meteorological process is divided by a hierarchical clustering algorithm, and the types of meteorological processes are judged by a dynamic time planning algorithm. Finally, the wind power prediction model of a long short-term memory (LSTM) network is established based on different meteorological processes. The results show that this method can provide a reference for improving the accuracy and interpretability of short-term wind power forecasting.