Ultra-Short-Term Wind Power Prediction Based on LVDBN

Abstract

Wind power forecasting has always been a major issue in wind power research. In order to improve the rapidity and accuracy of wind power forecasting, this study combines the least absolute shrinkage and selection operator (Lasso) and the variational mode decomposition (VMD) algorithm to select and extract the features of historical wind power time series, and carries out wind power forecasting through deep learning. The results show that the Deep Belief Network (DBN) prediction method which combines Lasso and VMD decomposition algorithm is obviously improved in calculation speed and accuracy compared with the prediction method which simply uses the original wind power time series or other models with the same data.