A hybrid prediction of wind speed based on variational mode decomposition method and long short-term memory

Abstract

As wind speed directly influences wind energy generation, the improvement of the accuracy of wind speed forecasting algorithms has significant technological and economic impacts on wind energy generation. In this paper, a hybrid prediction model of wind speed based on variational mode decomposition (VMD), Nash-Sutcliffe coefficient of efficiency (NSCE) and long short-term memory (LSTM), referred to as VMD-NSCE-LSTM, is proposed to predict the wind speed. Firstly, the VMD is used to decomposes the wind speed time series into amplitude modulation and frequency modulation. Secondly, the long short-term memory (LSTM) is used to forecast each intrinsic mode function (IMF) of the VMD. Finally, the forecasting results of IMF are combined according to the evaluation criteria of the Nash-Sutcliffe coefficient of efficiency (NSCE). The simulation results show that the hybrid VMDNSCE-LSTM models can significantly improve the accuracy of wind speed prediction.