Short-term wind speed multistep combined forecasting model based on two-stage decomposition and LSTM

摘要

In order to better extract and study the characteristics of the wind speed in time-domain and frequency-domain, so as to solve the time-domain randomness and frequency-domain complexity problems of the wind speed signal, a combined short-term prediction model (WD-VMD-DLSTM-AT), which is based on two-stage decomposition (WD + VMD), double long-short-term memory network (DLSTM) and attention mechanism (AT), is proposed; on this basis, a multi-input multiple output (MIMO) codec model based on attention mechanism (MMED-AT) is proposed for multiple short-term wind speed step forecast. Through experimental comparison and analysis, the proposed combined forecasting model has the smallest statistical error and the best prediction accuracy; the MMED-AT models based on the combined model can obviously eliminate the cumulative error of recursive multistep prediction and further improve the stability of multistep prediction.