Based on two levels of decomposition and the length of the memory network multi-step combined forecasting model of short-term wind speed

Summary: In order to better extract and learn the characteristics of wind speed in time and frequency domains, solve the time domain randomness and frequency domain of wind speed signalTo solve the complexity problem, a new method based on wavelet decomposition(WD), variational modulostate decomposition(VMD), long and short-term memory(LSTM)Network and Attention Mechanisms(AT)short-term wind speed combined forecasting model(WD-VMD-DLSTM-AT).span>On this basis, a multiple-input multiple-output based attention mechanism is proposed(MIMO)encoding-decoding multi-step prediction model(MMED-AT).Through experimental comparison and analysis, the proposed combined prediction model has the best statistical error and can significantly improve the prediction accuracy in short-term wind speed prediction.Based on