Ultra-short-term prediction of wind power based on EMD and DLSTM

In the background of the increase in wind power installed capacity in power systems, high-precision ultra-short-term wind power prediction is an important basis for ensuring reliable operation of the system. In order to make full use of the effective information in historical data to further improve the prediction accuracy of wind power generation, a prediction method combining empirical mode decomposition(EMD) and deep long-term memory(DLSTM) is proposed, constructing a multi-scale combined prediction model(EMD-DLSTM). In the process of model construction, firstly, the EMD method is used to decompose the wind power sequence into intrinsic mode function(IMF)of different scales and a trend term to weaken the volatility of the wind power sequence and obtain more regular components. Then, the DLSTM network is used to model the decomposed wind power sequence and its features, and then weight the predicted values of each component to obtain the true predicted wind power. The empirical analysis of a wind farm data in Laizhou shows that the EMD-DLSTM method proposed in the paper has better prediction ability than the existing forecasting methods.