Research on Short-Term Wind Speed Prediction Based on Deep Learning

Wind speed prediction is of great significance in power system operation. In this paper, a hybrid neural network algorithm model based on Grey Wolf Optimization (GWO) is proposed to improve the accuracy of wind speed prediction. Firstly, the original wind speed data is decomposed into intrinsic mode components by Singular Spectrum Analysis (SSA). Then, the Long Short Term Memory (LSTM) model is established to predict the sub-sequence respectively. Lastly, the weight of each sub-sequence is optimized by GWO algorithm, and the final prediction result is obtained by performing weighted fusion to superimpose the sub-sequences. Accumulating experiments indicate that the SSA-LSTM-GWO model has better predictability, compared with the existing models such as EEMD-LSTM.