A New Hybrid Forecasting Architecture of Wind Power Based on A Newly Developed Temporal Convolutional Networks

Wind energy planning, dispatching and control in the generation and conversion of wind power generation industry are becoming more and more important for today’s electrical power system, which needs accurate and stable wind speed prediction A new nonlinear hybrid model, SSA-TCN is proposed to improve the accuracy of wind power prediction. Singular Spectral Analysis (SSA) is introduced into the proposed method to decompose the original wind power sequence into four parts: trend, primary detail components, secondary detail components and noise. Moreover, a new Temporal Convolutional Networks (TCN) network is adopted in this paper instead of the current most advanced LSTM network. The output of TCN model has a longer actual memory range and is better able to cope with the gradient disappearance. In order to verify the advancement of SSA-TCN, the experimental part shows that the performance comparison between SSA-TCN, TCN and LSTM. The result shows that SSA-TCN has the best performance of all.