Short-term Wind Power Prediction Method Based on TCN-GRU Combined Model
Haotian Guo, Lin Pan, Jian Wang, Xin-Bo Fan, Jin Li, Zhe Liu
Abstract
The target of carbon peak and carbon neutralization puts forward higher requirements for the accuracy of wind power prediction. In order to accurately predict the wind power timing, a short-term wind power prediction method based on the hybrid neural network model of temporal convolution network (TCN) and gate recurrent unit (GRU) was proposed. TCN was used to extract the sequential features of wind power time series and unidirectional space features, and GRU was used to further extract the sequential features of wind power series. The two models were trained together. The wind power timing series of a single fan and a single wind farm were collected respectively. After the pre-processing, the corresponding data sets were constructed for modeling respectively to verify the reliability of the proposed method. The results show that compared with the traditional model, this method has lower sensitivity to the time window, higher accuracy, and has certain engineering application value.