Ultra-short-term wind-power forecasting based on an optimized CNN-BILSTM-attention model

The accurate forecast of wind power is crucial for the stable operation and economic dispatch of renewable energy power systems. To improve the accuracy of ultra-short-term wind-power forecast, we propose an improved model combining a convolutional neural network (CNN), bidirectional long short-term memory, and an attention mechanism network. First, the basic principle of the proposed model is introduced along with its merits in ultra-short-term wind-power forecast. Then, relevant data are processed based on the Pearson similarity criterion, and relevant feature parameters for wind-power forecast are optimized. Finally, the proposed model is analyzed based on the public dataset of the Baidu KDD Cup 2022 wind-power forecast competition and actual data from a wind farm in Shandong. Results show that the proposed model can effectively overcome the shortcomings of traditional forecast methods in terms of overfitting, feature extraction, and parameter tuning. Furthermore, the model exhibits higher forecast accuracy and stability.