Wind Power Forecasting Based on CNN-LSTM-Attention

To improve the accuracy of wind power forecasting, this paper proposes an integrated model combining a Convolutional Neural Network (CNN), a Long Short-Term Memory (LSTM) network, and an Attention mechanism. First, using a wind power dataset provided by the Guohua Xinjiang Wind Farm, data preprocessing was performed, including generating samples with a sliding window approach and applying normalization. Then, CNN was used to extract local features, LSTM was used to capture temporal dependencies, and the Attention mechanism was employed to dynamically assign feature weights. Finally, predictions were made on the test set, and the model’s performance was evaluated. Experimental results show that the CNN-LSTM-Attention model outperforms traditional CNN, LSTM, and CNN-LSTM models in terms of RMSE, MAE, MAPE, and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> indicators. This demonstrates the effectiveness of CNN-LSTM-Attention in complex time-series forecasting and provides an efficient and reliable solution for wind power forecasting, offering theoretical and technical support for wind farm operations optimization.