Abstract

Background: Wind power generation is becoming increasingly important as a renewable energy source, but its intermittent nature poses significant challenges to power system operation. Accurate forecasting of wind power output is crucial to ensure a stable and efficient power grid.

Objective: The scientific aim of the work is to propose a novel wind power forecasting method by combining the squeeze-and-excitation network (SENet), convolutional neural network (CNN), and long short-term memory (LSTM) network into a hybrid model. The subject of the research was to obtain a more accurate and reliable wind power forecasting approach and to explore the effectiveness of the proposed SENet-CNN-LSTM model in improving the forecasting performance of wind power generation.

Methods: Firstly, the isolated forest algorithm is used to detect abnormal values in wind power historical data, and the linear interpolation method is used to fill in the missing data. Secondly, CNN is used to extract the spatio-temporal characteristics of wind power data, SENet is used to assign different weights to the extracted feature information, and LSTM’s unique gating mechanism is used to memorize and forget the information. Finally, taking the measured historical data of a wind power farm as the sample, six algorithms including CNN, LSTM, CNN-LSTM, SENet- CNN, SENet-LSTM, and SENet-CNN-LSTM are used to predict the example.

Results: The prediction results show that the SENet-CNN-LSTM model achieves an RMSE of 1.781MW, MAE of 1.894MW, and MAPE of 9.032%. Compared to the other five prediction models, SENet-CNN-LSTM exhibits significantly improved performance with lower RMSE, MAE, and MAPE values.

Conclusion: The prediction accuracy of wind power based on the SENet-CNN-LSTM model is significantly improved, which provides an important analysis basis for the safe operation of wind farms and the economic dispatching of power grids.

Keywords: Wind power, channel attention mechanism, convolutional neural networks, long short-term memory, isolated forest.