Research on CNN-LSTM based on attention mechanism for wind power generation prediction

Due to the intermittent and random nature of wind power generation, the instability of the power grid can reduce the safety of its operation. Consequently, peak regulation of the grid system has become one of the new challenges for maintaining a stable grid operation. This research suggests an attention mechanism-based CNN-LSTM architecture-based wind power prediction model to address this problem. For projecting the output of wind turbines, this model considers a number of meteorological variables, including the speed and direction of the wind, air pressure, and temperature. The CNN-LSTM architecture is used to extract critical feature information from the data, while the attention mechanism assigns different weights to highlight the most critical features for more accurate wind power prediction. On the basis of the data gathered from a wind farm in northwest China, studies for predicting winds were done, and the prediction results of different models were compared using the goodness-of-fit and RMSE as evaluation indices. According to the findings, the CNN-LSTM model that is based on the attention mechanism has a goodness-of-fit of 0.89 and an RMSE of 18.03, proving that the model can effectively predict wind power. Therefore, the proposed method has a certain application value.