Short-term prediction of wind turbine power based on improved CNN-LSTM hybrid model

Precise short-term prediction of wind turbine power is crucial for optimizing power dispatching, enhancing power grid reliability, and improving wind power consumption capabilities. This paper presents a short-term prediction approach for wind turbine power based on an improved CNN- LSTM hybrid model. Initially, the wind turbine data is preprocessed, and then the structure of the traditional model is innovatively improved. Additionally, the impact of meteorological factors such as wind direction and temperature on wind turbine power prediction is taken into account. The results indicate that the improved CNN-LSTM hybrid model outperforms the traditional CNN-LSTM and CNN-GRU models in key evaluation metrics like RMSE, MAE, and $\mathbf{R}^{\mathbf{2}}$. This improvement significantly boosts the short-term prediction accuracy of wind turbine power, offering a more reliable and efficient solution for such predictions.