Ultra-short-term Wind Power Prediction Based on Bayesian Optimisation XGboost-LSTM

Artificial intelligence approaches are used in the realm of energy security to increase wind power prediction accuracy and lessen the effect of wind energy uncertainty on wind power generation. Model performance and input features are significantly correlated, which is expected given the temporal, periodic, and non-stationary nature of wind power impacting factors. An ultrashort-term wind power prediction model based on the Bayesian optimization of the XGboost-LSTM is developed in order to enhance the model’s generalization performance and prediction capacity as well as to offer dependable and stable prediction data for wind power generation. First, the model input features are extracted using the extreme gradient boosting tree (XGboost) approach. Second, the entire XGboost-LSTM model is optimized using the Bayesian hyperparameter tuning approach. Finally, by integrating the example data of a wind farm in Bili-uhe, Inner Mongolia, the method’s correctness and efficacy are confirmed by contrasting it with other combined models and a single model. By using the MAE, RMSE, MAPE, R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, and the fitting effect of the wind power curve as the indexes, the results demonstrate that the suggested model has a better fitting effect on the original data and a higher prediction accuracy.