A Short-Term Wind Power Forecast Method via XGBoost Hyper-Parameters Optimization
Xiong Xiong, Xiaojie Guo, Pingliang Zeng, Ruiling Zou, Xiaolong Wang
Abstract
The improvement of wind power prediction accuracy is beneficial to the effective utilization of wind energy. An improved XGBoost algorithm via Bayesian hyperparameter optimization (BH-XGBoost method) was proposed in this article, which is employed to forecast the short-term wind power for wind farms. Compared to the XGBoost, SVM, KELM, and LSTM, the results indicate that BH-XGBoost outperforms other methods in all the cases. The BH-XGBoost method could yield a more minor estimated error than the other methods, especially in the cases of wind ramp events caused by extreme weather conditions and low wind speed range. The comparison results led to the recommendation that the BH-XGBoost method is an effective method to forecast the short-term wind power for wind farms.