Wind Power Prediction Under Extreme Weather Conditions of Low Temperature Based on TimeGAN and GWO-BiLSTM

• Published in: 2023 IEEE/IAS Industrial and Commercial Power System Asia (I&CPS Asia) pp. 2319 - 2325
• Main Authors: Song, Jiajiong, Peng, Xiaosheng, Yang, Zimin, Qin, Guoyuan, Chen, Ruofan, Wei, Peijie, Xiong, Yuhan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 07.07.2023
• Subjects: cold wave weather; GWO-BiLSTM; Heuristic algorithms; Predictive models; sample data expansion; Temperature distribution; Time series analysis; TimeGAN; Training; Uncertainty; Wind power generation; wind power prediction
• DOI: 10.1109/ICPSAsia58343.2023.10294871

The existing wind power prediction models rarely consider the impact of cold wave weather events, and are usually not suitable for such extreme weather conditions, which makes it difficult to meet the operation requirements of power system dispatch department. In order to improve the accuracy of wind power prediction under low temperature extreme weather conditions, this paper proposes a wind power prediction method based on TimeGAN sample data expansion and GWO-BiLSTM. First, the target data set is obtained by screening original data set according to the definition of cold wave weather event. Then, TimeGAN algorithm is used to expand the sample set of the meteorological and wind power data during the cold wave period. Finally, the wind power prediction model is established based on GWO-BiLSTM. The results show that the RMSE and MAE of wind power prediction under low temperature extreme weather conditions have decreased by 2.89% and 2.25% respectively, which effectively improves the prediction accuracy.