CTGANBiLSTM Interval Power Prediction Method for Wind Turbines Based on Feature Selection

Abstract Aiming at the current problem of insufficient prediction accuracy and efficiency, this paper proposes an interval power prediction method for wind turbines. Firstly, for the problem of slow computing speed due to feature redundancy, Pearson coefficients and Spearman coefficients are used for feature screening, based on which the model combines bi-directional long and short-term memory network (BiLSTM) and conditional generative adversarial network (CTGAN), and introduces an attention mechanism to adaptively learn the weights of different features to improve the prediction accuracy. The results show that compared with the baseline model, the MAE and RMSE of the method are reduced by 2.825 % and 12.87 % on average in different seasons, respectively, and finally, the reliability and accuracy of the proposed model are examined at different confidence levels.