Ultra-Short-Term Wind Power Forecasting Using an NSGA-II-MVMD-BiLSTM Hybrid Model

Wind power forecasting is crucial for ensuring the stability and reliability of power systems as wind energy penetration increases. This study presents a hybrid model, NSGA-II-MVMD-BiLSTM, designed for ultra-short-term wind power prediction. The model employs NSGA-II to optimize the key MVMD parameters and uses BiLSTM to predict decomposed components. To verify the generalization ability of the proposed method and to correct the duplicated Turbine 10 results in the original manuscript, all Turbine 10 experiments were rerun using a chronological 90%/10% split. The result report the metrics from the retained evaluation run for each model; they are not mean values over repeated runs. A Transformer baseline was also added as a recent attention-based model. The proposed NSGA-II-MVMD-BiLSTM model achieved the best performance.