A BiLSTM-Based Hybrid Model for Forecasting of Wind Power Output in Complementary Systems

Accurate wind power prediction is critical for stabilizing integrated water-wind-solar-storage (WWSS) systems. This research develops a forecasting model leveraging Bidirectional Long Short-Term Memory (BiLSTM) to capture bidirectional temporal dependencies, outperforming traditional Backpropagation Neural Networks (BPNN) in modeling complex wind power dynamics. Validation with Southwest China operational data shows BiLSTM achieves MAPE <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$=0.142$</tex> and SMAPE <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$=0.132$</tex>, reducing prediction error by <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$&gt;50 \%$</tex> during output peaks compared to BPNN (MAPE=0.248). The framework enhances coordination reliability for multi-energy complementary systems.