Research on capacity configuration of wind–hydrogen coupling systems based on the NSGA-II algorithm

Wind power uncertainty and load fluctuations impose stringent requirements on the capacity planning and stable operation of wind–hydrogen coupling systems. This study develops an integrated prediction–control–planning framework for multi-objective capacity optimization under ultra-short-term wind power variability. First, a wind power forecasting model is developed by integrating variational mode decomposition optimized by the sparrow search algorithm with long short-term memory networks. The model achieves a mean absolute error of 0.2278 MW and a normalized mean absolute error of 5.06%, normalized by the rated capacity of the wind farm. Second, a three-mode power balance control strategy is designed to coordinate wind turbines, electrolyzers, fuel cells, and hydrogen storage tanks. Third, a bi-objective capacity planning model is formulated to simultaneously minimize total investment cost and an integrated power deviation (PD) metric, which combines the loss of power supply probability and the power fluctuation ratio. A standard non-dominated sorting genetic algorithm II is then employed to obtain Pareto-optimal capacity configurations. An offline case study is conducted using time-stamped historical wind speed data sampled at 15 min intervals, with wind speed converted into aggregated active power through the wind turbine power curve. The results reveal a clear trade-off between economic performance and grid compatibility. A representative compromise solution—comprising 30 wind turbines, 4 electrolyzers, 6 hydrogen storage tanks, and 5 fuel cells—requires an investment of 2.65 × 107 Chinese yuan while maintaining the PD at 2.13%.