Secure forecasting-driven scheduling of hydrogen-integrated renewable power systems under cyberattacks

Accurate demand and renewable-generation forecasts are essential for reliable operation of hydrogen-integrated power systems, where electrolyzers, liquid organic hydrogen carrier logistics, storage, and PEM fuel cells provide flexibility for renewable utilization. This paper develops a cyber-resilient forecasting-driven scheduling framework for coordinated power–hydrogen systems. Forecasting quality is first assessed using actual measurements, high-quality forecasts, and medium-quality forecasts; medium-quality forecasts increase total-cost deviations by up to 22.08%. A two-stage detection–recovery framework then protects demand, photovoltaic, and wind forecasting signals against denial-of-service and ramp-type false data injection attacks. Dedicated Long Short-Term Memory (LSTM) models detect corrupted inputs and reconstruct compromised measurements before coordinated scheduling. Demand-targeted attacks are the most disruptive, while wind- and PV-targeted attacks increase operating costs by up to 29.55% and 12.76%, respectively. After detection–recovery, operating-cost deviations fall below 6%. The framework improves hydrogen-enabled flexibility, renewable utilization, and operational resilience under cyber-compromised forecasting conditions.