On the Resiliency of Grid-Connected Hydrogen Refueling Stations Against Distributed Denial-of-Service Cyberattacks

The deployment of Hydrogen Refueling Stations (HRSs) supports the decarbonization of integrated power and transportation systems by supplying green hydrogen-powered vehicles, utilizing excess renewable energy, and providing grid services. However, this integration requires data exchange with the Energy Management System (EMS), exposing it to potential cyberattacks that threaten safety, service continuity, and grid stability. Despite these risks, cybersecurity assessment and the development of cyber-resilient solutions for HRSs remain largely unexplored. Therefore, this paper proposes a novel cyber-resilient EMS for HRSs within an integrated transportation and power system. First, the impact of DoS attacks on individual operational parameters, such as hydrogen demand, tank State of Charge (SoC), and grid service signals, is analyzed to assess vulnerabilities and quantify economic losses. Then, a Distributed Denial-of-Service (DDoS) attack model is introduced, strategically optimizing the attack budget to target critical information signals while maximizing financial disruption. To mitigate these threats, a two-tiered defense mechanism is developed. To reconstruct missing data, a Time-Frequency Spectrum-based Deep Neural Network (TFS-DNN) is proposed. The model applies Short-Time Fourier Transform (STFT) to capture short-term temporal correlations and integrates Long Short-Term Memory (LSTM) networks to enhance long-term forecasting accuracy. Second, a two-stage robust optimization framework ensures secure and optimal HRS scheduling. The first stage determines safe operational limits to maintain power system stability, while the second stage employs convex robust optimization to derive an optimal scheduling strategy that accounts for forecast uncertainty and ensures transportation demand fulfillment, grid service provision, and financial sustainability.