Researchers have unveiled a novel, three-stage adaptive cyberattack framework, Cyberattack Design Based on CNN-DQN-Blockchain Technology for Targeted Adaptive Strategy (CDB-TAS), specifically engineered to exploit vulnerabilities within hybrid hydrogen-electric power networks. This groundbreaking model addresses the limitations of existing static cyberattack methodologies by incorporating dynamic adaptability, multi-stage coordination, and sophisticated obfuscation mechanisms, presenting a significant evolution in the threat landscape for critical energy infrastructure.

The CDB-TAS framework operates through distinct phases designed for maximum impact and stealth. The initial Preliminary Reconnaissance Phase leverages a Convolutional Neural Network (CNN) to identify the most vulnerable buses within the grid through real-time anomaly detection. This intelligence then feeds into the Escalation Phase, where a Double Deep Q-Network (Double DQN) dynamically refines the attack strategy, adapting to real-time grid responses and demand profiles to optimize disruption. Finally, the Sustained Attack Phase maintains high-intensity disruptions while employing continuous feedback adaptation to minimize detection, ensuring prolonged system instability. Uniquely, the framework integrates a private blockchain network, not for defensive purposes, but as an attacker-side obfuscation layer, concealing attack metadata and facilitating decentralized coordination among malicious nodes.

Simulations conducted on a synthetic 2000-bus hybrid hydrogen-power system, meticulously modeled after the Electric Reliability Council of Texas (ERCOT), underscore the potent capabilities of CDB-TAS. The framework demonstrated the ability to induce up to a 15% voltage drop at critical buses, such as Bus 3103, and disrupt over 600 megawatts (MW) of load across 50 substations. Crucially, CDB-TAS achieved a 23.4% higher disruption efficiency compared to baseline attack models, all while maintaining significantly lower anomaly detection rates, highlighting its advanced stealth capabilities.

The increasing integration of Information and Communication Technologies (ICT) and Internet of Things (IoT) devices has transformed traditional energy grids into complex, automated, and data-centric networks. While this digitalization enhances control and efficiency, it simultaneously introduces new cybersecurity vulnerabilities, particularly in emerging multi-energy systems that combine conventional power with hydrogen networks. Existing static or single-phase cyberattack models often fail to adequately exploit or defend against these evolving complexities due to their lack of dynamic adaptability and coordination. This research marks a pivotal advancement by presenting the first integrated framework combining CNN, reinforcement learning, and blockchain from an adversarial perspective, offering unprecedented insights into the evolving threat landscape.

This comprehensive study provides critical intelligence for grid operators, cybersecurity professionals, and policymakers, guiding the development of future cyber-resilience strategies. Understanding these advanced adversarial tactics is paramount to fortifying the security posture of multi-energy systems against increasingly sophisticated and adaptive cyber threats.