Enhancing the Stealth of Load Redistribution Attacks: A Novel Cluster-Driven Approach

This letter proposes a novel cluster-driven bilevel Load Redistribution (LR) attack model in cyber physical power systems. The main goal is on overloading multiple transmission lines while maintaining a high level of stealth. The proposed model employs cluster-based approach to form distinct clusters, each representing different load patterns within the power system. This enables the design of attack vectors that seamlessly blend with these load patterns, thereby challenging advanced anomaly detection systems. The stealth of the attack is quantified by the Stealth Distance Index (SDI), ensuring that injected false data remain within the maximum permissible distance from the centroid of its corresponding cluster. The proposed model is validated on the modified IEEE 14-bus and 30-bus systems, demonstrating enhanced stealth capabilities in LR attacks.