Load redistribution attacks on power system operation and stability

Abstract

The integration of advanced communication technologies and metering infrastructures has led to the creation of a complex cyber network that enhances communication within traditional power grids. While this integration boosts the overall efficiency of power systems, it also introduces significant vulnerabilities to cyber-attacks. Rather than focusing on purely physical attacks, malicious actors can exploit communication channels to inject false data into measurement and control signals, thereby disrupting system operations. Among these, Load Redistribution Attacks (LRA) have gained prominence as a specific type of False Data Injection Attack (FDIA), posing serious threats to power system stability and security.

The thesis presents several key contributions. First, a new LRA model is introduced that incorporates both reactive power and voltage measurements, expanding the scope of the attack from increasing system costs to corrupting specific nodal voltages. To simplify the resulting optimization problem, a Linear Power Flow (LPF) model is employed, enabling efficient attack execution. Second, a stealthy LRA method is developed that eliminates the need for network admittance information. By using the Power Transfer Distribution Factor (PTDF) matrix and employing ridge regression, this method allows attackers to launch effective LRAs based solely on available SCADA data, without requiring detailed network knowledge.

Further, this thesis explores the potential of LRAs to degrade small-signal stability (SSS) in power systems. A novel strategy is introduced that bypasses constant voltage model-based SSS analysis, enabling attackers to destabilize the system while remaining undetected under conventional stability assessments. Lastly, the impact of high EV penetration on power systems is analyzed, with a proposed consecutive LRA model targeting load spikes caused by synchronized EV charging. This multi-slot bi-level optimization approach accounts for transmission and distribution network interactions and is solved using a greedy algorithm.

In summary, this thesis contributes to a deeper understanding of cyber-physical security risks in power systems by proposing novel attack models and providing insights into developing robust defense strategies. The findings highlight the need for more comprehensive stability analyses and proactive measures to protect future power grids from emerging cyber threats.