Mitigating electric power system vulnerability to worst-case spatially localized attacks

This paper proposes an approach to mitigate power system vulnerability against worst-case spatially localized attacks (SLAs), which are defined as the failure of a set of system components, distributed in a spatially localized area, due to natural hazards or malicious attacks, while other components outside of the area do not directly fail. This problem is mathematically formulated as a tri-level defender-attacker-defender model, where the inner level optimizes the power dispatch to minimize system vulnerability (quantified as power demand drop), the middle level identifies the most disruptive spatially localized attack, and the outer level makes an optimal mitigation decision, including protecting vulnerable components and building new lines, to reduce the SLAs-induced vulnerability. This model is exactly solved by a proposed decomposition algorithm. Case studies on the IEEE 14 bus test system demonstrate the effectiveness and efficiency of the proposed approach.