In this paper, we present a comprehensive study of smart grid security against cyber-physical attacks on its distinct functional components. We discuss: (1) a function-based methodology to evaluate smart grid resilience against cyber-physical attacks; (2) a Bayesian Attack Graph for Smart Grid (BAGS) tool to compute the likelihood of the compromise of cyber components of the smart grid system; (3) risk analysis methodology, which combines the results of the function-based methodology and BAGS to quantify risk for each cyber component of the smart grid; and (4) efficient resource allocation in the smart grid cyber domain using reinforcement learning (extension of BAGS tool) to compute optimal policies about whether to perform vulnerability assessment or patch a cyber system of the smart grid whose vulnerability has already been discovered. The results and analysis of these approaches help power engineers to identify failures in advance from one system component to another, develop robust and more resilient power systems and improve situational awareness and the response of the system to cyber-physical attacks. This work sheds light on the interdependency between the cyber domain and power grid and demonstrates that the security of both worlds requires the utmost attention. We hope this work assists power engineers to protect the grid against future cyber-physical attacks.
展开▼