Resilient Distributed Economic Dispatch in Smart Grids

We study resilient distributed economic dispatch in smart grids where some of the agents are compromised by attackers. The objective is to distributively compute the optimal power generation amount of each normal agent such that the balance between power generation and demand is satisfied within all normal agents despite the erroneous effects from malicious agents. To this end, we propose a novel algorithm for smart grids with directed communication topologies. Our algorithm detects the misbehaving agents and recovers the initial local demands of normal agents such that the final outcome of our algorithm is exactly the optimal solution to the resilient distributed economic dispatch problem. Specifically, our algorithm can succeed not only when the adversarial agents are neighboring and collaborating, but also under serious situations where nearly half of the agents turn malicious in the network. Moreover, our algorithm is fully distributed and is suitable especially for large-scale smart grids as each agent requires only the information sent by its direct in-neighbors. Lastly, numerical examples using IEEE standard test systems verify the efficacy of our algorithm.