Analyzing And Mitigating Cyberattacks In Power Distribution Systems Using ML Techniques
Abstract
The security of cyber-physical systems is crucial for the protection of power distribution systems. The remote-controlled switches in the distribution network are intended to be altered by the direct switching attacks that are being planned. Certain designs may result in voltage issues and power outages due to their predominantly radial mode of operation. The parameters of the assailant are believed to be known by current optimization methods, which demonstrate their interaction with the power system operator (defender). Due to this, they become less beneficial. Combining data from decentralized security systems with centralized processes is an approach that is increasingly employed to identify coordinated cyberattacks. This may pose an issue for one of the components. This results in the development of novel mathematical models for both the defense and the assailant. The models determine the attacker’s objectives by comparing assaults that have been discovered in a decentralized manner, without assuming that they are aware of the attacker’s system configuration.