Ensuring the Functionality of Digital Protection Devices in the Event of Cyber-Attacks on Microgrids With Distributed Energy Resources

The research aims to develop method for method for verifying data used in digital systems for protection, automation and control of microgrids with distributed energy resources. The research relies on the probabilistic methods, machine learning methods. Research result: the information support of microgrids protection schemes is considered, possible cyber attacks are analyzed, the successful implementation of which may result in a violation of the functionality of digital protection devices of microgrids. A data verification method has been developed using unsupervised learning methods, including isolation forest and the k-nearest neighbors method, which effectively identifies and corrects measurement errors during attacks on the information infrastructure of microgrids protection systems under false data injection attacks. The scientific novelty lies in the fact that an approach has been proposed that ensures the stability of microgrids protection systems during cyber attacks and, thereby, prevents false alarms and failures of protection devices.