Power Consumption-based Anomaly Detection for Relay Protection

With the rapid development of the smart grid in recent years, more and more smart devices are applied in the smart grid. As a critical monitoring and control device, the protection relay helps improve safety, reliability, and simplify of the smarter grid. However, due to the lack of effective anomaly detection and protection methods, the relay protectors are vulnerable to different types of attacks. In order to achieve the anomaly detection of the relay protector, we propose a monitoring method based on the power consumption side channel. By collecting the power consumption of the relay protection, we can acquire the features which reflecting the running state of devices from the power consumption and train a machine learning (ML) model. Finally, the anomaly state can be detected by the ML model. The test results on the relay protection platform in the laboratory show that the accuracy of anomaly detection can be up to 99.4%.