A Unsupervised Machine Learning Protection System For Large Scale Iot Grids




The electrical grids are more dependable, secure, and efficient thanks to smart grid technology.For effective and dependable power distribution, new vulnerabilities are raised by its strong reliance on digital communication technology. The unsupervised anomaly in this paper.The idea of measurement correlation-based detection has been put forth. The objective is to create a scalable a large-scale anomaly detection engine for smart grids that can distinguish between an actual fault and a commotion and a clever cyber-attack. The suggested technique utilises feature extraction by Using symbolic dynamic filtering (SDF) to lighten the computational load and find causal relationships among the subsystems. The outcomes of the simulations on the bus systems support the performance of the suggested method under various operating circumstances. The outcomes demonstrate accuracy of 99%, a 98% true positive rate, and a less than 2% false positive rate.