Securing Smart Grids with Machine Learning

The electricity system is evolving to become more flexible, sustainable, and distributed. As the grid becomes smarter, so too do the tools that can exploit vulnerabilities in its digital backbone. Recently in IEEE Transactions on Smart Grid, Ismail et al. reported an electricity theft detection method using deep neural networks that can achieve 99.3% detection rates with less than 0.22% false positives.