Abstract
Ensuring secure and reliable operations of the power grid is a primaryconcern of system operators. Phasor measurement units (PMUs) are rapidly beingdeployed in the grid to provide fast-sampled operational data that shouldenable quicker decision-making. This work presents a general interpretableframework for analyzing real-time PMU data, and thus enabling grid operators tounderstand the current state and to identify anomalies on the fly. Applyingstatistical learning tools on the streaming data, we first learn an effectivedynamical model to describe the current behavior of the system. Next, we usethe probabilistic predictions of our learned model to define in a principledway an efficient anomaly detection tool. Finally, the last module of ourframework produces on-the-fly classification of the detected anomalies intocommon occurrence classes using features that grid operators are familiar with.We demonstrate the efficacy of our interpretable approach through extensivenumerical experiments on real PMU data collected from a transmission operatorin the USA.