D-PMU Data Generation and Anomaly Detection Using Statistical and Clustering Techniques

Anomaly and event data analysis is necessary for enhanced situational awareness to help with decision support. Rapid growth in the deployment of synchrophasor technologies provides rich data sets to make it possible. In particular, the distribution systems with high penetrations of distributed generations, associated control, and distributed automation provide multiple components to respond to internal or external events. Along with the recent technological advancements in sensor and controller equipment, the installation of inverter-based resources and controllable loads makes event analytics more challenging. Early adoption of distribution phasor measurement units (D-PMUs) to record disturbances provides synchronized high-resolution data. This work focuses on comprehensive studies of unsupervised anomaly detector schemes, considering the statistical and clustering methods as base anomaly detectors. Rigorous testing and analysis of detection methods have been conducted using the real field equivalent simulated network model. A comparative study is also performed using carefully selected validation metrics to illustrate the distinguishing characteristics of each method for visualizing their usefulness.