Using PMU Data for Anomaly Detection and Localization

In Chap. 8 , we showed how to operate a system in SCADA timescale. However, we need faster timescale for some applications. With the continuous deployment of phasor measurement units (PMUs), we can analyze some part of the network at faster timescale. In this chapter, we will use two examples to show how to identify events and plan for the future. The goal is to illustrate the flow of information for situational awareness as discussed in Chap. 2 . To identify an existing event, we will show how to locate the source of forced oscillations. For forecast events, we use dimension reduction methods and the linear dynamical system theory for performance guarantees. Since the PMU data are quite dense due to the extremely high data resolution, we introduce principal component analysis (PCA) and its variational form, e.g., robust PCA (RPCA). They show how to conduct fast data analytics within a short period of time.