Real-time event detection and feature extraction using PMU measurement data

Wide-area situational awareness of grid operators and regional reliability coordinators for large-scale interconnected power systems are essential to prevent broad cascading outages. The need to enhance the power system situational awareness has been increasingly acute over the last decade as the grid evolves dramatically. The fast installation of phasor measurement units (PMUs) over the world generates large amounts of data available for event detection and feature extraction applications. As such, we devote this work to present a methodology for analyzing volumes of PMU measurement data and extracting information for effective event detection and decision making in a real-time fashion. To do so, we apply the principal component analysis (PCA) to process the obtained measurements in order to detect abnormal system behavior and adopt some visualizations to better present the extracted information related to the event. This PCA-based method is capable to directly show the location, magnitude, type and some other information related to the event to help the system operator produce better informed decisions. It can also show the similarity of bus dynamic behavior information hidden in the time-varying data. We illustrate the application of the PCA-based method using the Illinois 42-bus and the WECC 1,511-generator system models. To address the heavier computational burdens as the system size increases, we further propose a partitional PCA (PPCA) based method with improvement in computation time compared to the PCA-based method. Case studies are carried out on a large-sacle system to demonstrate the effectiveness of the PPCA-based method for real-time event detection.