Feature Extraction and Visualization of Power System Transient Stability Results

Abstract

Transient stability simulation studies of large-scale power systems often generate large amounts of data. This can make it difficult for power system planners to understand the overall system response or to identify portions of the system with unusual signals. In this paper we present a novel approach that utilizes clustering to extract common features in the voltage magnitude and frequency response signals, and to identify outliers. The geographic visualization of the results is also discussed. Results are demonstrated using the IEEE 118-bus system and a 16 000-bus real-world system model.