Pattern Recognition‐Based Approach for Dynamic Vulnerability Status Prediction

This chapter presents a novel post‐contingency pattern recognition method for predicting the dynamic vulnerability status of an Electric Power System. The methodology begins with the determination of post‐contingency dynamic vulnerability regions (DVRs) using Monte Carlo (MC) simulation and empirical orthogonal functions (EOFs), which allow finding the best pattern functions for representing the particular dynamic signal behavior. By applying the introduced DVR concept, the chapter describes a novel approach to determining the post‐contingency transient, voltage, and frequency stability (TVFS) vulnerability status of an Electric Power System. The chapter also presents a comprehensive approach for predicting the power system's post‐contingency dynamic vulnerability status. Based on the requirement of dynamic vulnerability status prediction, a two‐class classifier is structured that sorts the power system vulnerability status into “vulnerable” or “non‐vulnerable” using the pattern vectors obtained from the pattern recognition method as inputs of support vector classifier (SVC).