Supervised Learning Paradigm Based on Least Square Support Vector Machine for Contingency Ranking in a Large Power System

摘要

In modern emerging power system many contingencies and critical operating conditions present a potential threat to system’s stability. An intelligent designer at energy management center requires a paradigm which can not only predict such cases but also suggests an effective strategy for preventive control. This paper presents a least square support vector machine (LS-SVM)-based classifier to identify and rank the critical contingencies in a standard IEEE-39 bus Network (New England). This paradigm works in two stages. In first stage, the identification of two indices, i.e., voltage reactive performance index PI_(VQ) and MVA line loading index PI_(MVA) is carried out and in next stage the classification of contingencies is carried out. The proposed approach shows promising results when compared with recent contemporary techniques.