Application of Machine Learning on Power System Dynamic Security Assessment

Abstract

This paper addresses the on going work of the application of Machine Learning on Dynamic Security Assessment of Power Systems. Several techniques, which have been applied for the Dynamic Security Assessment of the Greek Power System are presented. These techniques include off-line Supervised learning (Radial Basis Function Neural Networks, Support Vector Machines, Decision Trees), off-line Unsupervised learning (Self Organizing Maps) and online Supervised learning (Probabilistic Neural Networks). Results from the application of these methods on operating point series from the Greek Mainland system and the Power System of Crete island show the accuracy and versatility of the methods.