Combined support vector classifiers using fuzzy clustering for dynamic security assessment

This paper addresses the problem of dynamic security classification of electrical power systems using class pattern recognition with a system of combined classifiers, where each classifier is a support vector classifier (SVC) and each of the SVCs is trained on a subset of the data. The subsets are specified by the fuzzy C-means clustering algorithm (FCM). The strength of the combined classifier stems from the combination of the single classifiers. As a test-bed we have used real data from the power system of Crete, Greece.