A Novel Adaptive Out of Step Protection in Synchronous Generators Using Support Vector Machine Algorithm

Abstract

The out of step condition occurs mainly due to the propagation of fault throughout the power system network. The fault may be due to power swings, phase faults, loss of excitation, under or over frequency. Particularly, in synchronous generators, the loss of excitation causes the out of step condition which leads to the wide-area blackout. In this paper, the synchronous generator’s frequency, phase and magnitude of voltage are directly observed from the PMUs connected to its respective buses. The responses of synchronous generator for different operating conditions are observed directly from the PMU. This procedure is developed and tested in MATLAB/Simulink software. The PMU data is extracted to classify the loss of excitation(LOE) fault, three-phase faults and the normal operating condition using one of the premiers supervised machine learning algorithms called Support Vector Machine algorithm. The propound algorithm is done in MATLAB environment and tested in IEEE 14 bus system. The results project the performance of the algorithm for fault classification in synchronous generators.