Análise de estabilidade transitória utilizando máquinas de aprendizado : uma abordagem baseada em redes neurais artificiais e SVM

Abstract

An electrical energy system is subject to a series of disturbances of differents grades of severity. It is designed and operated in a way that is stable for a given set of contingencies, involving the various types of possible short circuits in the system. The study of the stability of electric power systems allows to predict the behavior of such systems after a contingency for any load and generation conditions. The problem of transient stability in an electric power system revolves around maintaining equilibrium in the system under normal conditions and also of the ability to return to such conditions after being subjected to a severe transient disorder. The response of the system to a severe transient disturbance depends on the excursions at the rotor angle and the non-linearity of the power-angle ratio of the generators. When applying artificial neural networks and support vector machines (SVM) to recognize patterns of electrical systems simulations, it is possible to infer about the stability of it. This work aims to study and evaluate the transient stability of a 34-bus simulated electrical power system with the softwares Anarede and Anatem, using artificial neural networks and SVM, presenting and comparing application results in simulations, as well as retrieving results from related pre-existing works. In this work, the methods used are effective for the task of recognizing instabilities in the simulated system.