Utilização da transformada de Wavelet para detectar variações anormais de freqüência em sistemas de geração distribuída

Abstract

This dissertation focuses the importance of registering eventual frequency variations in Electric Power System which may cross the established limits for normal operation. These can cause serious problems in the operation of sensitive equipments connected to the electric grid, such as the turbo-generators employed in thermoelectric plants. For power system frequency supervision it is used frequency relays which act on protection devices when abnormal conditions are detected. In this context, the major objectives of this work consist of proposing the implementation of an algorithm capable of detecting frequency disturbances characterized by under or over frequency situations and quantifying the time during which the system was operating in such conditions. To achieve this objective the wavelet technique called multiresolution analysis was used for detection of generated disturbances. The data obtained can then be treated in order to prepare a predictive maintenance, mainly in steam turbines which are equipments more subject to damages when in long lasting operation under abnormal frequency conditions.