Metodologia para detecção e classificação de faltas em transformadores de potência por meio da Transformada Wavelet Discreta

Abstract

This thesis presents the development of a methodology for the detection and classification of electrical transients in power transformers, for protection purposes. For this, we analyzed several electrical transients involving the power transformers, in order to distinguish the different types of internal faults of the inrush current. The methodology was validated through a test system implemented in the ATP/EMTP software. From this, simulation data were acquired considering several operating conditions of the power transformer. The developed methodology monitors and analyzes the Detail Coefficients (CD) for different levels of decomposition by the Discrete Wavelet Transform of the differential current. The method extracts the characteristics and calculates the variation of the CD levels for the classification of transients. The results showed that during the internal fault conditions the CD presents high and constant peak values, different from what happens in other transients. The methodology proposed in this work is able to identify and correctly discriminate the internal fault conditions of other transients. This property makes the proposed method promising for power transformer protection purposes.