Análise do desempenho de metodologias para detecção de correntes de inrush em sistemas elétricos de potência

Abstract

The transformer is one of the most important parts of a power system and its high cost must be taken into account. Depending on its severity, internal faults can cause equipment damage and destruction, so it is necessary to protect it from these situations. The presence of protection may not completely prevent the damage caused, but it avoids as much as possible the problems that can aggravate the transformer’s operating principle. This demonstrates that a reliable protection system is of paramount importance for the transmission of electrical energy. However, even an efficient protection system can be exposed to unusual occasions. A good example of it is the presence of the transformer’s magnetizing current when energized without load, known as inrush current, which confuses the relay operation, causing a false tripping of the protection. In this context, the objective of this work is to compare the performance of inrush current detection methods present in the specialized literature. To this end, a database containing current signals obtained from simulations of a test system implemented in ATPDraw software was used to analyze the performance of the methods, which were implemented in MATLABr. The following techniques had their performances analyzed: Linear Prediction Method (LPM), Method Based on Waveform Asymmetry Analysis (MBWAA), Waveform Longitudinal Symmetry Coefficient Method (WLSCM) and Adapted Morphological Gradient Method (AMGM). From these simulations, it was possible to obtain the response of each method against different sampling rates, short circuits events, switching angles and for transformer energization with and without load. In this way, it was possible to determine which of the techniques best performs the function of detecting inrush currents under the simulated operating conditions. Keywords: Perfomance. Inrush Current. Inrush detection. Electric Power Systems. Power Transformers.