A novel technique for internal fault detection of power transformers based on moving windows

摘要

Power transformer differential relays may incorrectly operate due to inrush condition or transformer external faults lead to current transformer (CT) saturation. In this paper, a new technique is presented to distinguish internal fault current from magnetizing inrush current in power transformers. The proposed technique applies two moving windows which both of them estimate magnitude of transformer differential current using different methods. The first moving window method is based on full-cycle Fourier algorithm (long window) which uses one cycle samples of the current and provides precise estimation. The second one is originated from least error square (LES) method that has five sample point input data (short window). LES method has fast response, but its estimation might not be quite accurate. The proposed technique (which is a combination of two mentioned methods) presents a new criterion for discrimination between internal fault currents and inrush currents using the differences between output values of these two estimators. Meanwhile, obtained results demonstrate precise operation of the proposed algorithm for different conditions such as CT saturation and over-flux condition; therefore, it can be a multi-objective technique for transformer protection. Also, a general scheme is presented in order to improve differential protection. In this paper, a real part of 400/230 kV Iranian national grid is simulated using PSCAD software for evaluating the performance of the proposed algorithm. Also, additional validation of the proposed technique is tested offline using data collected from a three-phase, 5 kVA, 60 Hz, laboratory prototype transformer. Copyright © 2013 John Wiley & Sons, Ltd.