A GDSC-Based Technique to Distinguish Transformer Magnetizing From Fault Currents

Abstract

This paper presents a novel technique to distinguish a transformer magnetization inrush current from a fault current. The proposed scheme is based on the behavior of fundamental and second-harmonic positive- and negative-sequence components. These four components are computed by using generalized delayed signal cancelation (GDSC). After a transient analysis for the GDSC filters, related to the behavior of those four components detection, some features were proposed. These features depict the trajectory change of the most relevant components in the $\alpha \beta$ plane and their amplitudes for short-circuit discrimination, rather than waiting for filters convergence or waiting one cycle to analyze waveform symmetries. Some features instantly change after an electrical event, making possible a distinction by analyzing a short number of samples. This is the main competitive advantage of the proposed technique. In order to evaluate the proposed method performance, 2200 test cases based on an actual power system were simulated. Also, current-transformer (CT) saturation influence on the technique, computational burden, and embedded implementation were analyzed. In about 95% of the results obtained, the response time of the proposed method was around one fourth of a fundamental cycle.