DC Offset Removal Algorithm for Improving Location Estimates of Momentary Faults

Abstract

Fault location algorithms require the input of voltage and current phasors when estimating the distance to a fault. In the case of momentary faults, phasor calculations are often complicated by the presence of an exponentially decaying dc offset. Fourier filters and cosine filters, popularly used by most phasor estimation algorithms in relays, are successful in filtering out most, but not all, of the exponentially decaying dc offset. This dc offset affects the accuracy of voltage and current phasors and may result in a significant error in location estimates. Therefore, this paper presents a dc offset removal algorithm to improve the fault location estimates of momentary faults. The algorithm uses the rms-wavelet method for fault detection and estimates voltage and current phasors using non-linear least squares methods. The proposed method uses variable window size in calculating phasors and estimates a single, but more accurate fault location than multiple locations estimated by the Fourier and cosine filters. The method is validated using simulated and actual field data. The location estimates are shown to be accurate within 5.6% in all cases.