An Improved Algorithm to Remove DC Offsets From Fault Current Signals

Fault current signals that are processed by digital relays consist of dc, fundamental, and harmonic components. Filtering algorithms are necessary to eliminate the dc and harmonic components from these signals. Several algorithms have been proposed for this task which vary in their accuracy, response time, and computational burden. The conventional discrete Fourier transform (DFT) can eliminate harmonics and is commonly used to estimate the fundamental frequency phasor, but its accuracy is lower as it does not filter the dc offset. Other algorithms including variants of DFT attempt to improve the accuracy and response time. This paper proposes a technique that takes into account the exponential variation of the dc offset and more accurately determines the fundamental component. The effectiveness of this method is evaluated by simulation on a two-machine system and also compared against existing phasor measurement methods. Simulations confirm that the proposed method can more accurately estimate the fundamental component compared to the existing methods.

I. Introduction

Protection relays require the input voltage and current signals to be free from harmonics and DC components so that fault conditions can be more accurately detected. Filtering algorithms are necessary to extract the fundamental component from input signals that contain harmonics and DC components. The conventional Discrete Fourier Transform (DFT) is most commonly used to extract the fundamental components from the measured waveforms [1]–[6]. The DFT can eliminate the harmonic components but not the DC component. The DC component is a non-periodic signal that has a large frequency spectrum. It contributes to the overshoot and oscillations that are present in the estimation of the fundamental component when using the DFT giving rise to an error of around 15.1% [2].