Comparing Discrete Wavelet Transform (DWT) with Discrete Fourier Transform (DFT) implemented for digital relays

This paper critically assesses the implementation of algorithms for Phasor Estimation using the Discrete Wavelet Transform (DWT) in a digital relay. The performance of different wavelets using varying levels of recursion in the implementation are presented. It finds that the computational power required to implement Phasor Estimation using the Discrete Wavelet Transform is similar to using the Discrete Fourier Transform (implemented using the FFT). It found that Phasor Estimation using the Discrete Wavelet Transform offers no advantage when detecting the fundamental phasor in a signal with harmonic content. Changing the wavelet used or the level of recursion had little effect on the accuracy.

I. Introduction

Most digital relays use phasor estimation techniques based on the Discrete Fourier Transform (DFT). These filtering algorithms work well in normal conditions when the input waveforms are purely sinusoidal, however the current and voltage signals during the fault interval often contain large harmonics and an exponentially decaying DC offset which can cause large inaccuracies in calculating the fundamental components. The Discrete Fourier Transform is also slow to converge as it requires a full cycle of data to calculate the phasors accurately.