Evaluation of Transmission Line Fault Based on Various Mother Wavelets

Abstract

Multi-resolution analysis and data feature extraction have received a lot of attention in signal processing. The time-frequency analysis method provides information on joint distribution in both the time and frequency domains and is a potent mathematical tool for analysing time-varying non-stationary signals. The Short-Time Fourier Transform (STFT) is one of the standard time-frequency distribution functions. However, the wavelet transform offers great frequency resolution at low frequencies and high time resolution at high frequencies, which offers constant, equally spaced time-frequency localisation. In this study, to extract the optimal feature vector, the single-phase ground short-circuit fault signal was obtained in the MATLAB environment.  Two methods were applied to determine the most suitable wavelet family. The optimal resolution level was determined using Shannon entropy, while the Minimum Description Length (MDL) method was used to select the most suitable mother wavelet family. Accordingly, various wavelet families, including db8, sym5, coif5, bior1.3, and rbio3.1, were tested in discrete wavelet analysis. The results demonstrate that, when the high- and low-frequency components of the fault signal are analyzed in the feature vector extraction using the db8 wavelet, the similarity of the approximation coefficients to the main signal is not significantly affected. Moreover, the feature vector enables the most transparent and accurate identification of transient fault components, particularly in terms of the critical detail coefficients.