A Novel Transient Kurtosis Measurement Method Based on Adaptive QMF for Fault Location in Noisy Grids

Abstract

The performance of traditional traveling-wave fault location methods deteriorates under strong noise and signal dispersion, primarily characterized by waveform broadening and the impeded extraction of key fault characteristics. To address this measurement challenge, this article proposes a fault location approach based on transient kurtosis measurement (TKM). This method defines the transient kurtosis (quantified by a fourth-order norm ratio) as the target measurand to quantify the sharpness of the fault transient. Instead of using fixed-basis wavelets, an adaptive quadrature mirror filter (QMF) is constructed by reconstructing the lattice coefficients to match the transient characteristics of the fault. This process ensures the maximization of the measured kurtosis via a hybrid optimization algorithm (BFGS and gradient ascent). Experimental results demonstrate the quantitative advantage of the proposed method over conventional Daubechies (db) wavelet techniques: specifically, field tests confirm that the typical fault location error is reduced from 30.27 % to 3.32 % under high-impedance fault conditions, quantifying the method’s superior precision in capturing reflected waves.