Fault localization in Smart Grid using wavelet analysis and unsupervised learning

Abstract

A wavelet based fault localization method in Smart Grid (SG) systems is proposed in this paper. In SG systems, voltage, current, frequency and phase measurements can be collected in real-time using phasor measurement units (PMUs). Based on the wavelet analysis of these measurements, the signal features can be extracted by computing the maximum wavelet transform coefficients (WTCs) and further processing them with a new hybrid clustering algorithm. The clustered signal features then form a fault contour map which can be used to locate faults in the SG system accurately. Both long-term and short-term faults of transmission line, transformer, generator, and load, which are major components of SG systems, are simulated in PSCAD and PowerWorld using the IEEE New England 39-bus system to verify the proposed method. The numerical results demonstrate the feasibility and effectiveness of our proposed method for accurate fault localization in SG systems.