Multi-resolution Analysis Algorithm for Fast Fault Classification and Location in Distribution Systems

Abstract

This paper presents a new method for fault classification and location based on the Discrete Wavelet Transform decomposition and signal reconstruction - a type of Multi-Resolution Analysis. The designed signal-processing stage, which encompasses various signal transforms, plus the aforementioned decomposition in several frequency bands and the calculation of the signals’ energy, provides a consistent generalization of the features that characterize the fault signal. Then, this data is fed into ensemble Machine Learning algorithms. The results show that this method is reasonably accurate while requiring a tiny amount of fault data, expanding the capabilities of Traveling Wave relays to achieve an accurate fault classification and location in just microseconds.