Harmonic Selection-based Analysis for High Impedance Fault Location Using Stockwell Transform and Random Forest

Abstract

High Impedance Faults (HIFs) originate from the contact between an energized conductor and a high impedance surface. In distribution systems, the HIFs location is an issue that has not been completely solved due to the low fault current and varying impedance, which inhibits traditional fault location techniques from correctly functioning. Thus, this paper assesses the potential of the Random Forest algorithm to be employed to locate HIFs in power distribution systems. The main idea is based on the frequencies extracted by the Stockwell Transform from the phase and neutral currents measured only at the system substation using real HIF signals, thus performing a power quality data analysis. The results are promising, with high identification rates, even with noisy current signals. Additionally, the methodology can help researchers to better select their datasets for supervised-learning-based HIF location methods.