Decision Tree-Based Methodology for High Impedance Fault Detection

Abstract

This paper presents a high impedance fault (HIF) detection method based on decision trees (DTs). The features of HIF, which are the inputs of DTs, are those well-known ones, including current [in root mean square (rms)], magnitudes of the second, third, and fifth harmonics, and the phase of the third harmonics. The only measurements needed in the proposed method are the current signals sampled at 1920 Hz. It will reduce the cost of hardware compared with methods that use high sampling rates. A new HIF model is also used. The data of current signals are from the simulation of Electromagnetic Transients Program (EMTP). The DT algorithm trained can successfully distinguish the HIFs from most normal operations on simulation data, including switching loads, switching shunt capacitors, and load transformer inrush currents. Testing on experimental data is recommended for future work.