Hybrid artificial neural network and finite impulse response based relay algorithm for the detection of stochastic high impedance faults

High impedance faults (HIF) are faults which are difficult to be detected by overcurrent protection relays. This paper presents a practical pattern recognition based relay algorithm for electric distribution high impedance fault detection. The scheme recognizes the distortion of the voltage and current waveforms caused by the arcs usually associated with HIF. The analysis using Finite Impulse Response (FIR) filter bank yields three-phase voltage and current in the low frequency range which fed to a classifier for pattern recognition. The classifier is based on the algorithm using neural network approach. A HIF model was also developed, where the random nature of the arc was simulated using MATLAB.