Detection of high impedance arcing faults using a multi-layer perceptron

Abstract

The authors present an arcing fault detector, motivated by the advances in neurocomputing in pattern recognition, that uses a simple preprocessing algorithm. A feedforward three-layer perceptron was trained by high-impedance fault, fault-like load, and normal load current patterns, using the backpropagation training algorithm. The neural network parameters were embodied in a high-impedance arcing fault detection algorithm, which used a simple preprocessing technique to prepare the information input to the network. The algorithm was tested by traces of normal load current disturbed by fault currents on dry and wet soil, an arc welder, computers, and fluorescent lights. The algorithm showed good performance in identifying faults disrupted by arc noise as well as good discrimination between faults and fault-like loads. >