Detection of high impedance faults in overhead multi grounded networks

This article presents a methodology for detection of high impedance faults (HIF). HIF occurs when e.g. a cable makes contact with objects of high electric resistance, resulting in a nonsignificant increase in current. Such faults cannot be detected by traditional protection devices that operate due to overcurrent. The developed methodology is based on making use of variables from a power quality meter (PQM), such as harmonic and sequence components of the currents measured at a substation. Such variables compose descriptors for decision making in artificial neural networks (ANN). In order to test the proposed method HIF events were simulated as well as other common network events using the ATP software. The proposed framework is part of a HIF detection system to be implemented in a substation of an electric utility in the state of Säo Paulo.