Computerized system for detection of high impedance faults in MV overhead distribution lines

This paper presents a computerized system for detection of high impedance faults (HIF) in MV distribution feeders. This system has been developed in an R&D Project of AES Eletrpaulo. A HIF occurs when an energized conductor makes contact with objects with a high resistance and there are no significant increases in current. It is common to associate HIF to situations in which there is fallen primary conductor, creating risk of fatal accidents due its proximity to people. This is the main focus of this project. This type of fault cannot be detected by traditional overcurrent protection devices. The method developed herein relies on the use of variables obtained from power quality meters installed at the feeders in the substation, measuring quantities such as current harmonics, sequence components amongst others, with which some indices are calculated to detect the occurrence of a HIF event using artificial neural networks (ANN). The proposed system was implemented and installed in an AES Eletropaulo substation.