A neural net based approach for fault diagnosis in distribution networks
Conference Paper
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
A neural network based approach is proposed for the fault diagnosis in electric distribution networks. The method allows the detection of low or high impedance faults in the presence of arcing in grounded or ungrounded systems that are normally not detected by conventional overcurrent protection devices. The method uses a signal preprocessor and a supervised clustering based diagnosis system. The effectiveness of the fault diagnosis method is demonstrated for a field test system using data recorded at two four-wire utility distribution systems.
name of conference
IEEE Power Engineering Society. 1999 Winter Meeting (Cat. No.99CH36233)