An intelligent system for automatic detection of high impedance faults in electrical distribution systems

The paper presents a new methodology for automatic detection of high impedance faults in primary distribution feeders. An intelligent system continuously monitors the feeder response to a short pulse signal injected at the substation outlet. Any significant change in the feeder response not observed during normal operations is prone to be a fault. The supervisory system continuously monitors the feeder and processes the gathered information by means of the Intelligent System. Basically, it is composed of a case database and a rule database that operates with fuzzy numbers. To validate the technique presented here, several tests have been performed in a real feeder of CEMIG at Caratinga, MG, Brazil. The results shown in the paper demonstrate that the method is effective for most of the cases and encourage further investigations.