Automated power system event detector and classifier

In this paper a methodology for the on-line detection and classification of AC power system events measured at the interconnecting point between two parts of a power network is presented. These events can be isolation faults, broken lines, switching capacitors, starting motors, energizing transformers, and swinging turbines. The method uses the three phase-to-ground voltages and the three phase currents. Information about the net configuration as well as the neutral point treatment of the power system is not needed. The developed method comprises a signal pattern recognition unit embedded into a fuzzy expert system. The final result of the event evaluation is a list of all likely events including information about their direction, ranked according to their probability of recognition.