Derivation of simulative fault data from normal operating data for on-line monitoring and diagnostic system

More and more on-line monitoring and diagnostic system have been applied in power system to ensure the reliability of its HV power equipment. The diagnostic system has to be built up according to the actual fault patterns of the equipment. However, because of their low on-site failure rate, the real fault data are usually scarce, which restricts the validity verification of diagnostic method and corresponding algorithm. Hence, the method of simulative fault data derived from the real normal operating data was presented to provide a solution. Based on the measuring data from a 110 kV HV bushing on-line monitoring system, some fault data were simulated. In the system an artificial neural network (ANN) is constructed as the diagnostic algorithm, which is employing the adaptive resonance theory (ART). It is concluded that applying the method of simulative fault data was convenient for constructing the diagnostic system.