Field verification of a novel and simple partial discharge detection method for generator applications

Abstract

This study introduces a simple and novel partial discharge (PD) detection method for generator condition monitoring applications. The proposed method uses statistical operator ‘Variance’ which is proportional to the power of an electrical signal. It was validated by field measurement conducted at four in service hydro-generators: two having heavily aged windings and the other two having identical new. First, accuracy of detecting PD pulses was verified by comparing with wavelet-based method. The proposed method detected all the PD pulses with a ‘Precision’ higher than 90% and recall of 100%. As the second step phase resolved PD (PRPD) patterns generated using the proposed method were compared with commercial PD monitoring systems. It was found that the overall similarity was above 0.75. Finally, extracted PD pulses were used to calculate the polarity wise dissipated PD energies. The calculated energies were correlated with identified PD sources through series of off-line tests and visual inspections. It was observed that discharges due to loosen windings and corona activity were well correlated with estimated PD energies. All these validation processes indicated that the proposed method effectively and accurately detect PD pulses and can be used as a diagnostic tool for generator applications.