Partial discharge recognition and localization on transformers via fuzzy logic

The partial discharge (PD) measurement as a sensitive method for early detecting of insulation faults becomes, additionally to the conventional procedures for supervising power transformers, more and more important. Besides, a central problem during the measurements is caused by the outer disturbances, which complicate the recognition and assessment of the PD. In this paper, first the problems of PD measurement on site are described, whereby the occurring noises are classified. The main theme of the paper is a demonstration of the possibilities and experiences in suppressing the pulse shaped noises with the help of fuzzy logic algorithms. After a short introduction to fuzzy logic, its ability to separate the PD from pulse shaped noises is investigated using synthetical signals. Furthermore, the effectivity of this method is examined with PD pulses, which are injected at different points of a transformer coil and decoupled via the bushing. This way, the trial to determine the locations of the pulses with the help of fuzzy logic is successfully performed. A superposition of the measured signals with synthetically generated ambient noises of well known amplitude shows the limits of the method used