Comparison of different fourth order Hilbert fractal antennas for partial discharge measurement

Three Hilbert fractal antenna designs are proposed in this work to capture and classify common types of partial discharge (PD) in an oil insulated system. Each antenna design shows unique characteristics in terms of resonant frequencies, inception voltage, classification capabilities and noise performance. Three types of PD signals are artificially generated; namely, corona, surface and sharp PD. The captured signals from each antenna design are analyzed then fed to a trained artificial neural network for classification. A recognition rate of 97% is achieved when classifying the different types of PD using one of the proposed antennas. Moreover, the SNR of signals captured from each antenna design are analyzed to determine the best antenna for PD detection under intense noisy environments.