Recognition on ultra-high-frequency signals of partial discharge by support vector machine

This paper presented a novel approach to recognize ultra-high-frequency (UHF) signals of partial discharges (PDs). Four artificial insulation defect models were designed to generate PD UHF signals, which were detected by a Peano fractal antenna in experiments. Wavelet packet (WP) decomposition was used to decompose PD UHF signals into multiple scales. A group of energy parameters and fractal dimensions of PD UHF signals were computed and used as the input parameters of a support vector machine (SVM), which was used as the PD pattern classifier. For verifying the results of this approach, a back-propagation neural network (BPNN) was also used for pattern recognition of PD UHF signals. The recognition results showed that the SVM and the proposed parameters were qualified for PD pattern recognition and the SVM had advantages over the BPNN for the purpose.