An automated recognition system of ultra-high-frequency PD in transformers

The ultra-high-frequency (UHF) technique of PD detection has newly been put forward to study the discharge properties in transformers. In this paper, the design of an automated recognition system of UHF PD is described in detail. The system hardware includes UHF antenna and spectral analyzer. And the software consists of four parts, namely, data acquisition, discharge patterns, feature extraction from the measured patterns and pattern recognition using neural network (NN) with genetic algorithms (GA). Application of the system to some typical PD models is also presented and discussed. Experimental results show that the designed system has superior performance to measure the time and amplitude of the high-frequency (adjustable center frequency between 500-1000 MHz) and narrowband (5 MHz) signals emitted by PD, and it has strong functions of data processing, display, data analysis and recognition.