Combined use of intelligent partial discharge analysis in evaluating high voltage dielectric condition

This paper describes the results of synthesised high voltage impulse tests, conducted on surrogate dielectric samples. The tests conducted under laboratory conditions, were performed using contoured electrodes submersed under technical grade insulating oil. An escalating level of artificial degradation within surrogate samples was assessed, this correlated against magnitude and frequency of events. Withstand of partial discharge activity up to a point of insulation breakdown was observed using a conventional elliptical display partial discharge detector. Measurements of PD activity were simultaneously captured by virtual scope relaying data array captures to a desktop computer. The captured data arrays were duly processed by an artificial neural network program, the net result of which indicated harmony between human-guided opinion and the software aptitude. This paper describes work currently being undertaken for the identification and diagnosis of faults in high voltage dielectrics in furthering development of AI techniques.