PD behavior of gas and pattern recognition in co-axial cylinder electrodes

In high voltage apparatus, for example GIS, partial discharge (PD) can occur because of impure particles showing up, especially, some kinds of metallic particles appearing during service. The PD failures are dangerous and detrimental for the GIS, because they could destroy the insulating system of the high voltage apparatus, resulting in a great deal of economic damage and personal safety issues. So the diagnosis, monitoring and prediction of the PD failures are of concern for electrical engineers. Experiments have been made for simulation of the GIS structure and condition service under different kinds of applied voltage and existance of many sorts of metallic particles which possess variable materials (for instance, copper, stainless steel, steel and so on) in different diameters, lengths and geometry configurations. Through the practical experiments in the laboratory the relationships between inception voltage of PD, PD quantity, length, materials and geometry have been obtained. Meanwhile the PD signals have been recorded by the measurement systems. The position of the particles located shows different waveform and behavior, so that it could supply a way to detect the PD failure characteristics, location and attributes, and it helps analysis, judgement and assessment of the service conditions. The recognition system of PD pattern has been designed for recognizing PD failures according to artificial neural network. experiments and practices demonstrate that the method is very useful and effective.