Partial discharge pattern recognition and associated technologies

The electrical detection of individual partial discharges from high power plant insulation has proved an effective method of monitoring insulation integrity (since, irrespective of the causative stress, degradation generally results in partial discharge activity) and a stress condition which can, in itself, cause severe degradation and failure. It has been recognised for some time that individual discharge magnitudes can indicate the relative size of degradation sites and that the magnitude/phase relationships of these pulses on the power cycle and the way these vary with applied voltages and time can provide valuable data on the nature and form of degradation prevalent. The application of digital technology to this form of measurement technique has enhanced its potential further and created a new generation of discharge detectors. The development of artificial neural networks, and their link to pattern recognition processes, has made their application to partial discharge pattern recognition and associated correlations with degradation an obvious one. Given this application and their use in a variety of condition monitoring situations, the authors provide a brief paper for the general audience on this technology