Modelling of discharge inception and extinction in dielectric voids using artificial neural network

This work attempts at modelling of discharge inception and extinction voltages in dielectric voids applying artificial neural network with supervised learning. The effect of void thickness, the ratios of the void diameter and dielectric thickness to the void thickness are considered. The artificial neural network (ANN) is trained by the digitally simulated data obtained by a solution of empirically derived voltages across voids of different shapes and sizes. The results obtained from the ANN in a range of practical dielectrics are found to be correct within a few % indicating its effectiveness as an efficient tool in estimation