Modelling and prediction of high voltage insulation breakdown under impulse voltages using intelligent algorithms

Abstract

Electrical insulation is one of the most important components in power newlinesystems. It has been reported that the major cause of the failure of power system newlineis the failure of insulation systems. Frequent failures cause frequent interruptions newlineand results in less reliability. The damage to electrical insulation is caused by the newlinephenomenon called dielectric breakdown. Different types of electrical insulation newlineused in power system include solid, liquid, or gaseous insulation systems. The newlinemechanism of dielectric breakdown is different for solid, liquid, and gaseous newlinedielectric materials. Hence, the prediction of dielectric breakdown differs for each newlineof these materials. newlinePredictive methods for dielectric breakdown are so important to maintain newlinereliability, insulation coordination, better design of insulation systems, and newlinepreventive maintenance. Air and cross liked polyethylene are widely used newlineinsulating materials in power systems. Because of these facts, investigations into newlinethe breakdown phenomena in the air(gaseous) and cross-linked polyethylene newline(solid dielectric), have been done in this thesis. Machine learning methods are newlinenowadays very effectively used to solve many engineering problems and their newlineapplication in high voltage engineering is recent. In the first part of the thesis, newlinemachine learning models that effectively predict the breakdown mechanism of air