Prediction of Breakdown Strength of Non-cellulosic Insulating Materials using Artificial Neural Networks

In this research work, two sets of experiments are performed in the high voltage laboratory to measure different electrical parameters of insulating materials like breakdown strength, relative permittivity, loss tangent, etc. for a few non-cellulosic materials like teflon, nomex, laminated nomex, glass bonded mica, epoxy resin bonded mica paper, and epoxy resin bonded fiber glass. Also, Artificial Neural Networks (ANNs) models are proposed for the prediction of breakdown strength, considering its dependency on other electrical parameters like relative permittivity, loss tangent, volume resistivity, and thickness of material. The ANN model results are compared with those obtained experimentally and also with the values already predicted from an empirical relation suggested by J. W. Swanson and F.C. Dall. The reported results indicated that the breakdown strength predicted from the ANN model is in good agreement with the experimental values.