Condition based monitoring of gapless surge arrester using electrical and thermal parameters

Abstract

A new method to assess the condition of metal-oxide surge arresters is presented. The thermal image and third harmonic leakage current are used as an indicator. The correlation between the leakage current and temperature of the arrester is processed using a neural network. The temperature profile of arrester, ambient temperature and humidity were as input to the neural network and the peak value of the third harmonic resistive current as a target. Results are presented with the training of neural network close to the target and testing result is 98% successfully.