MOSA Monitoring Technique Based on Analysis of Total Leakage Current

Abstract

Currently, the majority of metal-oxide surge arrester (MOSA) monitoring techniques are based on total leakage current decomposition of their capacitive and resistive components. However, these techniques are subject to some financial, technical, and practical limitations, which can hamper their usage on the field. In this paper, a new monitoring technique for the zinc-oxide surge arrester is proposed. The technique is based on the feature extraction of measured total current signals and on the analysis of these characteristics by artificial neural networks. The measured total currents were obtained from station class surge arresters submitted to their maximum continuous operating voltage. In the lab tests, six different kinds of failures were simulated on the tested MOSA with the purpose of evaluating the capacity of the technique to distinguish different operating conditions of surge arresters. Hit ratios greater than 98% were obtained in the identification of the operating conditions. The results show the viability of the technique on MOSA monitoring procedures.