An intelligent system for acoustic inspection of outdoor insulators

Abstract

Condition monitoring of outdoor insulators is crucial to the integrity of distribution and transmission overhead lines. The objective of this paper is to use an Artificial Neural Network (ANN), along with a commercial acoustic sensor to measure and classify the different types of arcing on outdoor insulators. Experiments were performed, where both corona and dry band arcing were generated under lab test conditions which mimicked reality as closely as possible. The sound produced by corona, dry band arcing and acoustic noise was recorded using a commercial acoustic sensor. The problem of detecting corona, dry band arcing, or noise constituted a three-class pattern recognition problem, which is considered in this paper. The acquired acoustic signal was transferred to a low frequency signal using an envelope detection technique. Both the 100 and 150 Hz components of the envelope were used as input feature vectors for the developed ANN. Results show an average of around 90% success rate in classifying the measured acoustic signals.