Statistical Significance of Wavelet Extracted Features in the Condition Monitoring of Ceramic Outdoor Insulators

Abstract

The components of high voltage transmission lines such as towers, insulators, wires, and accessories are continuously subjected to severe environmental conditions. As a result, it is necessary to monitor their health condition to prevent any sudden interruption in the supplied load and to allocate the maintenance investments where they are highly needed. Defective discs of ceramic insulators essentially contain partial discharge (PD) activities; i.e., the presence of PD activities may characterize the insulator's poor condition. The detection of radio frequency (RF) waves emitted from the PD activities is an emerging technique to monitor the insulator's condition during its operation. In this paper, various artificial defects are introduced to the insulator discs and the corresponding RF signatures are captured using a high frequency sensor under the normal operating voltage. Several features are extracted from the captured signals after their decomposition using the discrete wavelet transform. The analysis of variance test has been adopted to evaluate the significance of each feature and level in identifying the defect type. This step facilitates the training of an intelligent classifier that will automatically distinguish the insulator strings that has to be replaced along the line.