Publication

Abstract : Polymeric insulators deliver superior pollution performance owing to their excellent hydrophobic characteristics, which aid in forming discrete water droplets on the surface of insulators. However, due to severe humidification, the insulator surface gets susceptible to aging and temporarily loses its hydrophobic property. As a result, continuous water films are formed, leading to dry-band arcing and subsequent flashover of the insulator. Hence, accurate identification of hydrophobicity grade (HG) is necessary to preclude such events of flashovers as well as maintain longer operating life of the insulator. Considering these facts, a novel graph-image processing-based remote HG identification framework is proposed in this contribution. To this end, water droplet images on the surface of 11kV silicone rubber (SiR) insulators corresponding to different HGs were captured and subsequently converted to graphs using the image visibility graph (IVG) algorithm. The variation in structural connectivity patterns of varying HGs was quantified by extracting topological features from the IVGs and was subsequently classified using the support vector machines (SVM) classifier. The employed HG recognition scheme delivered appreciably good recognition accuracies, suggesting that the proposed methodology can be potentially used for real-time remote condition monitoring of insulator surfaces.