부산대학교 도서관

Icing can deteriorate the electrical performance of overhead power line insulators, thus increasing the risk of ice flashovers during operation. Evaluating insulator icing conditions and risks in images requires recognizing shed parameters. This paper proposes a method for analyzing field-monitoring icing insulator images to position, quantify, and automatically calculate graphical shed parameters. The method integrates image GrabCut, denoising, and enhancement to segment and transform insulators into measurable contour lines. By analyzing the contour peaks and valleys, the method identifies the ends, roots, and units of the sheds on both sides of the graphical axis. It then calculates graphical shed parameters, including shed overhang, shed spacing, shed contour length, and shed end-root distance. This method effectively processes typical insulator monitoring images, revealing that icing changes the insulator's contour peaks and valleys, shifting the graphical axis, shed ends, and shed roots. Graphical shed parameters change with insulator icing types and characterize the distribution uniformity. Heavy rime icing increases graphical shed contour length, reducing graphical shed end-root distance. The proposed parameters and method enable the quantification of the morphology, distribution and degree of various icing types in images, costing a minimum of 4.4 seconds for a single image, meeting engineering application requirements.