Image Recognition and Monitoring of Insulator Icing Based on NU-SVC

Xingbo Han,Peng Sun,Jie Wang,Bin Xing,Xingliang Jiang
DOI: https://doi.org/10.1007/978-981-99-0451-8_22
2023-01-01
Abstract:Insulator icing identification and warning are basic scientific problems in line icing research, but the existing research results fail to fully applied in line icing monitoring and condition assessment. In this paper, based on multi-scene and multi-type insulator icing images, the methods of defogging the icing image, identifying the icing type and classifying the icing risk level were studied. By improving the dark channel dehazing algorithm, the image dehazing restoration is realized. Image segmentation was used to analyze the characteristics of various types of insulator ice-covered morphology, and the image texture features of insulator surface ice-covered structure under different ice-covered types were summarized. Based on the features of grayscale symbiotic matrices (GLCM), the classification of image texture and the recognition of insulator ice-covered type were realized. A discriminant model of icing severity was established by identifying the proportion of insulator volume. Finally, the NU-SVC classification and recognition algorithm is adopted to establish an early warning model of insulator icing risk level based on classification credibility and recognition credibility. The results show that: The algorithm in this paper ensures that the image is not distorted under the condition of rapid defogging, which can not only realize image enhancement under dim background, but also realize defogging processing under fog and rain. Based on the recognition model of GLCM features and insulator icing volume, the risk level of insulator icing can be accurately identified. The classification and recognition model based on NU-SVC can abstract a small number of feature parameters from a large number of insulator image samples to predict the icing category of unknown samples. The recognition accuracy of the model is 92%, and the evaluation mechanism is reliable, which can be used as the intelligent monitoring of the icing image of insulators.
What problem does this paper attempt to address?