Abstract
Due to the excellent insulation performance of composite insulators, they are widely used in transmission and distribution lines, which also brings problems such as a large workload of composite insulator fault detection and an urgent need to improve the degree of intelligence. In this paper, a method for identifying abnormal temperature rise defects of composite insulators based on infrared image threshold segmentation is proposed. First, train and use the YOLOv4 convolutional neural network to detect the composite insulator in the complex infrared image, and initially locate the area where the composite insulator is located; then use the affine transformation correction algorithm to adjust the insulator mandrel to the horizontal direction to determine the temperature rise position of the defect; Then use TRIANGLE method to perform image threshold segmentation on the composite insulator area to obtain a binary image of the insulator, according to which the insulator in the infrared image is segmented; an infrared image contrast enhancement algorithm based on fuzzy algorithm is proposed, and the temperature rise area of the insulator is obtained by threshold segmentation; According to the infrared characteristics of temperature rise of the defects of composite insulators, the types of defects and the degree of failure of composite insulators can be judged. Finally, the identification and diagnosis of the abnormal temperature rise of the composite insulator is realized based on the visual information and the temperature rise information of the abnormal temperature rise area.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have