Hydrophobicity grade discrimination of high voltage composite insulators based on water trace image processing

Abstract

At present, when the hydrophobicity (HC) of high-voltage composite insulators is graded, there is a certain probability that the noise point information will be lost in the denoising process of the water trace image, which makes the image features nonlinear and diversified, affecting the autonomous learning of the algorithm and resulting in low accuracy of the HC discrimination results. Therefore, a method based on water trace image processing for HC grade discrimination of high-voltage composite insulators was designed. The water trace of the high-voltage composite insulator was converted into a gray image, which was enhanced after smoothing and binarization. The dynamic threshold was used to segment the target and background, extract the image edge feature information, and construct a Bayes network to determine the HC level of the high-voltage composite insulator. The results of the performance test show that the average accuracy of the designed discrimination method is 97.89% for different grades, which verifies the reliability of the method in practical applications. The conclusion of this study is helpful to replace insulators in time, improve the safety performance of overhead transmission lines, and prevent accidents.