Composite Insulator Level Of Hydrophobicity Identification Technology Based On Feature Quantities Of Water Droplets Image

With the developing of power grid construction in China,there are more and more composite insulators appling in the project of transmission line due to their small size,light weight,good stain resistance and so on.However,because of service time of composite insulators and environmental factors,the material of composite insulators will be aging and the performance of hydrophobicity will be decreasing.And the risk of pollution flashover on the the composite insulator will be raised.Therefore,it becomes extremely important to identify the hydrophobicity of composite insulators regularly to ensure the safety and stable operation of transmission lines.Up to now,in the practical engineering project for the identification of hydrophobicity of composite insulators,appling the digital image processing technology to identify the hydrophobicity of composite insulators becomes more and more popularly.However,there are some shortcomings in the selection of hydrophobic image detection area,the extraction of water feature quantity,and the identification of the hydrophobicity using the existing technology.Thus the performance of identification in the practical engineering project is not satisfied.In order to improve the accuracy of digital image recognition of composite insulator hydrophobicity,the following research work is carried out:1)For the problem of extracting the water droplets feature quantities in the hydrophobic image,an algorithm based on improved watershed is proposed.The process is as follow: Firstly,depend on shed positioning and intelligent cropping,the hydrophobic detection area is obtained.Then,the mathematical morphology is used to mark the inside and outside of the water droplets.At last,the watershed algorithm is employed to get feature quantities,such as the number of the water droplets,the area of the water droplets,the coverage of the water droplets,the maximum water droplets area,the maximum water droplets coverage,the maximum side of the water droplets,the shape factor of the water droplets and so on.The algorithm not only reduces the over-segmentation in the process of the watershed,but also avoids to determine the value of threshold in the traditional edge detection.Therefore,the accuracy of the feature quantity of the water droplet is improved.2)Focus on the identification of the hydrophobicity of composite insulators,a method based on the Back Propagation(BP)neural network is proposed.According to the difference on the shape of the water droplets between HC1?HC3 and HC4?HC6,a large number of hydrophobic image samples are trained by the BP neural network to establish a mathematical correlation between the different water feature quantities and hydrophobicity.Then,hydrophobicity of the composite insulator can be easy to be identified by the water feature quantities.3)To verify the above method,a set of composite insulator level of hydrophobicity identification system was designed in the lab.The system includes an image acquisition device and a level of hydrophobicity recognition software.The image acquisition device is composed of an electric spray-head,a camera,a lighting device,etc.And images of different hydrophobicity of composite insulators can be acquired in the lab.Based on a modular design method,the level of hydrophobicity identification software is designed to extract the feature quantities of the water droplet image and identifiy the level of hydrophobicity.According to the level of hydrophobicity identification in the lab and a 500kV DC transmission line project using unmanned aerial vehicle(UAV),the results show that the level of hydrophobicity identification technology of composite insulator based on the feature quantities of the water droplet image can be applied in the practical engineering project effectively.