Research On The Discrimination Method For The Hydrophobicity Of Composite Insulators Based On Image Processing

With the rapid development of UHV transmission line construction,the operational status requirement for composite insulators is also increasing,the running composite insulators are susceptible to some conditions such as corona discharge,ultraviolet radiation,pollution,rain and snow,the insulator skirt material will be gradually aging,which decreases its hydrophobic performance and increases the probability of pollution flashover.In order to improve the safety and stability of power system,it is very important to detect the hydrophobicity of composite insulators in real time and find out the aging phenomenon in time.The research center in this paper is the hydrophobic state of composite insul ator,In view of the particularity of the environment of composite insulator hydrophobic image,the grayscale distribution of droplet image of composite insulator surface is improved through multi-scale Retinex and wavelet de-noising.The results show that the distribution of the gray areas of the hydrophobic image is significantly improved,and the details of the water bead are more visible,and the interference of the background noise to the image is effectively suppressed,and the shape of the target water is more prominent.In view of the defect of traditional CV model and the defect of the traditional Canny algorithm,the two segmentation algorithms are improved and applied to the hydrophobic image segmentation of composite insulators,and then the morp hological operation is applied to fill the correction.The segmentation results show that the two segmentation algorithms are able to complete the segmentation of hydrophobic images well.After improving the hydrophobic image after the modified CV model,the contour of the water bead is more clear,and the contour of the white water bead can be corrected by the morphological opening and closing operation,and the segmentation effect is more obvious.After the improved Canny algorithm,the hydrophobic image was processed,and the closed contour of the water bead was greatly increased,and the broken line produced during the segmentation was greatly reduced,and the segmentation results could meet the subsequent grading requirements.In the end,we need extract three characteristic parameters in the separation of the composite insulators,and then input the data from the profile of the characteristic parameters into the probability neural network that has been trained to do the hydrophobic level recognition,and analyze the effect of different expansion speed on the desired learning results simultaneously.The classification results show that the neural network can realize the accurate determination of the hydrophobic grade of the composite insulator,and the total rate of recognition up to ninety-five percent by setting the right rate of expansion,which meets the needs of the practical application.