Research Of Fault Diagnosis System Of Transmission Line Based On The Image Processing

Computer vision detection techniques are getting more and more widely used in intelligent inspection and online monitoring in the power equipment with the development of technologies in smart grid and power system automation.The key components in transmission lines(insulators,transmission towers,etc.)are frequent failure components.Once a failure occurs,it will seriously threaten the safe and reliable operation of transmission lines.Therefore,regular supervision and inspection are essential.Fault analysis of aerial images of transmission lines has become a research hotspot in recent years.This paper mainly focuses on the detection of power transmission elements and faults in aerial images,and studies the detection of faulty targets such as insulator dropouts and transmission tower bird nests.The main contents of this paper are as follows:1)The aerial image has a complex and diverse background.Based on the analysis of the transmission line image characteristics,an improved Mask R-CNN object segmentation algorithm is proposed to initially segment the image objects.Then,the characteristics of the insulator and the contour characteristics of the insulator after the string is dropped.A contour extraction method based on immune genetic snake was proposed to extract the insulator contour,and the result was subjected to a least squares ellipse fitting.The position of the insulator was dropped by calculating the distance between the center points of the insulator slices after the fitting,there by realizing the insulator fault detection.Experimental results show that the method has high accuracy and strong anti-interference ability,and can meet the requirements of practical applications.2)Due to the high similarity of the background features of the transmission tower and the trees and towers in the background,it is easy to cause false detection.Therefore,a fusion corner of the candidate target image segmented based on the improved Mask R-CNN algorithm is proposed the transmission tower identification method based on the linear and linear features realizes the accurate positioning of the transmission tower,and then the segmented image of the transmission tower is input to the target detection system for bird nest feature recognition training,and finally the bird nest fault detection on the transmission tower is realized.The influence of each part of the detection system on the measurement accuracy is analyzed,and corresponding solutions are given.Experimental results verify the accuracy and effectiveness of this method.3)Design and implementation of transmission line fault diagnosis system based on image processing.Based on the requirements analysis and function design of the functions to be implemented in the transmission line fault diagnosis system,this paper mainly introduces four major modules: system login module,picture acquisition and target classification module,target detection model training module,and fault diagnosis module.System users can the type of transmission element is appropriately selected by the fault diagnosis algorithm,the appropriate parameters are configured,the selected data is processed,and the processed results are displayed in a visual form.The Py Charm,Tensor Flow and Python are used to development environment,framework and programming language respectively.The results of experimental comparative analysis show that the fault diagnosis system has better accuracy,reliability and shorter time.In addition,the algorithm has an accuracy up to 94% in the fault detection of insulator dropping,and 95% in the fault detection of bird nest in the transmission tower.