Research On Identification Method Of Transmission Line Icing Based On Binocular Vision

Transmission lines are an important part of power transmission in power systems.Ice covering can cause accidents such as conductor galloping and tower collapse,which endanger the safe operation of power systems,and have a negative impact on the production work of various industries.Therefore,timely detection of line icing status and icing thickness measurement have become an important means to ensure the safe operation of transmission lines.Traditional transmission line icing detection methods are mostly manual observation,simulated wire method or weighing method to realize icing state detection.In view of the problems of high labor intensity,low accuracy of results,and inability to obtain icing data in time,the traditional methods exist.A method based on binocular vision is proposed to identify the ice-covered state of transmission lines,the ice-covered images of transmission lines are collected online through a binocular camera installed on high-voltage towers.The image processing technology is used to calculate the ice-covered thickness of transmission lines,and the threedimensional coordinate data of the ice-covered area,this method has the advantages of high precision,low cost,and conforms to the development direction of smart grid.The main contents include:(1)The principle and model of binocular stereo vision are analyzed.The parallel binocular vision model is used to study the identification method of ice-covered transmission lines.The key steps of the research include binocular vision calibration,stereo matching of ice-covered images,measurement of ice-covered thickness,and acquisition of three-dimensional coordinates of ice-covered areas.(2)The binocular vision calibration method is studied.Aiming at the problems that the traditional Zhang’s calibration method is sensitive to the initial value of camera model parameters and the calibration results are unstable,a calibration parameter optimization method based on sine and cosine algorithm is proposed.The sine and cosine algorithm is used to form the initial population of the initial value field obtained by Zhang’s calibration method,and iterative optimization is carried out.The calibration experiment results show that,the average reprojection error of the binocular vision calibration method proposed is only 0.0462 pixels.Based on the original calibration method,more stable and accurate calibration results can be obtained.(3)Research on the stereo matching algorithm of icing images.Aiming at the problem that the center point of the traditional Census transform window is easily disturbed by external noise and prone to mis-matching,a matching algorithm based on multi-feature fusion cost calculation and multi-step disparity optimization is proposed.The three features of Census transformation,color information and gradient information are combined,and the similarity measure function for matching cost calculation is jointly constructed.The multi-step optimization strategy is used to optimize the disparity map of the ice-covered image,passing the standard test of the Middlebury test platform Compared with other algorithms,the matching algorithm proposed has higher matching accuracy.(4)Through the built binocular stereo vision system and the ice-covered transmission line model,the ice-covered thickness measurement of the transmission line and the acquisition of the three-dimensional coordinates of the ice-covered area are realized.The experimental results show that the relative errors of the measured icing thickness and the three-dimensional coordinates of the icing area are within 5% and 2%,respectively,which verifies the accuracy of the ice-covered identification method proposed.Also develops the interface of the transmission line icing monitoring system,which has the functions of icing detection,icing thickness measurement,icing threedimensional coordinate data acquisition and storage,etc.,which provides an effective solution for transmission line icing detection.