Font Size: a A A

Defect Detection Methods For Transmission Line Inspection

Posted on:2019-08-02Degree:MasterType:Thesis
Country:ChinaCandidate:N X QiFull Text:PDF
GTID:2382330548470490Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Power system is the lifeblood of the country,while the transmission line is an important part of the power system.Transmission line inspection is very important for ensuring the safe operation of the power grid.With the development of science and technology,intelligent inspection technology has gradually replaced the traditional manual inspection and the use of helicopters and UAV for inspection has become an important method for inspection of transmission line.In this paper,the method of detection and determination of transmission line defects are studied based on the images or video collected by helicopters or UAV.The main work is as follows:To detect the defects of the transmission line,it should position the object firstly.By analyzing the performance and application of multiple target detection algorithms,the defect detection method is designed based on the YOL02,which can detect and locate all defects in one inference.After the adjustment of the network structure and the optimization of parameters,the algorithm is experimented by using the transmission line defect dataset.The results show that this method can identify and locate insulators and bird nests,and the bird damage can be detected by locating the nest.Based on the detection and localization of insulators,this paper studies the method of determining the string breakage defect.By analyzing the characteristics of the insulator images,this paper proposes a method for determining the defect based on pattern feature matching.After correcting the direction of the insulator string,the method can determination the string breakage defect by extracting the HOG feature of insulators and matching one by one.The experiments on insulator images have verified the effectiveness of this method.
Keywords/Search Tags:Line Inspection, Deep Learning, CNN
PDF Full Text Request
Related items