Research On Cable Surface Defect Detection Method Based On Image Processing Technology

China ’s power industry has entered a stage of rapid development.China ’s power grid construction and cable manufacturing industry has surpassed most European and American mainland countries and has gradually become the world ’s largest power cable manufacturing country.However,the quality and standards of power cable production and on-site installation are not up to standard.The manual visual inspection form of electric power workers is gradually unable to meet the various problems in the cable production and installation process,and the cable inspection results based on manual visual inspection are not reliable.Image processing technology is gradually applied to various industrial fields,especially in the field of power cable outer protective layer detection.In order to realize the application of image processing technology to the detection of cable surface defects,I have done a lot of research.The main work is as follows:(1)Based on the development status of cable detection at home and abroad and the actual needs of people,the overall scheme design of cable surface detection based on image processing technology is completed.Combined with the characteristics of the scheme and the actual situation,a simple image acquisition platform based on UAV is selected to complete the collection and production of cable image data sets.(2)According to the actual cable erection environment,and the UAV cable image acquisition process is often affected by the environment,or the UAV ’s own image acquisition components are overheated,there will be noise-containing cable images.Through the targeted analysis of the characteristics of the cable,the image of the cable surface is preprocessed by image graying,image denoising and edge detection.By comparing and analyzing the algorithms in image preprocessing,the experimental results show that the improved algorithm has the advantages of reliability,high efficiency and certain feasibility.(3)Starting from the actual cable characteristics,and the cable information extraction.The SVM and BP classification algorithms are studied,and the improved GMO-BP and GMO-SVM multi-classification algorithms are proposed.At the same time,four classification models are used to test the classification accuracy of three kinds of cable defects.The average classification accuracy of SVM classification algorithm is 89.7%,and the average classification accuracy of BP classification algorithm is86.3%.Because the classification accuracy of SVM and BP classification algorithm is low,the grey wolf optimization algorithm is used to improve SVM and BP.The average classification accuracy of GMO-SVM multi-classification algorithm is 93.1%,and the average classification accuracy of improved GMO-BP classification algorithm is 97.3%.This paper only considers the recognition accuracy of cable defects.Therefore,the GMO-BP classification algorithm can better meet the fine detection requirements of power cable surface defects during power inspection.(4)In order to facilitate the maintenance personnel to detect the cable,the software of the cable detection host computer is completed,and the human-computer interaction interface based on MATLAB is designed.The system scheme proposed in this paper satisfies the expected goal as a whole and has certain engineering practicability.