Open AccessDefect detection for aluminium conductor composite core X-ray image with deep convolution network
Author(s) -
Wei Rui,
Hanlai Wei,
Chen Da-bing,
Lizhe Xie,
Zheng Wang,
Yining Hu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1633/1/012166
Subject(s) - ampacity , conductor , convolution (computer science) , convolutional neural network , core (optical fiber) , transmission line , electrical conductor , computer science , composite number , electric power transmission , line (geometry) , image (mathematics) , aluminium , distortion (music) , materials science , artificial intelligence , structural engineering , electrical engineering , artificial neural network , engineering , composite material , telecommunications , algorithm , mathematics , amplifier , geometry , bandwidth (computing)
The Aluminum Conductor Composite Core (ACCC) has been considered one of the solutions for massively increasing requirements for the electricity power transmission in China due to its superiority in weight, strength and ampacity. Yet the popularize of ACCC lines suffer from damages caused during the construction, which may result in line broke in the future. In this paper, an automatic defect detection method based on Deep Convolution Network is proposed. Image classification framework with Inception-Resnet structure as backbone is applied. With the online self-designed robot, the proposed method can effectively detect the defects such as fracture, splitting and distortion, with a recall rate over 90%.