None
Authors: Xin’an Yuan (China University of Petroleum (East China)) , Wei Li (China University of Petroleum (East China)) , Xiaokang Yin (China University of Petroleum (East China)) , Jianming Zhao (China University of Petroleum (East China)) , Guoming Chen (China University of Petroleum (East China))
The traditional alternating current field measurement (ACFM) technique identifies the defect by the butterfly plot, which is easily disturbed by the lift-off variations of the probe. This paper presents a novel intelligent recognition method for the inspection of defects using the ACFM technique. Firstly, the magnetic field in Z direction (Bz) is discovered as the insensitive signal to the lift-off. Secondly, the image gradient field inversion algorithm is presented to reconstruct the surface profile of the defect. Thirdly, the convolutional neural networks (CNN) deep learning algorithm is proposed to achieve intelligent recognition and classification of defects. The results show that the crack, irregular crack and corrosion defect can be recognized and classified accurately.
Keywords:
How to Cite: Yuan, X. . , Li, W. . , Yin, X. . , Zhao, J. . & Chen, G. . (2019) “Visual and intelligent recognition of defects in underwater structures using ACFM technique”, Review of Progress in Quantitative Nondestructive Evaluation.(0).