Back to Search Start Over

Influence of Hysteresis Effect on Contrast of Welding Defects Profile in Magneto-Optical Image

Authors :
Xiangdong Gao
Deyong You
Yanxi Zhang
Nanfeng Zhang
Congyi Wang
Nvjie Ma
Source :
IEEE Sensors Journal. 20:15034-15042
Publication Year :
2020
Publisher :
Institute of Electrical and Electronics Engineers (IEEE), 2020.

Abstract

Under the excitation of alternating magnetic field, the disadvantage of Magneto-optical Imaging (MOI) Non-destructive Testing (NDT) is no obvious difference between the light and shade of micro defects (sub-surface cracks) image. It will lead to great difficulties in feature extraction and recognition of welding defects. In response to this problem, the magneto-optical image (induced magnetic field) characteristics of welding defects under an alternating magnetic field are examined in this study. The variation of induced magnetic fields of different welding defects (cracks, pits, unfused, etc.) is studied by simulation and MOI to optimize the nondestructive detection of MOI under alternating excitation. Both the simulation experiment and the MOI detection experiment are conducted to study the variation of the induced magnetic field under the excitation frequency of 50 Hz and sampling frequency of 75 frames/s. First, the variation law of the induced magnetic field for the welding defect is analyzed by simulation. Then, by studying the color moments of different defect magneto-optical image data sets, the periodic variation of the same defect magneto-optical image under alternating excitation is analyzed. To prove the correctness of the periodic law summarized by the color moment, the cosine similarity is used to test the periodic variation of the image. The results show that the contrast of welding defects on the magneto-optical image changes periodically. According to this rule, the image with the strongest contrast is selected in magneto-optical image data set for defect feature extraction and recognition to overcome its shortcomings.

Details

ISSN :
23799153 and 1530437X
Volume :
20
Database :
OpenAIRE
Journal :
IEEE Sensors Journal
Accession number :
edsair.doi...........a1d1f4b5eee55a4bf78d8b13c7111450
Full Text :
https://doi.org/10.1109/jsen.2020.3009478