水下
声学
电流(流体)
磁场
原位
材料科学
电气工程
工程类
电子工程
计算机科学
地质学
物理
气象学
海洋学
量子力学
作者
Xin’an Yuan,H J Wang,Wei Li,Xiaokang Yin,Xihe Zhang,Jianxi Ding,Jianchao Zhao,Qinyu Chen,Jun Wu,Xiao Li,Dong Hu,Yu Gao
标识
DOI:10.1109/tim.2024.3383056
摘要
Alternating current field measurement (ACFM) technique has been widely used in defect inspection of underwater metal structures. However, due to the high cost of inspection, it is difficult to carry out nondestructive testing (NDT) regularly to obtain the health status of key nodes of structures. And the quantification of defects depends on simple empirical formula, which leads to large quantization error. In this paper, a flexible ACFM magnetic sensor array for in-situ inspection of underwater structural cracks is proposed. A in-situ inspection probe based on the sensor array is manufactured and can be fixed on the surface of underwater curved structures. The probe can realize visual imaging of cracks in underwater carbon steel structure to obtain the damage status of key nodes of the structure regularly. According to the inspection image, a crack quantification method based on multi-physical features fusion convolution neural network, which is used for quantification of crack length, depth and angle. The experimental results show that the in-situ inspection probe can realize the visual imaging of cracks. The quantification method can accurately evaluate the crack dimensions. The average length error, average depth error and average angle error are 0.41 mm, 0.169 mm and 1.131° respectively.
科研通智能强力驱动
Strongly Powered by AbleSci AI