To address the issue of low accuracy in diagnosing irregular defects under multisource noise, a multiscale convolutional attention neural network (CANN) magnetic flux leakage detection method was proposed. First, the improved graph convolutional neural network was proposed to extract the long-range correlation between pixel-level and structural-level features of the leakage magnetic signal, and noise features were introduced to reconstruct the leakage magnetic signal to reduce the complex fluctuations caused by multi-source noise. Then, the decoupling fully connected attention mechanism was introduced into the traditional convolutional neural network (CNN) to aggregate the central features of the reconstructed signal and further extract the tiny features of irregular defects. Finally, the mean square error was used to determine the threshold value to detect irregular defects on the pipe surface. The results show that the average detection accuracy of the proposed method for irregular defects reaches 96.7%, which is 5.5% higher than the best detection accuracy of the existing methods.
WANG Ze, YIN Wei-wei, LANG Xian-ming, GUO Xiao-ting, CAO Jiang-tao, LIU Qiang, MIAO Xin.
Irregular Defect Magnetic Flux Leakage Detection Method Based on Improved Convolutional Neural Network[J].
Science Technology and Engineering, 2026, 26(12): 5144-5152 DOI:10.12404/j.issn.1671-1815.2504293
随着重叠IoU的增加,浅层特征受到噪声的干扰越来越大。所有方法的精准率、召回率和准确率分别随着IoU的增加而降低,在实际工业应用中,IoU为0.75时的检测结果更符合检测标准。引入了一个每秒浮点运算(floating-point operations per second, FLOPS)来间接评估每个网络模型的复杂性点。因此,设置在IoU为0.75情况下,如表1所示为不同方法缺陷检测的对比结果。
GuoYing, YangLijian, ZhaoBaishun, et al. Research status of leakage detection technology for long-distance oil pipelines[J]. Journal of Liaoning University of Petroleum & Chemical Technology, 2022, 42(4): 25-31.
WeiHaotian, DongShaohua, LiYong, et al. Residual magnetic signal detection method of steel pipeline crack based on spatial resolution[J]. Science Technology and Engineering, 2024, 24(28): 12135-12142.
[6]
LangX M, YuanL, LiS Y, et al. Pipeline multipoint leakage detection method based on KKL-MSCNN[J]. IEEE Sensors Journal, 2024, 24(7): 11438-11449.
WangKun, HanWenhua, WangHaihang. Recon-struction of three-dimensional irregular defects based on improved trust region algorithm[J]. Chinese Journal of Science Instrument, 2021, 42(10): 129-137.
ChenLiqiong, MeiHoujin, HuHongxuan. Weld defect detection based on improved Faster R-CNN method[J]. Science Technology and Engineering, 2025, 25(5): 2027-2033.
[11]
ZhouC, WeiC Z, WangW. A new detection method based on magnetic leakage theory and BP neural network for broken steel strands in ACSR conductor[J]. IEEE Sensors Journal, 2022, 22(20): 19620-19634.
[12]
FengJ, LiF M, LuS X, et al. Injurious or noninjurious defect identification from MFL images in pipeline inspection using convolutional neural network[J]. IEEE Transactions on Instrumentation and Measurement, 2017, 66(7): 1883-1892.
YaoRui, LangXianming, YuanKaixin, et al. Detection model of oil pipeline tiny defects based on improved swin transformer[J]. Pipeline Technique and Equipment, 2024(3): 35-42, 47.
[15]
YuG, LiuJ H, ZhangH G, et al. An iterative stacking method for pipeline defect inversion with complex MFL signals[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(6): 3780-3788.
RuLishuang, HanWenhua. Fast method for calculating the magnetic charge distribution of 3D irregular defects[J]. Science Technology and Engineering, 2025, 25(1): 194-200.
ShenXiangkai, LiuJinhai. Defect and component detection method of oil pipeline based on heterogeneous diffusion model[J]. Control and Decision, 2025, 40(3): 937-945.
[20]
LangX M, HanF C. MFL image recognition method of pipeline corrosion defects based on multilayer feature fusion multiscale ghost-net[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-8.
HuangXin, QuWenzhong, XiaoLi. Acoustic emission recognition method for valve internal leakage based on convolutional attention mechanism[J]. Journal of vibration and shock, 2024, 43(9): 105-114.
[23]
ShenX K, LiuJ H, SunJ Y, et al. SSCT-Net: a semisupervised circular teacher network for defect detection with limited labeled multiview MFL samples[J]. IEEE Transactions on Industrial Informatics, 2023, 19(10): 10114-10124.
LiXiang, ZhaoWeirui. Research on the method for simultaneously detecting piston and tip-tilt errors of segmented telescopes based on multiple CNNs[J]. Chinese Journal of Science Instrument, 2024, 45(3): 188-197.
[26]
JiangL, ZhangH G, LiuJ H, et al. A multisensor cycle-supervised convolutional neural networks for anomaly detection on magnetic flux leakage signals[J]. IEEE Transactions on Industrial Informatics, 2022, 18(11): 7619-7627.
[27]
HanF C, LangX M, LiuM Y. An anchor-free pipeline MFL image detection method[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 1-8.
[28]
JiangL, ZhangH G, LiuJ H, et al. THMS-Net: A two-stage heterogeneous signals mutual supervision network for MFL weak defect detection[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-9.
[29]
MienyeI D, SunY, WangZ. Improved sparse auto-encoder based artificial neural network approach for prediction of heart disease[J]. Informatics in Medicine Unlocked, 2020, 18: 100307.
ZhaoZongchao, LiDongxing, ZhaoMengna. A tracking algorithm for dim target based on guided image filter and deep denoising autoencoder[J]. Science Technology and Engineering, 2020, 20(14): 5696-5701.
PDF
(15710KB)
