Defect diagnosis method for oil and gas pipelines based on fusion of Welch power spectrum and multi-scale residual network

doi:10.16265/j.cnki.issn1003-3033.2026.02.0599

Abstract

Abstract:

To address the problem of reduced defect classification accuracy caused by noise contamination in the bend detection signals of oil and gas pipelines, this paper proposes an oil and gas elbow defect diagnosis model based on Welch power spectrum feature enhancement and multi-head attention improved dual-branch multi-scale-residual collaborative network. Firstly, the Welch method was used to convert the collected time domain signal into a feature-enhanced power spectrum, showing the energy distribution of the defect signals at different frequencies. Secondly, the multi-scale network branch composed of parallel stacked convolutional layers was responsible for extracting the multi-dimensional features of the signal power spectrum, and the multi-head attention mechanism was used to establish long-term associations between features. Simultaneously, the residual network branch captured the detail information of the signal power spectrum. Finally, the deep concatenation layer fused the features extracted by the dual-branch network to achieve defect classification. The experiment results show that in a high-noise environment, the test accuracy of the proposed model is 91.6%. Compared with the models based on Kaiser windows and flat-top windows, the classification accuracy is improved by 1%~7.9%; compared with convolutional neural network (CNN) and long short-term memory network (LSTM), the accuracy is improved by 36.9% and 10.3% respectively.

Key words: Welch power spectrum, multi-scale residual network, oil and gas pipeline, elbow defect, high noise, intelligent diagnosis

CLC Number:

X937

LIAO Chunyan, LIANG Wei, LIU Shuanglei, HUANG Tianchang. Defect diagnosis method for oil and gas pipelines based on fusion of Welch power spectrum and multi-scale residual network[J]. China Safety Science Journal, 2026, 36(2): 136-144.

Figures/Tables 14

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Table 1

Fig.7

Fig.8

Fig.9

Fig.10

Table 2

Evaluation results of diagnosis models based on different windows on test set %

窗函数	评价指标
窗函数	精确率P	召回率R	$F 1$	准确率A
汉明窗	90.9	91.0	90.9	91.6
凯塞窗 $β = 0$	89.3	89.0	89.2	90.2
凯塞窗 $β = 5$	90.2	89.0	89.4	90.6
凯塞窗 $β = 8.6$	83.8	83.3	83.5	85.2
平顶窗	82.1	81.8	81.7	83.7

Table 2

Fig.11

Table 3

Evaluation results of different models on test set %

模型	评价指标
模型	精确率P	召回率R	$F 1$	准确率A
本文模型	90.9	91.0	90.9	91.6
CNN	45.8	51.8	47.9	54.7
LSTM	78.8	76.5	74.9	81.3

Table 3

References 19

[1]	姜垣良, 任庆滢, 任远, 等. 基于改进YOLO模型的中缅油气管道遥感图像高后果区识别方法[J]. 中国安全科学学报, 2025, 35(1):103-111. doi: 10.16265/j.cnki.issn1003-3033.2025.01.0632
	JIANG Yuanliang, REN Qingying, REN Yuan, et al. High-consequence area identification of remote sensing images of China-Myanmar oil and gas pipeline based on improved YOLO model[J]. China Safety Science Journal, 2025, 35(1): 103-111. doi: 10.16265/j.cnki.issn1003-3033.2025.01.0632
[2]	宫运华, 张喆. 油气管道事故应急失效原因及社会网络分析[J]. 中国安全科学学报, 2025, 35(1):209-215. doi: 10.16265/j.cnki.issn1003-3033.2025.01.1536
	GONG Yunhua, ZHANG Zhe. Failure causes of emergency response in oil and gas pipeline accidents and social network analysis[J]. China Safety Science Journal, 2025, 35(1): 209-215. doi: 10.16265/j.cnki.issn1003-3033.2025.01.1536
[3]	HUANG Songling, PENG Lisha, SUN Hongyu, et al. Deep learning for magnetic flux leakage detection and evaluation of oil & gas pipelines: a review[J]. Energies, 2023, 16(3): DOI:10.3390/en16031372.
[4]	王轲, 黄成橙, 王鹏. 深度学习在油气管道漏磁检测领域的应用[J]. 无损检测, 2024, 46(12):93-99.
	WANG Ke, HUANG Chengcheng, WANG Peng. Application of deep learning in magnetic leakage detection of oil and gas pipelines[J]. Nondestructive Testing, 2024, 46(12):93-99.
[5]	WU Linyu, LIANG Wei, SHA Duolin. A novel adaptive noise reduction method for field natural gas pipeline defect detection signals[J]. International Journal of Pressure Vessels and Piping, 2022, 199: DOI: 10.1016/j.ijpvp.2022.104761.
[6]	武琳玉, 梁伟, 沙夺林. 油气站场典型设备的焊缝缺陷检测与识别方法[J]. 中国安全科学学报, 2020, 30(11):108-113. doi: 10.16265/j.cnki.issn 1003-3033.2020.11.016
	WU Linyu, LIANG Wei, SHA Duolin. Weld defect detection and identification method of typical equipment in oil-gas station[J]. China Safety Science Journal, 2020, 30(11): 108-113. doi: 10.16265/j.cnki.issn 1003-3033.2020.11.016
[7]	SHA Duolin, LIANG Wei, WU Linyu. A novel noise reduction method for natural gas pipeline defect detection signals[J]. Journal of Natural Gas Science and Engineering, 2021, 96: DOI:10.1016/j.jngse.2021.104335.
[8]	WANG Shuaishuai, LIANG Wei, SHI Fang. Identification of coating layer pipeline defects based on the GA-SENet-ResNet18 model[J]. International Journal of Pressure Vessels and Piping, 2024, 212: DOI:10.1016/j.ijpvp.2024.105327.
[9]	李友家, 张忠伟, 焦宗豪, 等. 强噪声干扰下基于SVMD-FFCNN的深沟球轴承故障分类模型[J]. 机电工程, 2025, 42(4):686-696.
	LI Youjia, ZHANG Zhongwei, JIAO Zonghao, et al. Fault classification model of deep groove ball bearing under strong noise interference based on SVMD-FFCNN[J]. Journal of Mechanical & Electrical Engineering, 2025, 42(4):686-696.
[10]	KILADZE M. Application of bidirectional LSTM neural networks and noise pre-cleaning for feature extraction on an electrocardiogram[C]. International Conference on Actual Problems of Applied Mathematics and Computer Science (APAMC), 2022:421-429.
[11]	KAY K. The risk of bias in denoising methods: examples from neuroimaging[J]. PloS One, 2022, 17(7): DOI:10.1371/journal.pone.0270895.
[12]	PICHIKA S V V S N, KASAM V, RAJASEKHARAN S G, et al. Automatic signal denoising and multi-component fault classification based on deep learning using integrated condition monitoring in a wind turbine gearbox[J]. Journal of Vibration Engineering & Technologies, 2024, 12(7): DOI:10.1007/s42417-024-01380-6.
[13]	邢晓晴, 朱根民. Welch功率谱估计中窗函数的选择与算法分析[J]. 计算机时代, 2018(2):1-4.
	XING Xiaoqing, ZHU Genmin. Window function selection and algorithm analysis in Welch power spectrum estimation[J]. Computer Era, 2018(2):1-4.
[14]	杨婧, 程乃平, 倪淑燕. Welch算法在弱信号检测中的性能分析[J]. 计算机仿真, 2020, 37(5):235-240.
	YANG Jing, CHENG Naiping, NI Shuyan. Performance analysis of Welch algorithm in weak signal detection[J]. Computer Simulation, 2020, 37(5):235-240.
[15]	OPPENHEIM A V, SCHAFER R W, BUCK J R. Discrete-time signal processing[M]. Upper Saddle River: Prentice Hall, 1999: 475-487.
[16]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]. 31^st Annual Conference on Neural Information Processing Systems (NIPS), 2017: 5998-6008.
[17]	SZEGEDY C, LIU Wei, JIA Yangqing, et al. Going deeper with convolutions[J]. CoRR, 2014: DOI:10.1109/cvpr.2015.7298594.
[18]	李强, 马超, 黄民. 基于注意力的多尺度残差卷积网络轴承故障诊断[J]. 电子测量技术, 2025, 48(9) :19-26.
	LI Qiang, MA Chao, HUANG Min. Attention-based multi-scale residual convolutional network for bearing fault diagnosis[J]. Electronic Measurement Technology, 2025, 48(9) :19-26.
[19]	LIAO Chunyan, LIANG Wei, YAO Junming, et al. Intelligent defect diagnosis for oil and gas pipeline based on lightweight CBAM-Inception-Resnet[J]. Chemical Engineering Research and Design, 2025, 218: 548-571. doi: 10.1016/j.cherd.2025.05.030

缺陷类别	样本个数	各数据集代表信号的信噪比/dB
缺陷类别	样本个数	数据集1	数据集2	数据集3	数据集4
腐蚀	448	2.94	2.00	1.06	-0.42
划痕	369	13.78	5.63	0.29	-6.58
焊缝开裂	572	18.09	6.58	1.71	-1.04
正常	645	18.21	8.82	0.66	-4.11