|本期目录/Table of Contents|

[1]张昊宁,段庆全,付立武,等.基于惯导内检测数据的油气管道凹陷尺寸量化研究*[J].中国安全生产科学技术,2025,21(1):132-138.[doi:10.11731/j.issn.1673-193x.2025.01.017]
　ZHANG Haoning,DUAN Qingquan,FU Liwu,et al.Research on quantification of dent size in oil and gas pipelines based on inertial navigation internal detection data[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2025,21(1):132-138.[doi:10.11731/j.issn.1673-193x.2025.01.017]
点击复制

基于惯导内检测数据的油气管道凹陷尺寸量化研究*
分享到：

《中国安全生产科学技术》[ISSN:1673-193X/CN:11-5335/TB]

卷:
21
期数:
2025年1期
页码:
132-138
栏目:
职业安全卫生管理与技术
出版日期:
2025-01-30

文章信息/Info

Title:
Research on quantification of dent size in oil and gas pipelines based on inertial navigation internal detection data
文章编号:
1673-193X(2025)-01-0132-07
作者:
张昊宁; 段庆全; 付立武; 石彤; 谢婷; 李睿; 富宽; 王昊; 刘啸奔
（1.中国石油大学（北京） 安全与海洋工程学院,北京 102249;
2.油气生产安全与应急技术应急管理部重点实验室，北京 102249；
3.国家管网集团储运技术发展有限公司，天津 304457；
4.中国石油大学（北京） 油气运输安全国家工程研究中心，北京 102249；
5.国家管网集团科学技术研究总院分公司，天津 304457）
Author(s):
ZHANG Haoning;  DUAN Qingquan;  FU Liwu;  SHI Tong;  XIE Ting;  LI Rui;  FU Kuan;  WANG Hao;  LIU Xiaoben
(1.College of Safety and Ocean Engineering,China University of Petroleum,Beijing；
2.Key Laboratory of Oil and Gas Safety and Emergeney Technology,Ministry of Emergeney Management；
3.PipeChina Storage and Transportation Technology Development Co.,Ltd.;
4.National Engineering Research Center for Oil and Gas Pipeline Transportation Safety,China University of Petroleum (Beijing);
5.PipeChina Science and Technology Research Institute Branch)
关键词:
IMU; CNN-LSTM-Attention神经网络; 凹陷量化
Keywords:
inertial measurement unit (IMU);  CNN-LSTM-Attention neural network;  dent quantifi-cation
分类号:
X937；TE88
DOI:
10.11731/j.issn.1673-193x.2025.01.017
文献标志码:
A
摘要:
为研究基于惯性导航检测数据的长距离油气管道凹陷尺寸的智能识别和定量分析，提出1种基于CNN-LSTM-Attention混合神经网络的管道凹陷定量识别方法。首先针对惯性测量单元（IMU）检测数据进行预处理，利用主成分分析(PCA)算法进行数据降维，建立凹陷样本数据库，构建CNN-LSTM-Attention神经网络的深度学习模型，实现凹陷的量化识别。研究结果表明:本文提出的深度学习模型学习率为0.001时收敛较快，准确率高达92.4%，皆优于同类对比模型，并与2010—2018年数据进行对比分析，其对于凹陷长度以及宽度的误差均不超过真实值的10%，预测精度较高。研究结果可为管道安全运行评价提供理论支撑和技术支持。
Abstract:
To study the intelligent recognition and quantitative analysis of depression size in the long-distance oil and gas pipelines based on the inertial navigation detection data,a pipeline depression quantitative recognition method based on CNN-LSTM Attention hybrid neural network was proposed.The preprocessing was performed on the detection data of inertial measurement unit (IMU),and the principal component analysis (PCA) algorithm was used for data dimensionality reduction.A deep learning model of CNN-LSTM-Attention neural network was constructed to achieve the quantitative recognition of dents.The results show that the deep learning model proposed in this paper converges quickly when the learning rate is 0.001,with an accuracy of up to 92.4%,which are superior to similar comparative models.Through the comparative analysis on the data from 2010 to 2018,both the errors in the dent length and width are not exceed 10% of the true value,indicating a high prediction accuracy.The research results can provide theoretical and technical support for the evaluation of pipeline safety operation.

参考文献/References:

［1］曾绪财，张葛祥，任涛，等.管道内检测研究进展及展望［J］.交通信息与安全，2019，37(6):20-31.ZENG Xucai，ZHANG Gexiang.REN Tao，et al.Research progresses and prospects of pipeline internal inspection ［J］.Journal of Transport Information and Safety，2019，37 (6):20-31
［2］程诚.油气长输管道安全经济效益评价模型及应用［J］.中国安全生产科学技术，2021，17(增刊1):56-61.CHENG Cheng.Safety and economic benefit evaluation model of long-distance oil and gas pipeline and its application ［J］.Journal of Safety Science and Technology，2021，17 (Supplement 1):56-61.
［3］石彤,刘啸奔,张琳,等.管道IMU弯曲应变解算算法优化与全尺寸试验验证［J］.油气储运,2024,43(11):1269-1276.SHI Tong,LIU Xiaoben,ZHANG Lin,et al.Optimization of IMU-based bending strain solving algorithm and full-scale experimental validation［J］.Oil & Gas Storage and Transportation,2024,43(11):1269-1276
［4］DESJARDINS G,NICKLE R,SKIBINSKY D,et al.Comparison of in-line inspection service provider Magnetic Flux Leakage (MFL) technology and analytical performance based on multiple runs on pipeline segments［C］//International Pipeline Conference，2012.
［5］HUANG K.3-D defect profile reconstruction from magnetic flux leakage signatures using wavelet basis function neural networks［D］.Ames:Iowa State University，2000.
［6］赵晓明，李睿，陈朋超，等.中俄东线天然气管道弯曲变形识别与评价［J］.油气储运，2020，39(7):763-768.ZHAO Xiaoming，LI Rui，CHEN Pengchao，et al.Identification and evaluation on bending deformation of China-Russia Eastern Gas Pipeline［J］.Oil & Gas Storage and Transportation，2020，39(7):763-768
［7］何仁洋，刘艳贺，王海涛，等.基于IMU数据的应变解析技术及其应用［J］.科学技术与工程，2024，24(7):2683-2689.HE Renyang，LIU Yanhe，WANG Haitao，et al.Strain analysis technology based on IMU data and it’s application ［J］.Science and Engineering，2024，24(7):2683-2689
［8］AJMI C，ZAPATA J，MARTNEZ-LVAREZ J J，et al.Using deep learning for defect classification on a small weld X-ray imaged at a set［J］.Journal of Nondestructive Evaluation，2020，39(3):1-13.
［9］LE V K,CHEN Z,WONG Y W,et al.A complete online-SVM pipeline for case-based reasoning system:a study on pipe defect detection system［J］.Soft Computing,2020(1):16917-16933.
［10］GAO B X,ZHAO H,MIAO X Y.Multi-modal cascade detection of pipeline defects based on deep transfer metric learning［J］.Engineering Failure,Analysis,2024,160:108216.
［11］YAN B C,ZHENG J F,LI R，et al.Semi-supervised pipeline anomaly detection algorithm based on memory items and metric learning［J］.Nondestructive Testing and Evaluation，2023,38(5):753-766.
［12］CHEN K,LI H T,LI C S,et al.An automatic defect detection system for petrochemical pipeline based on cycle-GAN and YOLOv5［J］.Sensors 2022,22(20),7907
［13］崔国宁.基于深度学习的管道缺陷漏磁数据识别方法研究［D］.沈阳：沈阳工业大学，2023.
［14］刘啸奔，刘燊，季蓓蕾,等.基于IMU数据的管道弯曲变形段智能识别方法［J］.油气储运，2021，40(11):1228-1235.LIU Xiaoben，LIU Shen，JI Beilei,et al.Intelligent identification method of pipeline sections with bending deformation based on IMU data ［J］.Oil & Gas Storage and Transportation，2021，40(11):1228-1235.
［15］刘啸奔，张宏，夏梦莹,等.基于主成分分析和神经网络的管道泄漏识别方法［J］.油气储运，2015，34(7):737-740LIU Xiaoben，ZHANG Hong，XIA Mengying，et al.Pipeline leakage recognition based on principal component analysis and neural network ［J］.Oil & Gas Storage and Transportation，2015，34(7):737-740.
［16］郭梦琪.基于IMU检测数据的中俄原油管道凹陷深度定量识别方法研究［D］.北京：中国石油大学(北京)，2023.
［17］王琳，马林杰，徐建,等.基于深度学习的油气管道变形管段识别方法［J］.石油机械，2023，51(11):11-19.WANG Lin，MA Linjie，XU Jian,et al.Deformed section identification of oil and gas pipeline based on deep learning ［J］.China Petroleum Machinery，2023,51(11):11-19.
［18］陈颖，张仲伍.基于聚类分析和主成分分析的城市空气质量评价——以山西省11个地级市为例［J］.山西师范大学学报(自然科学版)，2020，34(4):72-78..CHEN Ying，ZHANG Zhongwu.Urban air quality assessment based on cluster analysis and principal component analysis:taking 11 prefecture-level cities in Shanxi province as an example ［J］.Journal of Shanxi Normal University(Natural Science Edition)，2020，34(4):72-78.
［19］曹梅，杨超宇.基于小波的CNN-LSTM-Attention瓦斯预测模型研究［J］.中国安全生产科学技术，2023，19(9):69-75.CAO Mei，YANG Chaoyu.Research on CNN-LSTM-Attention gas prediction model based on wavelet ［J］.Journal of Safety Science and Technology，2023，19(9):69-75.
［20］魏永合，宫俊宇.基于CNN-LSTM-Attention的滚动轴承故障诊断［J］.沈阳理工大学学报，2022，41(4):73-77.WEI Yonghe，GONG Junyu.Fault diagnosis in rolling bearing based on CNN-LSTM-Attention ［J］.Journal of Shenyang Ligong University，2022，41(4):73-77.
［21］王玉静，李少鹏，康守强,等.结合CNN和LSTM的滚动轴承剩余使用寿命预测方法［J］.振动.测试与诊断，2021，41(3):439-446,617.WANG Yujing，LI Shaopeng，KANG Shouqiang，et al.Method of predicting remaining useful life of rolling bearing combining CNN and LSTM ［J］.Journal of Vibration,Measurement & Diagnosis，2021，41(3):439-446,617.
［22］贾全烨，张强，宋博川.一种基于循环神经网络的电网客服语音文本实体识别算法［J］.供用电，2020，37(6):13-20.JIA Quanye，ZHANG Qiang，SONG Bochuan.A text entity recognition algorithm based on recurrent neural network for customer service voice of state grid ［J］.Distribution & Utilization，2020，37(6):13-20.
［23］CHAUDHARI S, MITHAL V，POLATKAN G，et al.An attentive survey of attention models［J］.ACM Transactionson Intelligent Systemsand Technology，2021，12(5):1-32.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期： 2024-06-20
* 基金项目： 国家重点研发计划项目(2022YFC3070100)；北京市科协“青年人才托举工程”项目(BYESS2023261)；国家自然科学基金项目(52304013)；国家管网科学研究与技术开发项目(WZXGL202105，WZXGL202104)；中国石油大学(北京)科研基金项目(2462023BJRC005);国家管网集团科技研发项目(CLZB202301)
作者简介： 张昊宁，硕士研究生，主要研究方向为管道内检测数据处理分析。
更新日期/Last Update: 2025-01-26