基于统计慢特征分析的输气管道流体状态在线监测实验研究

天津科技 ›› 2025, Vol. 52 ›› Issue (05): 5-10.

基于统计慢特征分析的输气管道流体状态在线监测实验研究

刘金海¹, 王俊峰¹, 刘学涛¹, 王璐瑶¹, 李凌涵², 梁光辉²

1.中海油能源发展股份有限公司采油服务分公司天津 300452;
2.天津大学电气自动化与信息工程学院天津 300072

收稿日期:2025-04-09 发布日期:2026-01-05

Experimental study on online monitoring of fluid flow state in gas pipelines based on statistical slow feature analysis

LIU Jinhai¹, WANG Junfeng¹, LIU Xuetao¹, WANG Luyao¹, LI Linghan², LIANG Guanghui²

1. CNOOC Energy Technology & Service-Oil Production Services Co.,Tianjin 300452,China;
2. School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China

Received:2025-04-09 Published:2026-01-05

摘要/Abstract

摘要： 输气管道经常面临复杂的输送环境工况,管道中可能存在积液,导致堵塞现象发生,影响输送效率和安全。针对输气管道的流体状态监测问题,采用基于超声兰姆波的非侵入式测量技术,提出统计慢特征分析（statistical slow feature analysis, SSFA）方法。先对超声检测信号进行滑窗处理,构建统计量模式矩阵,提取低阶和高阶统计信息;然后建立慢特征分析（slow feature analysis, SFA）模型对统计量模式矩阵进行特征提取,挖掘过程变化的动态信息,并计算统计量;最后采用贝叶斯推理法对监测统计量进行决策融合,提供直观精确的监测指标。超声测量实验表明,SSFA方法可以有效利用超声兰姆波信号准确监测输气管道的流动状态,相较于其他方法准确率最高。

关键词: 输气管道, 超声兰姆波, 统计量模式, 慢特征分析, 状态监测

Abstract: Gas pipelines often face complex transportation environment conditions,which may lead to the accumulation of liquid in the pipeline,affecting the transportation efficiency and safety. Aiming at the fluid state monitoring in gas pipelines,a non-invasive measurement technology based ultrasonic Lamb waves is utilized,and a statistical slow feature analysis （SSFA） method is proposed. Firstly,the ultrasonic signal is processed by sliding windows to construct the statistical pattern matrix,which extracts the low-order and high-order statistical information. Then,a slow feature analysis （SFA） model is established to extract the features of the statistical pattern matrix,mining the dynamic information of process changes. Besides,the monitoring statistics are calculated based on slow features. Finally,Bayesian inference is used to make decision fusion of monitoring statistics to provide intuitive and accurate monitoring indicators. Experiment results show that SSFA method can effectively use ultrasonic Lamb wave signal to accurately monitor the fluid state in gas pipelines,and its accuracy is the highest compared with other methods.

Key words: gas pipeline, ultrasonic Lamb wave, statistic pattern, slow feature analysis, state monitoring

中图分类号:

TE832

刘金海, 王俊峰, 刘学涛, 王璐瑶, 李凌涵, 梁光辉. 基于统计慢特征分析的输气管道流体状态在线监测实验研究[J]. 天津科技, 2025, 52(05): 5-10.

LIU Jinhai, WANG Junfeng, LIU Xuetao, WANG Luyao, LI Linghan, LIANG Guanghui. Experimental study on online monitoring of fluid flow state in gas pipelines based on statistical slow feature analysis[J]. TIANJIN SCIENCE & TECHNOLOGY, 2025, 52(05): 5-10.

参考文献

[1] 梁法春,王骅钟,陈婧,等. 复杂山地湿气集输管道积液动态累计规律[J]. 北京石油化工学院学报,2018,26(3):32-35,86.
[2] 李鹏,宋振坤,刘明仁,等. 煤层气集输管道积液判断及治理技术[J]. 中国煤层气,2021,18(3):35-36,22.
[3] 罗朝莉,朱冰,王波,等. 铝板表面裂纹的激光超声检测与信号处理研究[J]. 电子测量与仪器学报,2023,37(10):41-52.
[4] YANG L,JIANG H Q.Weld defect classification in radiographic images using unified deep neural network with multi-level features[J]. Journal of Intelligent Manufacturing,2021,32(2):459-469.
[5] MAO X F,LEI Y Z.Thickness measurement of metal pipe using swept-frequency eddy current testing[J]. NDT & E International,2016,78:10-19.
[6] SHI X W,TAN C,DONG F,et al.Oil-gas-water three-phase flow characterization and velocity measurement based on time-frequency decomposition[J]. International Journal of Multiphase Flow,2019,111:219-231.
[7] MURAKAWA H,MURAMATSU E,SUGIMOTO K,et al.A dealiasing method for use with ultrasonic pulsed Doppler in measuring velocity profiles and flow rates in pipes[J]. Measurement Science and Technology,2015,26(8):085301.
[8] 张瀚瑞,董峰,谭超. 液固两相流压降规律及超声法压降测量[J]. 仪器仪表学报,2021,42(10):46-54.
[9] KIEFER D A,PONSCHAB M,RUPITSCH S J,et al.Calculating the full leaky Lamb wave spectrum with exact fluid interaction[J]. The Journal of the Acoustical Society of America,2019,145(6): 3341-3350.
[10] 张浩亮,顾根堂,薛志亮,等. 煤层气集输管网积液在线监测及处理技术[J]. 中国煤层气,2023,20(6):22-25.
[11] 伍坤一,张露露,林宇,等. 四川某气田集输管线积液影响规律研究[J]. 西南石油大学学报(自然科学版),2024,46(1):136-146.
[12] 严帅,熊新. 基于KPCA和TCN-Attention的滚动轴承退化趋势预测[J]. 电子测量技术,2022,45(15):28-34.
[13] JIANG B B,HUANG D X,ZHU X X,et al.Canonical variate analysis-based contributions for fault identification[J]. Journal of Process Control,2015,26:17-25.
[14] FENG Z X,LI Y G,SUN B,et al.Abnormal operating condition identification of industrial processes based on deep learning with global-local slow feature analysis[J]. IEEE Transactions on Instrumentation Measurement,2023,72:3267347.
[15] 张汉元,田学民,邓晓刚. 基于SPA相似系数的故障识别方法[J]. 化工学报,2013,64(12):4503-4508.
[16] 胡宸,邬冠华,罗文斌,等. 噪声环境下空耦超声Lamb波铝板扫查成像定量研究[J]. 电子测量与仪器学报,2021,35(5):120-127.
[17] 王丙泉,赵勃. 复合材料板的空耦超声Lamb波原位应力测量方法[J]. 仪器仪表学报,2023,44(8):100-108.
[18] SHI Z X,BAO Y,TAN C,et al.A non-intrusive method for liquid level measurement in vertical pipeline using ultrasonic lamb wave[C]//2024 IEEE International Instrumentation and Measurement Technology Conference (I2MTC). Glasgow,United Kingdom:IEEE,2024:1-6.
[19] ZHANG Z,LAI X Z,LU C D,et al.Statistics pattern cointegration analysis-based bit bounce detection for drilling process[J]. IEEE Transactions on Industrial Electronics,2024,71(2):2039-2048.
[20] LI L H,HAN X Y,DONG F,et al.Zero-shot state identification of industrial gas-liquid two-phase flow via supervised deep slow and steady feature analysis[J]. IEEE Transactions on Industrial Informatics,2024,20(6):8170-8180.
[21] HUANG J,ERSOY O K,YAN X F.Slow feature analysis based on online feature reordering and feature selection for dynamic chemical process monitoring[J]. Chemometrics and Intelligent Laboratory Systems,2017,169:1-11.