计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (10): 134-141.doi: 10.11896/jsjkx.260400023

• 大模型与多模态智能感知 • 上一篇    下一篇

基于多模态数据特征融合的光纤外破事件识别方法

李莞坡1,3, 石文玉2, 张用明2, 徐鹤1,3, 张玉杰1,3   

  1. 1 南京邮电大学计算机学院 南京 210023
    2 安徽新华学院大数据与人工智能学院 合肥 230088
    3 江苏省高性能计算与智能处理工程研究中心 南京 210023
  • 收稿日期:2026-04-07 修回日期:2026-06-08 发布日期:2026-10-13
  • 通讯作者: 张用明(zyming@njupt.edu.cn)
  • 作者简介:(zyming@njupt.edu.cn.cn)
  • 基金资助:
    江苏省研究生科研实践创新计划项目(SJCX24_0315);安徽新华学院博士人才科研启动经费项目(2026kyqd43)

Method for Optical Fibers External Damage Event Recognition Based on Multi-modal DataFeature Fusion

LI Wanpo1,3, SHI Wenyu 2, ZHANG Yongming 2, XU He1,3, ZHANG Yujie1,3   

  1. 1 School of Computer Science,Nanjing University of Post and Telecommunications,Nanjing 210023, China
    2 College of Big Data and Artificial Intelligence,Anhui Xinhua University,Hefei 230088, China
    3 Jiangsu HPC and Intelligent Processing Engineer Research Center,Nanjing 210023, China
  • Received:2026-04-07 Revised:2026-06-08 Online:2026-10-13
  • About author:LI Wanpo,born in 2002,postgraduate,is a member of CCF(No.A34988G).Her main research interests include data mining and intelligent software.
    ZHANG Yongming,born in 1976,Ph.D,senior engineering,is a member of CCF(No.15973G).His main research in-terests include artificial intelligence technology and application,and so on.
  • Supported by:
    Postgraduate Research & Practice Innovation Program of Jiangsu Province(SJCX24_0315) and Doctoral Research Start-up Fund of Anhui Xinhua University(2026kyqd43).

PDF (PC)

67

摘要： 针对地下管线环境复杂、噪声干扰严重及光纤长距离信号衰减等问题,提出一种基于多模态数据特征融合的光纤外破事件识别方法,以降低第三方施工破坏事件的误报率。该方法将光纤信号的时序数据与时频图像数据相结合,构建多模态数据输入;提取时频域特征并建模光纤通道间的空间关联,以更全面地表征事件;引入一种分层注意力融合机制,利用模内注意力与模间注意力分别学习各模态内部的关键信息及模态间的互补性互信息,并利用门控单元进行自适应加权融合,从而构建了一个鲁棒性更强的光纤外破事件识别模型。实验结果表明,所提出的模型在人工挖掘、挖掘机、疑似事件等6类事件分类上有效提升了光纤振动信号事件识别的准确性,其召回率较传统机器学习与深度学习等基准模型分别提升了约20个百分点和30个百分点。

关键词: 多模态数据, 时频域特征, 注意力机制, 通道相关性, 光纤外破事件识别

Abstract: To address challenges including complex underground pipeline surroundings,severe noise interference,and long-distance signal attenuation of optical fibers,this paper proposes a multi-modal feature fusion-based identification method for external damage incidents of optical cables to reduce the false alarm rate induced by third-party construction damage.The proposed me-thod constructs multi-modal input by combining time-series data and time-frequency image data of optical fibers signals for comprehensive event characterization.It extracts time-and frequency-domain features and models spatial correlations across different fiber channels to fully depict event attributes.A hierarchical attention fusion mechanism is further introduced:intra-modal attention captures critical information within individual modalities,while inter-modal attention learns complementary mutual information across modalities;gated units are adopted for adaptive weighted fusion to establish a robust identification model for optical fiber external damage events.Experimental results demonstrate that the proposed model improves the classification accuracy of fiber vibration signals across six event categories including manual excavation,excavator operation and suspected incidents.Compared with conventional machine learning and deep learning baseline models,the recall rate of the proposed model is increased by 20 percentage points and 30 percentage points respectively.

Key words: Multimodal data, Time-frequency domain features, Attention mechanisms, Channel correlation, Optical fiber external damage event identification

中图分类号: 

  • TP391
[1] ZHAO L J,ZHANG X Z,XU Z Y,et al.Influencing Factors of IQ Demodulation Method in Distributed Acoustic Sensing(DAS) Systems[J].Acta Optica Sinica,2023,43(14):1-16.
[2] WANG L S,XIE X,YAO M,et al.Distributed Optical FiberAcoustic Wave Sensing and Its Progress[J].Study on Optical Communications,2025(5):1-10.
[3] LIU H,ZHANG Q,WU B Y,et al.Frequency-Shift Phase Ge-neration Carrier Demodulation for Noise Reduction in-OTDR[J].IEEE Photonics Technology Letters,2024,36(11):725-728.
[4] RONG J Y,ZhANG Y C,CUI T,et al.Frequency locking and measurement of feedback bandwidth of a fiber Fabry-Pérot cavity based on FPGA[J].Journal of Quantum Optics,2026,33(2):1-10.
[5] ZhANG L.Research and development of a 20 Gb/s PAM4 optical communication ASIC in high-energy physics experiment [D].Wuhan:College of Physical Science and Technology Central China Normal University,2022.
[6] LI Y Y,SHAO W J,WEI D S,et al.Design of a fiber bragggrating demodulation system based on dual-core DSP[J].Electronic Measurement Technology,2025,48(7):135-141.
[7] JIANG F,LI H L,ZHANG ZH H,et al.An event recognition method for fiber distributed acoustic sensing systems based on the combination of MFCC and CNN[C]//International Confe-rence on Optical Instruments and Technology:Advanced Optical Sensors and Applications.SPIE,2018:1-7.
[8] WU S H,REN Z H,HE S H,et al.Event recognition based on the distributed located fiber sensor[J].Intelligent Computer and Applications,2021,11(9):48-53.
[9] BAI Y,XING J C,XIE F,et al.Detection and identification ofexternal intrusion signals from 33km optical fiber sensing system based on deep learning[J].Optical Fiber Technology,2019,53:102060.
[10] LI S Z,PENG R Z,LIU Z L.A surveillance system for urban buried pipeline subject to third-party threats based on fiber optic sensing and convolutional neural network[J].Structural Health Monitoring,2020,19(6):1-12.
[11] YAN S,ZOU N M,SHANG Y,et al.LR-Net for Weak Vibra-tion Event Location and Recognition with DAS[J].IEEE Sensors Journal,2025,25(23):42954-42963.
[12] ZHANG S X,LI H,FAN C Z,et al.Adaptive decentralized AI scheme for signal recognition of distributed sensor systems[J].Opto-Electronic Advances,2024,7(12):240119.
[13] WANG D D,ZHANG H J,WANG P F,et al.A DAS Monitoring System for Underground Cables Using Multi-Scale Dynamic Mel Frequency Cepstral Features[J].IEEE Sensors Journal,2025,25(11):19323-19331.
[14] FANG J B,WANG J B,CHEN J X.A Graphical Symbol Detection Method for Substation Layout Drawings Based on Text-Image Multimodal Fusion[J].Computer Science,2026,53(1):206-215.
[15] CHEN Y A,MA Z L.A Review of Multimodal AlignmentMethods for Large Models[C]//Proceedings of the 11th National BIM Academic Conference.2025:368-374.
[16] WU X L,ZHANG L,ZHAI M,et al.Information visualization representation with coupled visual task information flow and physiological cognition[J].Journal of Huazhong University of Science and Technology(Natural Science Edition),2025,53(12):1-11
[17] DONG S X,FAN W,WU T,et al.Teaching Time Series to See and Speak:Forecasting with Aligned Visual and Textual Perspectives[J].arXiv:2506.24124v2,2025.
[18] LAI Y,YUAN S H,NASSAR Y,et al.Natural Multimodal Fusion-Based Human-Robot Interaction Application With Voice and Deictic Posture via Large Language Model[J].IEEE Robo-tics & Automation Magazine,2025,33(2):29-38.
[19] CHEN G G,LIU H,LI H T,et al.Research on Subgraph Ma-tching Contrastive Learning Methods for Multimodal Pre-training[J].Chinese Journal of Computers,2025,48(4):893-909.
[20] LI C Y,WONG C,ZHANG S,et al.LLaVA-Med:Training aLarge Language-and-Vision Assistant for Biomedicine in One Day[J].arXiv:2306.00890v1,2023.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发