Acta Optica Sinica, Volume. 45, Issue 10, 1028003(2025)

Pipeline Leakage Event Recognition in Distributed Optical Fiber Sensing Using One‑Dimensional Convolutional Neural Network with Fusion Input of Features and Multi‑Parametric Signals

Muping Song**, Ning Jia*, and Enxue Cui
Author Affiliations
  • College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310027, Zhejiang , China
  • show less
    Figures & Tables(13)
    Diagram of distributed optical fiber multi-parameter sensing system
    Flowchart of signal preprocessing
    Model of multi-parametric signal fusion recognition. (a) Total structure of model; (b) structure of convolutional unit; (c) structure of fusion unit
    Confusion matrices of three methods. (a) Input of features; (b) input of vibration signals; (c) fusion input of vibration signals and features
    Comparison of recognition effects of features, signals, and combination of features and signals. (a) Comparison of accuracy; (b) comparison of precision; (c) comparison of recall; (d) comparison of F-score
    Confusion matrices of three methods. (a) Input of single parameter temperature signal; (b) fusion input of multi-parameter vibration and temperature signals; (c) fusion input of multi-parameter dynamic and static signals
    Comparison of recognition effects of single parameter signal, multi-parametric signals and multi-parametric dynamic and static signals. (a) Comparison of accuracy; (b) comparison of precision; (c) comparison of recall; (d) comparison of F-score
    Confusion matrix of proposed method
    Comparison of effects of proposed method and other methods. (a) Comparison of accuracy; (b) comparison of precision; (c) comparison of recall; (d) comparison of F-score
    • Table 1. Details of typical events

      View table

      Table 1. Details of typical events

      Event typeVibration detailTemperature detail
      Slow leakageChange in time domain; periodicitySlow temperature rise
      Quick leakageChange in time domainQuick temperature rise
      Knocking or diggingPeriodicityRemain stable
      Temperature interferenceNoiseFluctuation
    • Table 2. Setting for collection parameters

      View table

      Table 2. Setting for collection parameters

      Parameter

      Dynamic

      signal

      Static signalFeatureTemperature signal
      Sampling rate /Hz100016.71
      Length100010002060
    • Table 3. Construction of dataset and sample number of each type

      View table

      Table 3. Construction of dataset and sample number of each type

      LabelEvent typeDynamic signalStatic signalFeatureTemperature signal
      TrainTestTrainTestTrainTestTrainTest
      1Slow leakage1000200100020010002001000200
      2Quick leakage1000200100020010002001000200
      3Knocking or digging1000200100020010002001000200
      4Temperature interference1000200100020010002001000200
    • Table 4. Recognition effects of proposed method

      View table

      Table 4. Recognition effects of proposed method

      Event labelPrecision /%Recall /%F-score /%
      198.04100.0099.01
      2100.0097.5198.74
      394.2097.5095.82
      496.9194.0095.43
    Tools

    Get Citation

    Copy Citation Text

    Muping Song, Ning Jia, Enxue Cui. Pipeline Leakage Event Recognition in Distributed Optical Fiber Sensing Using One‑Dimensional Convolutional Neural Network with Fusion Input of Features and Multi‑Parametric Signals[J]. Acta Optica Sinica, 2025, 45(10): 1028003

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Remote Sensing and Sensors

    Received: Jan. 21, 2025

    Accepted: Mar. 31, 2025

    Published Online: May. 19, 2025

    The Author Email: Muping Song (songmp@zju.edu.cn), Ning Jia (22231121@zju.edu.cn)

    DOI:10.3788/AOS250523

    CSTR:32393.14.AOS250523

    Topics