Infrared and Laser Engineering, Volume. 51, Issue 12, 20220179(2022)

Characterization and identification of static and dynamic hammer tail characteristics in infrared temperature field of leaking steam

Tao Wu, Cancheng Xiong, Cong Li, Yuanyuan Xu, and Shengping Li
Author Affiliations
  • Department of Mechanical Engineering, Shantou University, Shantou 515063, China
  • show less
    Figures & Tables(27)
    Settings of model boundary condition
    Steam leakage development process
    Influence of leakage pressure, leakage amount and leakage direction on temperature distribution
    Influence of turbulence scale on temperature distribution
    Temperature distribution of leakage steam under different leakage pressure, leakage volume and leakage direction
    Schematic diagram of leakage steam
    (a) Temperature layer a; (b) Temperature layer b; (c) Temperature layer c; (d) Judgment result of intersection area of a and b; (e) Judgment result of intersection area of b and c
    (a) Temperature layer ; (b) Temperature layer ; (c) Temperature layer; (d) Temperature layer ; (e) Temperature layer ; (f) Temperature layer ; (h) Difference area between temperature layer and ; (i) Difference area between temperature layer and ; (j) Difference area between temperature layer and (a) 温度层;(b) 温度层;(c)温度层;(d)温度层;(e)温度层;(f) 温度层;(h) 温度层和;(i) 温度层和;(j) 温度层和
    (a) Skeleton; (b) Fitting line; (c) Direction of leakage source prediction
    (a) Infrared image of leakage steam collected by FOTCRIC 686; (b) Extracted suspected steam area
    (a) Original image; (b) Adaptive median filter (window size 5×5); (c) Adaptive median filter (window size 11×11); (d) Adaptive median filter (window size 25×25)
    Temperature layer extraction effect
    Variable scale extraction processing effect
    Frame diagram of Mask R-CNN
    (a) Binarization processing; (b) Dimensional adjustment; (c) Labeling
    (a) Judged as hammer-tail; (b) Judged as not a hammer-tail
    Flow chart of steam temperature field identification algorithm
    Steam experimental platform
    Original image and two consecutive infrared image
    Variable scale extraction processing
    Results of Mask R-CNN network model identification
    Judgment results of diffusion characteristic
    Judgment results of dynamic hammer tail characteristics
    Prediction results of steam leakage direction
    • Table 1. The results of data set segmentation

      View table
      View in Article

      Table 1. The results of data set segmentation

      Label nameTotal dataTraining data (70%)Validation data (20%)Testing data (10%)
      Total data3772657537
      Hammer-tail3492447035
      Not a hammer-tail282152
    • Table 2. Main relevant parameter settings

      View table
      View in Article

      Table 2. Main relevant parameter settings

      Parameter nameParameter setting
      Number of FPN network layers and output layers [256, 512, 1024, 2 048],256
      Transform scale parameters of FPN network [0.25, 0.125, 0.0625, 0.03125]
      Number of RPN channels and threshold256,0.5
      Positive and negative IoU confidence 0.7,0.3
    • Table 3. The results of training

      View table
      View in Article

      Table 3. The results of training

      ModelThe number of iterationsLearning rateAccuracy rate
      Training one120.001250.9071
      Training two120.00250.8307
      Training three200.001250.9045
      Training four120.000620.8975
      Training five120.000800.8790
    Tools

    Get Citation

    Copy Citation Text

    Tao Wu, Cancheng Xiong, Cong Li, Yuanyuan Xu, Shengping Li. Characterization and identification of static and dynamic hammer tail characteristics in infrared temperature field of leaking steam[J]. Infrared and Laser Engineering, 2022, 51(12): 20220179

    Download Citation

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

    Category: Image processing

    Received: Mar. 14, 2022

    Accepted: --

    Published Online: Jan. 10, 2023

    The Author Email:

    DOI:10.3788/IRLA20220179

    Topics