NUCLEAR TECHNIQUES, Volume. 46, Issue 12, 120501(2023)

A neural network-based method for measuring the activity of waste drum

Minxiang SHU1, Chenyu SHAN2, Weiguo GU3, and Dezhong WANG3、*
Author Affiliations
  • 1College of Smart Energy, Shanghai Jiao Tong University, Shanghai 200240, China
  • 2China Nuclear Power Technology Research Institute, Shenzhen 518031, China
  • 3Machinery and Power Engineering College, Shanghai Jiao Tong University, Shanghai 200240, China
  • show less
    Figures & Tables(24)
    Schematic diagram of point source equivalent to ring source
    Diagram of measuring position
    Schematic diagram of collimator
    Schematic diagram of measurement system
    Detection efficiency of point sources at different angles when step is 20°
    Detection efficiency at different steps
    Results of count rate (a) Interpolation, (b) Numeral calculation
    Diagram of neural network training process
    Loss function after 100 000 iterations
    Diagram of neural network testing proces
    Process diagram of activity reconstruction of waste drum based on neural network
    Predicted and true radii of ring source
    Activity reconstruction value of single ring source
    Reconstructed value of total activity from multiple sources
    Activity reconstruction value of SGS (a) Single source, (b) Multi-sources
    Two grid division methods
    Activity reconstruction results of TGS (a) Single source, (b) Multi-sources
    Activity reconstruction results of STGS (a) Single source, (b) Multi-sources
    • Table 1. Probe size parameters

      View table
      View in Article

      Table 1. Probe size parameters

      部件Component尺寸Size / cm
      晶体直径Crystal diameter6.09
      晶体长度Crystal length5.18
      冷指直径Cold finger diameter0.87
      冷指长度Cold finger length3.82
      死层厚度Dead layer thickness0.07
      真空层厚度Vacuum layer thickness0.50
      铝壳厚度Aluminum shell thickness0.15
    • Table 2. Relative error at different steps (%)

      View table
      View in Article

      Table 2. Relative error at different steps (%)

      半径

      Radius / cm

      步长Step / (°)
      2451020
      5-0.47-0.15-1.020.02-0.16
      15-1.79-0.22-2.56-1.97-2.25
      300.48-0.65-1.22-1.17-0.69
    • Table 3. Count rate for detectors at different measurement positions

      View table
      View in Article

      Table 3. Count rate for detectors at different measurement positions

      半径

      Radius

      / cm

      偏移距离

      Offset distance / cm

      07142128
      054.7454.5054.0051.7233.58
      563.6563.2861.9757.7436.18
      1095.9995.3492.9576.5649.57
      15170.81169.55156.23121.9384.92
      20327.88317.73269.54214.53157.80
      25689.21632.38526.52411.81332.97
      301 427.171 352.021 164.03864.96763.96
      352 843.782 754.522 935.312 103.791 936.98
    • Table 4. Relative error of count rate between interpolation and numerical calculation (%)

      View table
      View in Article

      Table 4. Relative error of count rate between interpolation and numerical calculation (%)

      半径

      Radius / cm

      偏移距离Offset distance / cm
      07142128
      2.51.371.371.220.480.46
      7.5-2.15-2.29-2.33-3.06-2.39
      12.51.701.422.282.393.39
      17.5-0.88-0.60-1.880.75-0.53
      22.5-0.76-1.400.341.611.95
      27.5-0.54-0.14-3.59-2.49-3.14
      32.55.753.246.667.347.60
    • Table 5. Activity reconstruction results of single source

      View table
      View in Article

      Table 5. Activity reconstruction results of single source

      NGSSGSSTGS-3STGS-5TGS-60
      误差Error / %平均值Average4.2668.1544.5322.733.97
      最大值Maximum12.63114.54689.46251.8316.02
      标准差Standard deviation5.3375.32141.6748.444.97
    • Table 6. Activity reconstruction results of multi-source

      View table
      View in Article

      Table 6. Activity reconstruction results of multi-source

      NGSSGSSTGS-3STGS-5TGS-60
      误差Error / %平均值Average24.2748.0246.078.9528.61
      最大值Maximum42.4186.96133.2823.1480.63
      标准差Standard deviation27.9753.7171.5431.5356.38
    Tools

    Get Citation

    Copy Citation Text

    Minxiang SHU, Chenyu SHAN, Weiguo GU, Dezhong WANG. A neural network-based method for measuring the activity of waste drum[J]. NUCLEAR TECHNIQUES, 2023, 46(12): 120501

    Download Citation

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

    Category: Research Articles

    Received: Jun. 13, 2023

    Accepted: --

    Published Online: Mar. 7, 2024

    The Author Email: Dezhong WANG (王德忠)

    DOI:10.11889/j.0253-3219.2023.hjs.46.120501

    Topics