NUCLEAR TECHNIQUES, Volume. 48, Issue 1, 010601(2025)

Prediction of heat transfer parameters of nuclear reactor based on physical information machine learning algorithm

Dexiang KONG, Yichao MA, Jing ZHANG*, Mingjun WANG, Yingwei WU, Yanan HE, Kailun GUO, Wenxi TIAN, and Guanghui SU
Author Affiliations
  • School of Energy and Power Engineering, Xi'an Jiaotong University, Xi'an 710049, China
  • show less
    Figures & Tables(12)
    Framework of PIML model
    Structural diagram of MLP
    Structural diagram of decision tree
    Structural diagram of RF model
    HTC experimental data distribution of mass flux (a), pressure (b) and equilibrium quality (c)
    Relationship between the number of MLP hidden layer neurons and the error in the HTC model
    Relationship between the number of BPNN hidden layer neurons and the error in the HTC model
    Relationship between the number of RF decision trees and the error in the HTC model
    Prediction results of experimental data by six HTC models, MLP_Thom (a), MLP_JL (b), BPNN_Thom (c), BPNN_JL (d), RF_Thom (e) and RF_JL (f)
    Cumulative data fraction of MAPE in models associated with Thom (a) and Jens-Lottes (b)
    • Table 1. Calculation results of HTC model and formulas on experimental data

      View table
      View in Article

      Table 1. Calculation results of HTC model and formulas on experimental data

      HTC模型

      HTC model

      MAPE / %

      累计数据份额

      Cumulative data fraction / %

      <20%<35%<50%
      MLP_Thom8.2193.1899.62100
      MLP_JL7.9493.18100100
      BPNN_Thom6.0298.48100100
      BPNN_JL6.1198.86100100
      RF_Thom3.3898.1199.62100
      RF_JL3.1799.62100100
      Thom104.01.138.315.53
      Jens-Lottes27.6949.2468.1887.12
    • Table 2. Pressure extrapolation in RF_JL model

      View table
      View in Article

      Table 2. Pressure extrapolation in RF_JL model

      参量 Parameters数值Value
      模型训练范围Training range / MPa0.11~0.170.11~0.160.11~0.150.11~0.140.11~0.13
      验证范围Test range / MPa0.18~0.220.17~0.220.16~0.220.15~0.220.14~0.22
      训练数据量No. of training data244230212171116
      验证数据量No. of test data20345293148

      平均相对误差

      MAPE / %

      RF_JL模型RF_JL model9.859.2411.6712.6227.71
      Jens-Lottes关系式Jens-Lottes14.3616.4718.2221.7523.67
    Tools

    Get Citation

    Copy Citation Text

    Dexiang KONG, Yichao MA, Jing ZHANG, Mingjun WANG, Yingwei WU, Yanan HE, Kailun GUO, Wenxi TIAN, Guanghui SU. Prediction of heat transfer parameters of nuclear reactor based on physical information machine learning algorithm[J]. NUCLEAR TECHNIQUES, 2025, 48(1): 010601

    Download Citation

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

    Category: NUCLEAR ENERGY SCIENCE AND ENGINEERING

    Received: Jul. 12, 2024

    Accepted: --

    Published Online: Feb. 26, 2025

    The Author Email: ZHANG Jing (章静)

    DOI:10.11889/j.0253-3219.2025.hjs.48.240285

    Topics