Acta Optica Sinica, Volume. 45, Issue 16, 1612003(2025)

Scalar Field Velocimetry Based on Deep Optical Flow Neural Network

Mingtao Jiang1,3,4, Wenjie Xu1, and Wei Chen2、*
Author Affiliations
  • 1School of Mechanics and Engineering Science, Shanghai University, Shanghai 200400, China
  • 2School of Marine Engineering Equipment, Zhejiang Ocean University, Zhoushan 316000, Zhejiang , China
  • 3Nanyang Environment & Water Research Institute, Nanyang Technological University, Singapore 637141, Singapore
  • 4Marine Ecological Restoration and Smart Ocean Engineering Research Center of Hebei Province, Qinhuangdao 066000, Hebei , China
  • show less
    Figures & Tables(19)
    SIV-RAFT neural network
    Deformable convolutional networks
    Iterative attention feature fusion module
    Schematic diagram of drag flume experiment
    Centrelines of negatively buoyant jets
    Distributions of centreline mass fractions in vertical sections of 45° jets. (a) 20D; (b) 30D
    Scalar and velocity fields of 45° jet. (a)(b) Instantaneous scalar field images of two adjacent frames; (c) reference velocity field; velocity fields predicted from scalar fields using (d) SIV-RAFT, (e) PWC-Net, and (f) HS algorithms, respectively
    Error analysis results of jet velocity fields. (a) SIV-RAFT; (b) PWC; (c) HS
    Spectrum analysis of 45° jet. (a) E(k); (b) Eω(k)
    Centreline velocity decay curves of 45° jet
    Distributions of velocity in vertical sections at different centreline distances of 45° jets. (a) 20D; (b) 30D
    Velocity amplitude maps of DNS turbulent flow (t=30). (a) Velocity amplitude map of real DNS; (b) SIV-RAFT; (c) PWC; (d) HS
    Vorticity maps of DNS turbulent flow (t=30). (a) Instantaneous vorticity map of real DNS; (b) SIV-RAFT; (c) PWC; (d) HS
    Velocity spectra of DNS turbulent flow (t=30). (a) Horizontal direction; (b) vertical direction
    Velocity spectrum error of DNS turbulent flow (t=30). (a) Horizontal direction; (b) vertical direction
    • Table 1. Experiment conditions

      View table

      Table 1. Experiment conditions

      Caseθ /(°)U0 /(m/s)ρ0 /(kg/m3ρa /(kg/m3g0 /(m/s2ReFrD /mm
      L1300.531031998.30.324240158
      L2300.581031998.30.324797108
      L3450.531031998.30.324240158
      L4450.581031998.30.324797108
    • Table 2. Simulated working conditions of LES turbulence model

      View table

      Table 2. Simulated working conditions of LES turbulence model

      Caseθ /(°)U0 /(m/s)ρ0 /(kg/m3ρa /(kg/m3g0 /(m/s2ReFrNFr
      L1150.751031998.30.326109150
      L2150.501031998.30.324073100.5
      L3300.751031998.30.326109150
      L4300.501031998.30.324073100.5
      L5450.751031998.30.326109150
      L6450.501031998.30.324073100.5
      L7600.751031998.30.326109150
      L8600.501031998.30.324073100.5
    • Table 3. Description of SIV-RV-RAFT dataset

      View table

      Table 3. Description of SIV-RV-RAFT dataset

      Caseθ /(°)U0 /(m/s)T /sρ0 /(kg/m3ρa /(kg/m3ReQuantity
      L1150.7510‒501031998.361091000
      L2150.510‒501031998.340731000
      L3300.7510‒501031998.361091000
      L4300.510‒501031998.340731000
      L5450.7510‒501031998.361091000
      L6450.510‒501031998.340731000
      L7600.7510‒501031998.361091000
      L8600.510‒501031998.340731000
    • Table 4. Training and test results

      View table

      Table 4. Training and test results

      AlgorithmRMSE
      Training of jet

      Test of

      jet

      Training of

      DNS

      Test of

      DNS

      SIV-RAFT0.4250.8710.682
      PWC-Net0.5741.1070.886
      HS1.5351.118
    Tools

    Get Citation

    Copy Citation Text

    Mingtao Jiang, Wenjie Xu, Wei Chen. Scalar Field Velocimetry Based on Deep Optical Flow Neural Network[J]. Acta Optica Sinica, 2025, 45(16): 1612003

    Download Citation

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

    Category: Instrumentation, Measurement and Metrology

    Received: Apr. 24, 2025

    Accepted: May. 26, 2025

    Published Online: Aug. 18, 2025

    The Author Email: Wei Chen (wchen@zjou.edu.cn)

    DOI:10.3788/AOS251001

    CSTR:32393.14.AOS251001

    Topics