Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Acta Optica Sinica, Volume. 41, Issue 6, 0611004(2021)

Research on Active Optical Correction Algorithm Based on Deep Learning

Chao Kang1,2, Wenxiang Li1,2、**, Sheng Huang3, Hengrui Guan1,2, Jinbiao Zhao3, and Qingsheng Zhu1,2,3、*
Author Affiliations
  • 1CAS Nanjing Astronomical Instruments Research Center, Nanjing, Jiangsu 210042, China
  • 2University of Science and Technology of China, Hefei, Anhui 230026, China
  • 3CAS Nanjing Astronomical Instruments Co., LTD., Nanjing, Jiangsu 210042, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (14)
    Equations (0)
    References (28)
    Cited By (5)
    Tools
    Paper Information
    Topics
    Figures & Tables(14)
    Framework of DLCM algorithm

    Fig. 1. Framework of DLCM algorithm

    Download full size
    Framework and parameters of kinetic model

    Fig. 2. Framework and parameters of kinetic model

    Download full size
    Loss decline curve of strategy network

    Fig. 3. Loss decline curve of strategy network

    Download full size
    Framework and parameters of strategy network

    Fig. 4. Framework and parameters of strategy network

    Download full size
    Convergence process of evolutionary strategy algorithm

    Fig. 5. Convergence process of evolutionary strategy algorithm

    Download full size
    Actuator distribution map. (a) Support method of standard spherical mirror; (b) ANSYS simulation model

    Fig. 6. Actuator distribution map. (a) Support method of standard spherical mirror; (b) ANSYS simulation model

    Download full size
    Running time of DLCM algorithm

    Fig. 7. Running time of DLCM algorithm

    Download full size
    Comparison of effect of first correction. (a) Before correction; (b) corrected results

    Fig. 8. Comparison of effect of first correction. (a) Before correction; (b) corrected results

    Download full size

    • Table 1. Relationship between model network layers and accuracy

      View table

      Table 1. Relationship between model network layers and accuracy

      Number of layers3456
      Accuracy /%91.3595.1497.5497.60

    • Table 2. Relationship between model network dropout density and accuracy

      View table

      Table 2. Relationship between model network dropout density and accuracy

      Dropout density0.30.40.5
      Accuracy /%98.4198.3798.18

    • Table 3. Relationship between correction rate and convolutional layers and FC layers

      View table

      Table 3. Relationship between correction rate and convolutional layers and FC layers

      FC layerConvolutional layer
      12345
      2/%84.3784.9290.8196.1497.92
      3/%86.4188.5993.0198.4298.51
      4/%85.2088.3892.9798.3498.27

    • Table 4. Main parameters of spherical mirror

      View table

      Table 4. Main parameters of spherical mirror

      NameParameter
      Diameter /mm1000
      Thickness /mm80
      Radius of curvature /mm4000
      MaterialK9 glass
      Mass /kg174.445

    • Table 5. Hardware and software parameters of algorithm training platform

      View table

      Table 5. Hardware and software parameters of algorithm training platform

      NameParameter
      CPUIntel Core i7-4790 CPU @3.6 GHz
      Memory16 GB
      Graphics cardNVIDIA GeForce GTX980 Ti
      SystemWindows 7 professional
      EnvironmentPython3.7, PyTorch 1.5.1-GPU

    • Table 6. Comparison of calibration results of three algorithms

      View table

      Table 6. Comparison of calibration results of three algorithms

      MethodInitialstateResultNumberof timesSinglepromotionratio /%
      DLS0.27λ0.02λ246.26
      LS0.27λ0.04λ242.60
      DLCM0.27λ0.01λ196.30
      DLS0.56λ0.03λ327.98
      LS0.56λ0.02λ424.11
      DLCM0.56λ0.02λ196.42
      DLS0.86λ0.02λ519.54
      LS0.86λ0.05λ615.70
      DLCM0.86λ0.03λ196.51
      DLS1.21λ0.05λ519.17
      LS1.21λ0.04λ713.81
      DLCM1.21λ0.02λ249.17
    Tools

    Get Citation

    Copy Citation Text

    Chao Kang, Wenxiang Li, Sheng Huang, Hengrui Guan, Jinbiao Zhao, Qingsheng Zhu. Research on Active Optical Correction Algorithm Based on Deep Learning[J]. Acta Optica Sinica, 2021, 41(6): 0611004

    Download Citation

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

    Category: Imaging Systems

    Received: Aug. 25, 2020

    Accepted: Nov. 5, 2020

    Published Online: Apr. 7, 2021

    The Author Email: Li Wenxiang (lwxiang@mail.ustc.edu.cn), Zhu Qingsheng (85482014@163.com)

    DOI:10.3788/AOS202141.0611004

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology