Chinese Journal of Lasers, Volume. 51, Issue 17, 1701005(2024)

Laser Frequency Stabilization Method Based on Intelligent Identifying Absorption Peaks with Convolutional Neural Network

Benyong Chen, Yong Zhao, Yingtian Lou, Liping Yan*, Jiandong Xie, Liang Yu, and Jianjun Tang
Author Affiliations
  • Precision Measurement Laboratory, School of Information Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, Zhejiang , China
  • show less
    Figures & Tables(16)
    SAS laser frequency stabilization method based on CNN intelligent recognition of absorption peaks and phase automatic matching
    Saturation absorption spectroscopy signal after preprocessing
    Absorption peaks marking. (a) 87Rb F=1→F′ transition; (b) 85Rb F=2→F′ transition; (c) 85Rb F=3→F′ transition; (d) 87Rb F=2→F′ transition
    One-dimensional convolutional neural network model
    Convergence trend chart of loss function
    Predicted sequence number of absorption peaks
    Error of the predicted results
    Error signal under different phase delay values Δφ. (a) 0°‒150°; (b) 180°‒330°
    Schematic diagram of error signal demodulation based on phase delay automatic extraction and matching
    Phase delay and error signal before and after phase delay matching. (a) Phase delay; (b) error signal
    Amplitude changes of three signals before and after laser frequency locking. (a) Scanning control signal of PZT; (b) saturation absorption spectral signal; (c) error signal
    Frequency spectrum of beat signal
    Frequency fluctuation of beat signal
    ECDL laser frequency stability after locking
    • Table 1. Spectral frequencies corresponding to four groups of absorption peak signals[25]

      View table

      Table 1. Spectral frequencies corresponding to four groups of absorption peak signals[25]

      Peak groupTransitionFrequency v /MHzΔv /MHzPeak groupTransitionFrequency v /MHzΔv /MHz
      1:87Rb F=1→F′ transition0384234454.14:87Rb F=2→F′ transition1384227691.6
      CO0-1384234490.236.1CO1-2384227770.178.5
      1384234526.336.12384227848.678.5
      CO0-2384234568.742.4CO1-3384227903.454.8
      CO1-2384234604.836.1CO2-3384227981.978.5
      2384234683.278.43384228115.2133.3
      2:85Rb F=2→F′ transition13842320643:85Rb F=3→F′ transition2384229057.6
      CO1-2384232078.714.7CO2-3384229089.331.7
      2384232093.414.7338422912131.7
      CO1-3384232110.417CO2-4384229149.728.7
      CO2-3384232125.114.7CO3-4384229181.431.7
      3384232156.831.74384229241.760.3
    • Table 2. One-dimensional CNN parameters

      View table

      Table 2. One-dimensional CNN parameters

      LayerKernel size /strideQuantity of filtersOutput shape
      Input15550×1×1
      Conv 13/125550×1×2
      Maxpool 12/222775×1×2
      Conv 23/122775×1×2
      Maxpool 22/221388×1×2
      Conv 33/141388×1×4
      Maxpool 32/24694×1×4
      Conv 43/14694×1×4
      Maxpool 42/24347×1×4
      Conv 564/18347×1×8
      Maxpool 52/28174×1×8
      Dense 11180×1×1
      Dense 2124×1×1
    Tools

    Get Citation

    Copy Citation Text

    Benyong Chen, Yong Zhao, Yingtian Lou, Liping Yan, Jiandong Xie, Liang Yu, Jianjun Tang. Laser Frequency Stabilization Method Based on Intelligent Identifying Absorption Peaks with Convolutional Neural Network[J]. Chinese Journal of Lasers, 2024, 51(17): 1701005

    Download Citation

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

    Category: laser devices and laser physics

    Received: Oct. 19, 2023

    Accepted: Dec. 7, 2023

    Published Online: Aug. 31, 2024

    The Author Email: Yan Liping (yanliping@zstu.edu.cn)

    DOI:10.3788/CJL231308

    CSTR:32183.14.CJL231308

    Topics