• Acta Optica Sinica
  • Vol. 43, Issue 7, 0706003 (2023)
Wenjuan Sheng1、*, Haitao Lou1, and Gangding Peng2
Author Affiliations
  • 1College of Automation Engineering, Shanghai University of Electric Power, Shanghai 200090, China
  • 2College of Electrical Engineering and Telecommunications, The University of New South Wales, Sydney 2052, New South Wales, Australia
  • show less
    DOI: 10.3788/AOS221651 Cite this Article Set citation alerts
    Wenjuan Sheng, Haitao Lou, Gangding Peng. Dynamic Compensation of Tunable Filter Demodulation Error Based on Least Squares Support Vector Machine and Multi-Reference Gratings[J]. Acta Optica Sinica, 2023, 43(7): 0706003 Copy Citation Text show less
    Schematic diagram of dynamic modeling and compensation
    Fig. 1. Schematic diagram of dynamic modeling and compensation
    Schematic diagram of FBG demodulation system
    Fig. 2. Schematic diagram of FBG demodulation system
    Principle of centroid detection algorithm
    Fig. 3. Principle of centroid detection algorithm
    Temperature trend of reference grating in cooling mode
    Fig. 4. Temperature trend of reference grating in cooling mode
    Compensation result of dynamic model in cooling mode. (a) Absolute wavelength drift; (b) prediction error
    Fig. 5. Compensation result of dynamic model in cooling mode. (a) Absolute wavelength drift; (b) prediction error
    Temperature trend of reference grating in cooling-heating mode
    Fig. 6. Temperature trend of reference grating in cooling-heating mode
    Compensation result of dynamic model in cooling-heating mode. (a) Absolute wavelength drift; (b) prediction error
    Fig. 7. Compensation result of dynamic model in cooling-heating mode. (a) Absolute wavelength drift; (b) prediction error
    Compensation result of dynamic model in cooling mode (n=0, n=1). (a) Absolute wavelength drift; (b) prediction error
    Fig. 8. Compensation result of dynamic model in cooling mode (n=0, n=1). (a) Absolute wavelength drift; (b) prediction error
    Compensation result of dynamic model in cooling mode (n=2, n=3). (a) Absolute wavelength drift; (b) prediction error
    Fig. 9. Compensation result of dynamic model in cooling mode (n=2, n=3). (a) Absolute wavelength drift; (b) prediction error
    Compensation result of dynamic model in cooling-heating mode (n=0, n=1). (a) Absolute wavelength drift; (b) prediction error
    Fig. 10. Compensation result of dynamic model in cooling-heating mode (n=0, n=1). (a) Absolute wavelength drift; (b) prediction error
    Compensation result of dynamic model in cooling-heating mode (n=2, n=3). (a) Absolute wavelength drift; (b) prediction error
    Fig. 11. Compensation result of dynamic model in cooling-heating mode (n=2, n=3). (a) Absolute wavelength drift; (b) prediction error
    FBG No.0123
    Wavelength /nm1528.83931541.06211557.34601562.1832
    Table 1. Characteristic wavelengths of FBG
    nSCoolingCooling-heating
    MAXE /pmRMSE /pmCPU time /msMAXE /pmRMSE /pmCPU time /ms
    0039.1214.9418177.0258.17179
    1FBG033.5015.5017941.7025.33178
    FBG130.8412.6637.6820.16
    FBG226.0611.4333.3718.07
    2FBG0 and FBG18.024.9918129.0712.96180
    FBG0 and FBG25.012.6627.4212.04
    FBG1 and FBG23.792.0122.8811.83
    3FBG0,FBG1,and FBG22.531.301828.785.24182
    Table 2. Influence of location of FBG3 on model performance
    nSCoolingCooling-heating
    MAXE /pmRMSE /pmCPU time /msMAXE /pmRMSE /pmCPU time /ms
    0033.6513.7518069.2549.51181
    1FBG022.479.6017838.3825.36179
    FBG120.049.5031.0021.50
    FBG315.075.9029.0321.28
    2FBG0 and FBG112.713.7717918.689.14180
    FBG0 and FBG311.563.1415.288.89
    FBG1 and FBG37.712.7213.978.21
    3FBG0,FBG1 and FBG33.631.061817.843.33179
    Table 3. Influence of location of FBG2 on model performance
    nLwCoolingCooling-heating
    MAXE /pmRMSE /pmCPU time /msMAXE /pmRMSE /pmCPU time /ms
    05062.1032.75181106.2773.22181
    8048.8523.2780.5463.80
    10033.6513.7569.2549.51
    15050.5829.0517892.7560.57180
    8045.1622.7666.0945.36
    10022.479.6038.3825.36
    25042.6621.6918384.0955.89181
    8035.6019.6235.2621.60
    10012.713.7718.689.14
    35037.3514.1918276.7850.29182
    8027.6413.4519.939.51
    1003.631.067.843.33
    Table 4. Model performance under different window lengths and number of reference gratings
    Wenjuan Sheng, Haitao Lou, Gangding Peng. Dynamic Compensation of Tunable Filter Demodulation Error Based on Least Squares Support Vector Machine and Multi-Reference Gratings[J]. Acta Optica Sinica, 2023, 43(7): 0706003
    Download Citation