• Laser & Optoelectronics Progress
  • Vol. 59, Issue 1, 0106002 (2022)
Ning Mao, Jiangning Xu, Hongyang He*, and Miao Wu
Author Affiliations
  • College of Electrical Engineering, Naval University of Engineering, Wuhan , Hubei 430033, China
  • show less
    DOI: 10.3788/LOP202259.0106002 Cite this Article Set citation alerts
    Ning Mao, Jiangning Xu, Hongyang He, Miao Wu. Real-Time Compensation of Fiber Optic Gyroscope Zero-Drift Based on Online-SVR Model[J]. Laser & Optoelectronics Progress, 2022, 59(1): 0106002 Copy Citation Text show less
    Basic schematic diagram of SVR
    Fig. 1. Basic schematic diagram of SVR
    Principle of real-time acquisition of temperature change rate
    Fig. 2. Principle of real-time acquisition of temperature change rate
    Online compensation of FOG zero-bias
    Fig. 3. Online compensation of FOG zero-bias
    Original FOG output, temperature change rate, and temperature varying with time
    Fig. 4. Original FOG output, temperature change rate, and temperature varying with time
    Compensation effect of RBF neural networks. (a) Zero-drift compensation curve of gyro; (b) residual error
    Fig. 5. Compensation effect of RBF neural networks. (a) Zero-drift compensation curve of gyro; (b) residual error
    Compensation effect of SVR. (a) Zero-drift compensation curve of gyro; (b) residual error
    Fig. 6. Compensation effect of SVR. (a) Zero-drift compensation curve of gyro; (b) residual error
    Compensation effect of Online-SVR. (a) Zero-drift compensation curve of gyro; (b) residual error
    Fig. 7. Compensation effect of Online-SVR. (a) Zero-drift compensation curve of gyro; (b) residual error
    SchemeAverageStandard deviationMAERMSE
    Before compensation0.30821.1204
    RBF neural network0.08970.18880.17440.2091
    SVR-0.06430.02350.06430.0685
    Online-SVR-0.01730.02170.02530.0278
    Table 1. Comparison of statistical results before and after the compensation for FOG zero-drift data
    SchemeTime of all compensation pointsAverage time per compensation point
    RBF neural network58.273.9×10-3
    SVR15.801.1×10-3
    Online-SVR6.194.13×10-4
    Table 2. Compensation time
    SchemeNoise in angular random walk /[(°)h-1/2Noise in bias instability /[(°)h-1Noise in rate random walk /[(°)h-3/2
    Before compensation9.33×10-41.57×10-24.49×10-2
    RBF neural network1.61×10-34.46×10-34.41×10-2
    SVR1.83×10-42.90×10-31.71×10-2
    Online-SVR2.21×10-41.81×10-31.57×10-2
    Table 3. Comparison of error coefficients of noise before and after compensation
    Ning Mao, Jiangning Xu, Hongyang He, Miao Wu. Real-Time Compensation of Fiber Optic Gyroscope Zero-Drift Based on Online-SVR Model[J]. Laser & Optoelectronics Progress, 2022, 59(1): 0106002
    Download Citation