Adaptive Kalman filter based on random-weighting estimation for denoising the fiber-optic gyroscope drift signal

Reducing and suppressing the random noise and drift error is a critical task in an interferometric fiber-optic gyroscope (IFOG). In this paper, an improved adaptive Kalman filter (KF) based on innovation and random-weighting estimation (RWE) is proposed to denoise IFOG signals in both static and dynamic conditions. The covariance matrix of the innovation sequence is estimated using the random-weighted-average window. The KF gain is then adaptively updated by the estimated covariance matrix. To decrease the inertia of KF response in the dynamic condition, the covariance matrix of process noise is adjusted when discontinuous IFOG signals are detected by the innovation-based chi-square test method. The proposed algorithm is applied for denoising IFOG static and dynamic signals. Allan variance is used to evaluate the denoise performance for static signals. In the dynamic condition, root-mean-square error is considered as the performance indicator. Quantitative results reveal that the proposed algorithm is competitive for denoising IFOG signals when compared with conventional KF, RWE-based gain-adjusted adaptive KF, and RWE-based moving average double-factor adaptive KF.