Abstract: An adaptive real-time estimate algorithm is proposed to address the problem of gyro-stabilized platform drift. First, the major work mode of the gyro-stabilized and tracking equipment is analyzed. Then, a gyro drift model under the stabilization mode and the track mode is established, which indicates that the gyro low frequency noise composed of the constant bias and related bias is the main cause of stabilization error. The algorithm adopts the Kalman filter framework and employs filter convergence criterion combining Sage-Husa filter and weighted Sage-Husa filter to estimate the constant and related drifts of gyros. Experiment results show that the convergence time is less than 3 s and the estimation standard deviation is less than 0.02 (?)/s under the condition of non-exact system model and noise characteristics, which demonstrates the robustness and adaptive ability of the algorithm.

Key words: gyro-stabilized platform, gyro drift, Kalman filter, Sage-Husa filter, adaptive filter