Robust Adaptive Filter for Small Satellite Attitude Estimation.

The attitude kinematic equations and measurement equations are the function of the orbit parameters, so the standard Extended Kaiman Filter (EKF) can't give optimal attitude estimation because of the orbit errors, especially for a magnetometer-based small satellite. A robust adaptive Kaiman filter was designed to deal with the disturbance from the orbit estimation errors. The principle of robust estimation was adopted for the adaptive filters, and the covariance matrix and observational noise were adjusted to get proper Kaiman gains, and then the approximately optimal filtering results were obtained. Based on EKF stable theory, RAKF was proved uniformly and stochastically stable if the system is asymptotically controllable and asymptotically observable. Simulation results indicate that RAKF can improve attitude estimation precision from 0.3° to 0.2°, and the filter is approved to be more effective and robust for attitude estimation. [ABSTRACT FROM AUTHOR]