Roll Attitude Estimation with Moving Horizon Estimator and Unscented Kalman Filter

The real-time roll attitude estimation algorithm based on the nonlinear moving horizon estimator and unscented Kalman filter is proposed to solve the inconsistent and time-varying sampling problems of multi sensors on the high-spin projectile. Firstly, the traditional estimator is transformed to optimize the new captured roll velocity value instead of the start one of the moving horizon window. Secondly, UKF is applied to estimate the roll position value based on the AD sampling data from the geomagnetic sensors, which can effectively decrease the influence of the situations of sampling data loss, electrical disturbance, and so on. Finally, the whole roll attitude algorithm is simplified based on the trajectory characteristics without precision decrease. The estimation accuracy is compared with the fight test data, and it is proved that the proposed estimation method has the advantages of the high accuracy, good real-time performance, no accumulated error and strong robustness.