Artificial satellites attitude adaptive estimation using Kalman filter with dynamic model compensation

An optimal estimation algorithm is developed with the capability of real time applications under unfavorable conditions (acquisition phase, e.g.), when external torques are not completely known and have significant effect on attitude dynamics. To do so a procedure is used which combines extended Kalman filter with dynamical model compensation and adaptive state noise estimation techniques. The external torques (gravitational, magnetic, aerodynamic and solar radiation) are dynamically compensated in the estimator by a first order Gauss-Markov process and their global effect estimated together with the system state; simultaneously state noise is adaptively estimated to avoid divergence, by imposing statistical consistency residuals. The algorithm is tested with simulated data corresponding to a satellite with three axis motion and equipped with solar and Earth horizon sensors.