Research On Satellite Attitute Control System Parameter Identification And Recursive Control

As to long-life satellites, the gyros’ lifespan may be shorter than the lifespan ofsatellite. In order to minimize the dependence on gyros, a satellite attitude controlsystem which only calls for star sensor as attitude sensor is designed. This attitudecontrol system should give special consideration to the short time recoverable failure ofstar sensor. Fully taking advantage of the periodic characteristic of satellite in-orbitkinematics, an attitude recursive control system is promoted. Four aspects of contentswill be studied in order to realize this attitude recursive control system.The predictive filter algortithm and extended Kalman filter algorithm are used toestimate attitude angular rate for the satellite which only has star sensor as attitudesensor. The predictive filter which combined the advantage of Kalman filter and MMEestimator can calculate the model error in real-time. Extended Kalman filter algorithm isthe most common used algorithm as to nonlinear estimation problem. The performanceof predicitive filter and extended Kalman filter is compared with the most simple directdifference method. Then the best algorithm choice is selected for this situation. In orderto improve the accuracy of disturbance torque estimation, two in-orbit satellite inertiamatix estimation methods are promoted and compared.Most satellites have the payloads which have relative big rotating componentswhich are capable of produce big disturbance torque on the satellite. Two Stage Kalmanfilter is used to estimate this disturbance torque. Two Stage Kalman filter have a bigadvantage over the argumented Kalman filter whose calculation power is much bigger.Two Stage Kalman filter is derivated and simulated in the satellite attitude controlsystem.As the satellite attitude control system is stable, the amplitude of the noise in theattitude control commond torque which is produced by the controller is almost the samebig with the environmental disturbance torque. The traditional filter algorithm is invalidin this situation. In order to solve this problem, fully taking advantage of the extremelylow frequency characteristic of envirimental disturbance torque, a method is designedbased on the unkown input observer and the Fast Fourier Transform noise deduction.Compared with the traditional time-delay filter, this method does not have time delay.Then the high-precision satellite disturbance torque estimation and attituderecursive control model is derived. This model has much higher precision and is equallysimple. The attitude recursive control method is based on the fourth-order Runge-Kuttaintegration algorithm and estimated disturbance torque model. During the short timefailure of star sensor, the controller can still output attitude control command based on the attitude recursive method. The precision of this control system heavily depends onthe precision of identified disturbance torque model and inertial matrix.