INVESTIGATION OF MULTIVARIABLE CONTROL METHODS FOR ATTITUDE CONTROL OF AN ARTIFICIAL SATELLITE

Abstract. In this paper are investigated multivariable control methods for attitude control of an artificial satellite consisting of a rigid body and two flexible panels. The investigated techniques are Linear Quadratic Regulator (LQR) method, Linear Quadratic Gaussian (LQG) method and H–Infinity method. The satellite modeling was built following the Lagrangean approach and the discretization was done using the assumed-modes method. The equations of motion obtained were written in its modal state space form. In LQR method, the control law shows a good performance, however the method is only applied for system with absence of disturbances and where all the states are available. In reality that does not happen. On the other hand, the LQG method is more realistic, because nor all the states are available and the system presents noises. However, the performance of the system decays due to presence of the Kalman filter. The disadvantage of both the methods is the absence of a systematic procedure in the choice of weights matrix Q and R, and noises w and v. The H ∞ method has a distinct systematic with respect to the other two methods here applied. In comparison with LQR and LQG the H ∞ results in this work had been superior. However its great disadvantage is in the need of great ability and necessary experience to build the weights that are associated to the performance of the method H∞