Adaptive control of DGMSCMG using dynamic inversion and neural networks

Abstract

The paper presents a nonlinear dynamic model for DGMSCMGs, sometimes used as actuators to control the attitude of large satellites. The developed models describe the translation dynamics, the rotation dynamics of the Active Magnetic Bearing Rotor, as well as the rotation dynamics of the two mobile gimbals. Two control architectures are initially designed by using the dynamic inversion concept, proportional-integrator-derivative/proportional-derivative dynamic compensators, linear observers, and a neural network to compensate the effect of the dynamic inversion error. One also develops a similar adaptive control architecture consisting of a proportional-integrator dynamic compensator, a feed-forward neural network, and a linear observer. The latter system models two interconnected nonlinear servo-systems and controls the angular rates of the two mobile gimbals actuated by the attitude controller of the satellite. The validation of the novel control architectures is achieved in Matlab/Simulink, the obtained results proving a very good angular rate precision and the robustness of the control systems in relation to the external disturbances.