Performance analysis of linear and nonlinear control strategies for flexible space manipulators

The dynamics of a flexible space manipulator is highly coupled and complex, since it depends on orbital forces and perturbations, on orbital torques (gravity gradient and aerodynamic disturbance) and on the flexible behaviour of the links. A fully nonlinear model for a multibody system composed by a generic number of flexible links, suitable for an efficient numerical simulation, is detailed and implemented. Moreover, a simplified dynamics model is also developed, in order to be used in the design of the regulator for reducing computational cost. Dedicated control strategies are implemented, tested, and then compared in order to define their relevant performance, such as required total power, maximum torque level, maximum flexible oscillations, computational cost. Different approaches are followed to control the multibody system: (i) a proper nonlinear regulator (FLT -feedback linearization technique), (ii) a linear controller (LQR - linear quadratic regulator), (iii) a simple PD (proportional derivative) regulator. All control techniques are implemented and applied closing the loop of the complete non linear dynamics. Robustness of the controllers will be investigated by considering uncertainties regarding both the parameters describing external environment (target mass), and the multibody characteristics (actuators malfunctioning). Moreover, sensor noisy measurements are considered and estimated by means of an extended Kalman filter. As a result, a complete, realistic simulation tool for space multibody is realized.