Model-Free Predictive Fault-Tolerant Control for Spacecraft Roto-Translational Relative Motion

The problem of jointly controlling relative attitude and position of spacecraft in the presence of actuator fault is investigated in this paper. Following a description about drawbacks and limitations of the existing models and the control approaches based on them, a new formulation of the spacecraft relative motion is provided. Subsequently, the subspace predictive control framework, which is a powerful model-free approach, is extended in several dimensions, that is, adaptive nonlinear control, tolerance against abrupt faults and control allocation. Based on this generalized framework, three distinct data-driven fault-tolerant controllers for coupled, nonlinear and time-variant plants are developed. From the viewpoint of fault diagnosis, the only requirement of the control structure is to detect the occurrence time of faults. Furthermore, an internal data-driven fault diagnosis capability is introduced, which makes the control structure completely self-sufficient. The three controllers are then designed to solve the aforementioned problem, and their efficiency are verified via a multidimensional simulation scenario.