Research On Spacecraft Integrated Orbit And Attitude Dynamics, Control And Navigation

The modelling, control and navigation of spacecraft’s orbit and attitude is importantissues in aerospace missions. Therefore, a huge number of studies and results have beenobtained during recent decades. However, there is a crucial problem which most missionsshould face: modelling the orbit and attitude dynamics seperately in most flying missionsis an inaccurate approximate method which should be checked all the time. As a result,this dissertation concentrates on integrated dynamics modelling mothed, integratedcontrol strategy, and integrated relative navigation approach. The major contents of thisdissertation consist of the following parts.For the integrated modeling of spacecraft orbit and attitude motion, the mathematictool dual quaternion is firstly used to describe the orbit motion and attitude motionsimultaneously, and the integrated kinematics model is derived accordingly. Thedynamics of a single spacecraft is then proposed in the framework of dual number, whichcan describe the relationship between spacecraft and external forces and torques. Basedon the research of single spacecraft, modelling and analysis of relative orbit and attitudeare studied. In particular, the coupling effect due to the point deviating from the center ofmass is taken into account, and the dynamics model of relative motion between thepoints deviating from the center of mass is established. For the tracking control problemof spacecraft, a linear sliding mode controller and a PD-like robust controller areproposed. Lyapunov method and Barbalat lemma are used respectively to prove thestability of the system in the presence of disturbances and model uncertainties. Finally,simulations are done to demonstrate the effectiveness of the proposed controllers.For the control problem of spacecraft relative orbit and attitude, a terminal slidingmode control strategy is designed based on the logarithm of dual quaternion. Thecontroller can realize finite-time tracking of desired attitude states while being robustagainst model uncertainties and disturbances. When the information of disturbances andmodel uncertainties are unavailable, an adaptive finite-time control approach is proposed.Furthermore, the Lagrange form of spacecraft’s dynamics model is derived, and a fastfinite-time control law is developed. The convergence rate of linear sliding mode,terminal sliding mode and fast sliding mode is compared and analyzed. Finally, byconsidering spacecraft rendezvous and docking mission, numerical simulations are doneto illustrate the effectiveness of the proposed control methods.For the estimation problem of relative position and attitude, single vision basednavigation approaches are proposed based on multiple geometric features. At first,feature point, line and circle are described in the frame of dual number. In particular, the description of feature circle is based on a kind of dynamic definition. By using the singlevision theory, the measurement models based on feature point, line and circle aredeveloped, respectively. Then, employing the measurement models, Levenberg-Marquardt algorithm, Extended Kalman Filter algorithm and Unscented Kalman Filteralgorithm are ultilized to estimate the relative position and attitude. Finally, by numericalsimulations of the aforementioned algorithms, the accuracies of all the algorithms arecompared, and the influence of number of feature points and their distribution onestimation accuracy is analized.For model uncertainties and maneuverabilities of target spacecraft, an improvedstrong tracking filter is proposed to estimate relative position and attitude. This methodpossesses strong robustness against model uncertainties and competent tracking ability ofsuddenly-changed system states. In particular, this method is also effective if themeasurement matrix is not full-rank. Considering the measurement failures of featureinformation, a single coefficient robust Unscented Kalman Filter and a multiplecoefficient robust Unscented Kalman Filter algorithms are developed to avoid or alleviatethe failure of filters, where the multiple coefficient method can vary its coefficientsaccording to various measurement states. Finally, simulations of these approaches arepresented to validate their effectiveness.