Research On SVC Spacecraft Attitude Control Based On ADABOOST

With the development of the space technology, spacecraft as the base of science research is applied in more complex missions. Due to the unique workspace and structure, the system of spacecraft is usually dealing with many issues may influence the spacecraft attitude control such as the gravity gradient,solar radiation pressure,parameter perturbations and unmodeled dynamics. With the complex workspace and external disturbances, the spacecraft attitude control system is required to have the capabilities of processing them all to make sure all kinds of missions done with accuracy. It is researched that how to designed the appropriate attitude control system which would work in the complex space environment.It is begun with the mathematic modeling of spacecraft system. Assuming that the spacecraft is the structure of rigid body, hybrid coordinate method is applied in the modeling of spacecraft system. When the modeling is done, it is vital necessary to simplify it into the linear one for the benefit of the attitude control system design.Then, the method of sliding mode variable structure control(VSC) is introduced and applied in the spacecraft attitude control system. VSC is a discontinuous control method. It is applied in the spacecraft attitude control system for its strong robustness over parameter perturbations and external disturbances. It is verified that VSC is capable of the application in the spacecraft attitude control system through simulation. But some other issues are needed to be solved. Dealt with the parameter perturbations,the adaptive VSC method is a good solution to enhance the robustness for that the complex structure of spacecraft may cause the parameter perturbations. As for the chattering, the switch function is replaced by the hyperbolic tangent function to reduce the VSC chattering. It is vital necessary to combine the machine learning method and the VSC for the accuracy of spacecraft movability in complex space environment.After that, the ADABOOST method is introduced. Later it is introduced that how the basic theory of decision trees chooses the optimal features in the training dataset which is applied in the ADABOOST method as the basic classifier. Due to the mass amount of calculation of the weight, MODEST ADABOOST is advanced in the base of ADABOOST. It is verified that MODEST ADABOOST method is able to achieve a faster convergence and a lower training error through simulation.Finally, the MODEST ADABOOST method and the adaptive SVC method are combined to design the proper control law for the spacecraft dynamic mode. Some parameters which make the influence on the speed and precision of convergence are inferred form the Lyapunov function. These parameters are processed with MODEST ADABOOST to get the right classifier which could classify depending on the condition of the system. The classifiers are applied in the control law. It is verified that the adaptive VSC applied with MODEST ADABOOST could work in both rapidity and stability which makes the spacecraft follow the instructions with the external disturbance in a good state of control.