Research On Satellite Attitude Control Based On Data-driven Theory

This paper is based on Data-driven theory and takes satellite attitude control asbackground. Three controllers which could work well without the precise model ofthe controlled object are designed aiming at satellite attitude control system withnonlinearity, time-varying, parameter coupling, and random interference in thecontrol process. The simulations are performed in the Matlab/Simulinkenvironment.Firstly, the paper gives the kinetics and dynamics model of satellite based onEuler angle and quaternion and derives the kinetics equation based on the errorquaternion. Furthermore it gives the simulation framework of satellite attitudecontrol system.Secondly, a Mamdani type fuzzy controller is designed on the basis of fuzzylogic and the inputs is error quaternion and the output is the rotational speed offlywheel. The controller is used in the satellite attitude control simulation andcompared with the traditional PD controller, then gets the result from it.Thirdly, it is hard to determine the parameters of membership function. Inorder to solve this problem, an anfis controller is designed using neural network tofix and optimize the parameters. This paper summarizes the relation betweenMamdani type and Sugeno type fuzzy reasoning algorithm and performs thesimulation with the controller. The robustness of the controller is tested by changingthe moment of inertia and increasing interferential moment.Finally, the paper introduces a black-box control theory called MFAC anddesigns an attitude based on this theory. Then the paper describes the designprogress of the control algorithm and gives its stability and convergence conclusionunder certain assumptions. Explore the application prospects of this control methodin the complex satellite control and give a preliminary of the stability of thecontroller. The simulation results show the feasibility of model-free adaptivecontroller.