Distributed Cooperative Attitude Tracking Control For Multiple Spacecrafts Based On Sliding Mode/Neural Network

Spacecraft formation flying has huge technological advantages and broad application,and it has been a trend to use multiple spacecraft to replace the single spacecraft in missions.By having multiple simple rigid bodies working together, the formation control gains lots ofadvantages, such as low cost, good performance as well as high stability when performingsome accomplicated missions. For some applications, numerous rigid bodies should keeptheir attitudes consistent with each other in order to accomplish the given task. In this study,we investigate the distributed cooperative attitude tracking control problem for multiplespacecraft.Firstly, the sliding mode attiude control law is designed for spacecraft in the presenceof model uncertainty and external disturbance. To tackle with the slow system response,weak robustness in the reaching phase of sliding mode, an exponent time-varying slidingmode attitude control law is proposed. Meanwhile, the priori requirement for the upperbound of model uncertainty and external disturbance,which is difficult to be obtained, canbe removed by designing the adaptive exponent time-varying sliding mode, from which theupper bound can be real-time estimated.Then, it is well-known that time-varing sliding mode can only guaranteeasymptotically convergence, the singularity and slow convergence rate may encounterwhen using terminal sliding mode control law. This study will make a summaration aboutthe existing terminal sliding mode, and propose another kind of fast terminal sliding modemanifold, the spacecraft cooperative attitude tracking problem can be solved by designingthe control law based on the proposed fast terminal sliding mode manifold, and thesimulation results has proved the effectiveness of the proposed control law.Finally, considering that the mathematical model is indispensable for silding modecontrol, to remove the restriction, this psper propose neural network cooperative trackingcontrol law where the neural network is utilized to approximate the desire control torque.Furthermore, to speed up the system response convergence rate and avoid the controltorque saturation caused by the over-adaptation phenomenon, a modified adaptive law isproposed to revise the sliding mode manifold, by which the control torque saturationproblem can be improved, and the finite-time stability for the spacecraft formation flying can be also guaranteed. Meanwhile, the simulation results has demonstrated a strongrobustness and fault capability of the proposed control law.