Optimal tracking neuro-controller in satellite attitude control

In the paper, a control strategy for optimal attitude tracking control of a multivariable satellite system has been presented. The approach is based on radial basis function neural network (RBFNN) and classical PD controller for its initial stabilization. It is shown how the network can be employed as a multivariable self-organizing and self learning controller in conjunction with a PD controller for attitude control of a satellite. By using four reaction wheels and quaternion for kinematics representation, the attitude dynamics of the satellite has been presented. In contrast to the previous classical approaches, it is shown how this controller can be carried out in an on-line manner with adaptive learning capabilities. The problem of robustness, accuracy and speed of response has all been addressed and through an example, the applicability of the proposed control scheme is illustrated.