Application of Dynamic Neural Network Model for Magneto- Coulombic Attitude Control of Earth-Pointing Satellite

Abstract

An active control of the satellite attitude using magneto-Coulombic actuators is proposed along the circular orbit. Equation for the required charges on the Coulomb shells is derived, which is used to find the available control torque along the three body axes. It is proved that the eigenvalues of the averaged control matrix of the magneto-Coulombic system in the body reference frame for different orientations and angular velocities components will converge to the same values as in the case of the eigenvalues of the averaged control matrix in the orbital reference frame. A dynamic neural network based control law is adapted and trained for the proposed system. The neural network based dynamic neural network controller has the advantages of handling the non-linearity, adaptation and on-board training, and robust to large changes in the parameters in the plant, such as the moments of inertia. It is shown that the stabilization accuracy of the dynamic neural controller is higher as compared to the proportional-derivative based controller.