Adaptive neuro-control for spacecraft attitude control

The spacecraft attitude control which combines the concepts of artificial neural networks and nonlinear adaptive control, is investigated as an alternative to linear control approaches. Two capabilities of neuro-controllers are demonstrated using a nonlinear model of the Space Station Freedom. These capabilities are: 1) synthesis of robust nonlinear controllers using neural networks; and 2) adaptively modifying neuro-controller characteristics for varying inertia characteristics. The main components of the adaptive neuro-controllers include an identification network and a controller network. Both these networks are trained using the backpropagation of error learning paradigm. To ensure robustness of the neuro-controller, an optimally connected neural network is synthesized for the identification network. For the online adaptive control problem, a new technique using a memory filter for error backpropagation is introduced. The performances of the nonlinear neuro-controllers for cases listed above are verified using a nonlinear simulation of the Space Station. Results presented substantiate the feasibility of using neural networks in robust nonlinear adaptive control of spacecraft