Spacecraft attitude control using neuro-fuzzy approximation of the optimal controllers

•A neuro-fuzzy controller was developed for spacecraft attitude control.•The controller approximates the State-Dependent Riccati Equation controller.•An optimal feedback controller can be designed through the approximation.•The controller also mitigates the large computational load.•The stability and the performance of the controller are analyzed.

In this study, a neuro-fuzzy controller (NFC) was developed for spacecraft attitude control to mitigate large computational load of the state-dependent Riccati equation (SDRE) controller. The NFC was developed by training a neuro-fuzzy network to approximate the SDRE controller. The stability of the NFC was numerically verified using a Lyapunov-based method, and the performance of the controller was analyzed in terms of approximation ability, steady-state error, cost, and execution time. The simulations and test results indicate that the developed NFC efficiently approximates the SDRE controller, with asymptotic stability in a bounded region of angular velocity encompassing the operational range of rapid-attitude maneuvers. In addition, it was shown that an approximated optimal feedback controller can be designed successfully through neuro-fuzzy approximation of the optimal open-loop controller.