Vibration Control of Satellite Antennas via NMPC and NARX Neural Networks

This article introduces an integrated approach for vibration control in satellite plate antennas by combining advanced control techniques with smart materials. The proposed system integrates bonded piezoelectric actuators and sensors with two complementary control strategies: nonlinear model-predictive control (NMPC) and a Nonlinear AutoRegressive with eXogenous inputs (NARX) neural network. A comprehensive coupled attitude and structural dynamics model is developed specifically for flexible plate-type antennas, providing the foundation for precise control implementation. The innovation lies in the synergistic combination of NMPC's predictive capabilities with NARX's adaptive learning features. While NMPC leverages the system model to optimize future control actions, the NARX neural network serves dual purposes: functioning as an independent controller and enhancing the system's state estimation accuracy. This hybrid approach addresses key limitations of traditional control methods, particularly in handling model uncertainties and external disturbances during spacecraft attitude maneuvers. MATLAB/Simulink simulations demonstrate that the integrated NMPC-NARX system significantly outperforms both stand-alone NMPC and NARX-based controllers in vibration suppression. The results show superior robustness to modeling inaccuracies and enhanced adaptability to dynamic disturbances, marking a significant advancement in satellite control system design. This research establishes a new paradigm for achieving improved stability and operational efficiency in satellite systems operating under uncertain conditions.