This paper presents an adaptive attitude control system using artificial neural networks for self-tuning of PID-like controllers using the Model Reference Adaptive Control (MRAC) approach. The proposed control method effectively compensates the influence of the uncertainties of the moment of inertia parameters and external disturbances in the agile satellite dynamics, also deals with rapidly change from any initial angle to any final angle in the three axis attitude maneuver. The gains of PID controller are set first using Ant Colony optimization method in off-line calculations and then fed to on-line training for Neural Networks (NN). The output from NN controls automatically the gains of the original PID controller. A comparison between the proposed system and existing control techniques with the same parameters under the same conditions was performed. The important features of the proposed system are demonstrated by carrying out a simulation model under MATLAB/ SIMULINK environment.
展开▼
