Project for an engineering course in advanced aircraft control. I chose to implement a Feedback Error Learning architecture enhanced with a RBF Neural Network, for the attitude control of a teetering helicopter. I developed a simplified 6 DoF model of a Bell AH-1 and the controller itself in MATLAB/Simulink.
The file parameters.m contains the specifics of the helicopter while the settings for the controller are inside FEL_RBFNN.m. The complete non-linear model is contained in non_lin_dyn_model.slx, while the analytical linearization of the attitude dynamics around the hover condition, which is used to derive the control law and also to test the efficacy of the neural network, is implemented inside lin_att_model.slx. The script results.m performs an extensive test of the controller in terms of trajectory tracking and robustness against external disturbancies.
See the report pdf for a detailed explanation.
Fsessini/heli_FEL_controller
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|