In this paper, a neural controller with disturbance observer is presented, in order to control a quadrotor Unmanned Aerial Vehicle (UAV) subject to aerodynamic disturbances. As a novel feature, the controller is formulated in a way to be implemented directly to a second-order system, such as the one of the quadrotor. Feedforward neural networks are employed in the control system design to compensate for internal disturbances, while the external disturbances and approximation error of the neural network are estimated by a disturbance observer. Moreover, composite learning is used to improve the overall performance, by estimating the state variables in real-time and using the estimation error in the updating rules of both controller and the disturbance observer. An accurate disturbance modeling for the quadrotor is given, which considers wind and attitude changes, in order to evaluate the effectiveness of the controller. The controller successfully fulfills the task of trajectory tracking in the presence of wind and measurement noises, proving itself to be robust.
Quadrotor Composite Learning Neural Control with Disturbance Observer against Aerodynamic Disturbances
Castaldi, P
Ultimo
Conceptualization
2022
Abstract
In this paper, a neural controller with disturbance observer is presented, in order to control a quadrotor Unmanned Aerial Vehicle (UAV) subject to aerodynamic disturbances. As a novel feature, the controller is formulated in a way to be implemented directly to a second-order system, such as the one of the quadrotor. Feedforward neural networks are employed in the control system design to compensate for internal disturbances, while the external disturbances and approximation error of the neural network are estimated by a disturbance observer. Moreover, composite learning is used to improve the overall performance, by estimating the state variables in real-time and using the estimation error in the updating rules of both controller and the disturbance observer. An accurate disturbance modeling for the quadrotor is given, which considers wind and attitude changes, in order to evaluate the effectiveness of the controller. The controller successfully fulfills the task of trajectory tracking in the presence of wind and measurement noises, proving itself to be robust.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.