An enhanced anti-disturbance finite-time exponentially convergent parameter estimator for quadrotor UAVs with unknown...

This article presents an enhanced anti-disturbance finite-time exponentially convergent parameter estimator for quadrotor UAVs with unknown uncertain dynamics, reconstructing unavailable states and unknown uncertain parameters simultaneously, thereby enabling the formation of a closed-loop control scheme. Its construction process consists of three steps. First, the mathematical model of the quadrotor UAV is divided into two loop subsystems, the position and attitude loop, which are then reparameterized to form a state-affine system. Second, a linear regression equation is constructed by using the generalized parameter estimation-based observer (GPEBO) technique, which reformulates the state observer task as a parameter observer. Third, the GPEBO and the dynamic regressor extension and mixing estimator are invoked to ensure finite-time exponential convergence without imposing persistent excitation. In addition, a forgetting factor over time is incorporated to track possible varying parameters. Subsequently, replacing the reconstructed information into the designed controller yields a closed-loop control scheme that will be rigorously proven using Lyapunov theory. Finally, simulation results based on data from an online database (www.flyeval.com) are presented, demonstrating that the proposed scheme improves trajectory tracking accuracy by 22% and 58.94% compared to schemes based on an extended state observer and a classical gradient estimator, respectively. The developed observer can be extended to a large class of state-affine systems.