An enhanced anti-disturbance of quadrotor UAV via the adaptive reduced-order GPEBO

Abstract

This paper proposed an adaptive reduced-order generalized parameter estimation-based observer (GPEBO) to simultaneously reconstruct the unavailable states and unknown parameters for a quadrotor UAV to realize trajectory tracking control without velocity measurements. More specifically. Firstly, the model of a quadrotor UAV is built and divided into position and attitude loop subsystems. Secondly, a linear regression equation (LRE) is constructed using the GPEBO technique to transform the tasks of state estimation into parameter estimation, only utilizing the output information of the plant after the reparameterization of the system where a state affine nonlinear system. Then, the GPEBO technique is adopted again to reconstruct the parameters of LRE to obtain a new LRE that combines the dynamic regressor extension and mixing (DREM) estimator to generate a set of scalar LREs under extremely weak conditions, observability of the system, which are much weaker than persistent excitation (PE) conditions. In addition, a time-varying forgetting factor is incorporated into the developed observer to track possible time-varying parameters. Subsequently, the information that the developed observer estimates is introduced into the non-singular fast terminal sliding model controller (NFTSMC) based on complete information in an equivalent way to form a composite controller scheme. The stability proof of the closed-loop system will be rigorously proven via the Lyapunov stability theory. Finally, comparing the proposed control scheme with the classic active disturbance rejection control (ADRC), which is widely popular, an obvious merit is that the system output of the proposed scheme can quickly converge to the desired value, and the developed observer can also be extended to a broader class of state affine nonlinear systems.