Adaptive sliding-mode composite disturbance-rejection control for variable-mass sloshing dynamics in photovoltaic cleaning UAVs

Abstract

This paper addresses multi-source composite disturbances in photovoltaic cleaning UAVs during continuous operation, including fluid-structure sloshing in the tank, time-varying mass decay, and near-wall unsteady aerodynamic interference. We propose an adaptive sliding-mode composite disturbance rejection strategy with physical-prior boundary constraints. Traditional controllers based on constant-rigid-body assumptions often induce trajectory overshoot or even instability under such conditions because they cannot effectively compensate the coupled effects of transient fluid impacts and mass loss. Therefore, we first establish a six-degree-of-freedom variable-mass nonlinear UAV dynamic model that includes additional Coriolis-related terms. On this basis, a collaborative disturbance-rejection architecture integrating intelligent sensing and online estimation modules is proposed: On one hand, a nonlinear disturbance observer (NDO) and a robust projection-based identification law are designed to achieve online decoupling and compensation of external low-frequency gusts and internal parameter drift; on the other hand, sloshing-force extrema extracted from fluid-structure simulations are used as dynamic constraint thresholds to construct a constrained adaptive sliding-mode controller (ASMC) with anti-chattering characteristics. Lyapunov theory proves the uniformly ultimately bounded (UUB) property of closed-loop states. Numerical simulations show that the proposed strategy confines roll-angle deviation within 0.15 rad under strong liquid-surface agitation and achieves zero-error altitude tracking with steady-state error below ±0.015 m during monotonic mass decay induced by continuous discharge. In addition, the physical-prior mechanism suppresses actuator chattering and saturation caused by high-frequency switching at the control source. The proposed architecture combines low computational cost with strong robustness, providing effective theoretical and technical support for engineering deployment of special liquid-carrying UAVs.