关键词: 四旋翼无人机, 执行器故障, 在线优化, 跟踪控制, 容错控制

Abstract: The failure of the rotor makes the quadrotor unmanned aerial vehicles (UAVs) exhibit strong nonlinearity, fast time-varying and severe underactuation characteristics. Traditional control methods are difficult to maintain its flight stability and are highly likely to cause the UAV to lose stability and crash. Therefore, this paper proposes a multi-mode fault-tolerant control method based on the idea of hierarchical adaptation. This method innovatively establishes a hierarchical control framework consisting of dynamic weighting nonlinear model predictive control (DWNMPC) and adaptive incremental nonlinear dynamic Inversion (AINDI). The upper-layer DWNMPC controller devises a state-dependent weight adaptation mechanism. By dynamically adjusting the weights of each state in the cost function according to the attitude error of the UAV in real-time, it prioritizes maintaining attitude stability at the instant of failure to suppress rolling. After the system stabilizes, it smoothly transitions to the precise trajectory tracking task. To cope with model uncertainties and strong aerodynamic disturbances under severe failure conditions, the lower-layer AINDI controller is designed to perform online robust and adaptive corrections on the DWNMPC commands. This controller utilizes sensor measurements to compensate for unmodeled torques in real time. Additionally, it employs the recursive least squares (RLS) method with a forgetting factor to identify crucial parameters such as the moment of inertia online, thereby significantly enhancing the robustness of the system. Experimental results indicate that the proposed hierarchical adaptive fault-tolerant control method demonstrates good trajectory tracking capability under conditions of no fault, partial failure of a single rotor, and complete failure of the rotor. The control process relies solely on the UAV's onboard sensors for state estimation, highlighting its high versatility in actual physical environments. In the extreme case where a single rotor completely fails, causing the aircraft to spin at a high speed of approximately-10.5 rad/s, the root mean square error of trajectory tracking increased by only 0.0476 meters, 0.0545 meters, and 0.083 meters along the x, y, and z axes respectively compared to the no-fault condition, significantly enhancing the reliability of the UAV.

Key words: Quadrotor unmanned aerial vehicle, Actuator failure, Online optimization, Tracking control, Fault-tolerant control