Finite-time fault-tolerant tracking control for a QUAV with mixed faults and external disturbances based on adaptive global fast terminal sliding mode neural network control method

This paper addresses the finite-time tracking control problem for a class of quadrotor unmanned aerial vehicle (QUAV) subject to unknown mixed faults and external disturbances. The considered mixed faults include both input quantization and actuator faults. First, radial basis function neural networks (RBFNNs) are employed to approximate the unknown nonlinear dynamics of the QUAV system, with adaptive control laws designed for online weights updates. Second, since the neural network approximation errors and external disturbances can be treated as unknown but bounded constants, adaptive control laws are developed to estimate these parameters. Third, to address the design complexity caused by unknown control coefficients arising from mixed faults, a Nussbaum gain function is introduced. Subsequently, based on the designed global fast terminal sliding mode (GFTSM) functions, adaptive GFTSM neural network control strategies are proposed for position and attitude tracking control. Theoretical analysis confirms that these control strategies guarantee the QUAV system's position and attitude outputs converge to reference trajectories, with tracking errors reaching a very small neighborhood of zero within a finite time. Finally, the effectiveness of proposed control strategies is validated through an actual system.