In this article, a L1 neural network adaptive fault-tolerant controller is exploited for an unmanned aerial vehicle attitude control system in presence of nonlinear uncertainties, such as system uncertainties, external disturbances, and actuator faults. A nonlinear dynamic inversion controller with sliding mode control law is designed as the outer-loop controller to track the attitude angles quickly and accurately which reduces dependence on model accuracy. A L1 neural network adaptive controller of the inner loop is introduced to compensate the nonlinear uncertainties and have a good attitude tracking. The radial basis function neural network technique is introduced to approximate a lumped nonlinear uncertainty and guarantee the stability and transient performance of the closed-loop system, instead of converting it to a half-time linear system by the parametric linearization method. Simulation results demonstrate the effectiveness of the proposed controller.

在本文中，L1神经网络自适应容错控制器被用于存在非线性不确定性（例如系统不确定性，外部干扰和执行器故障）的无人机姿态控制系统。将具有滑模控制律的非线性动态逆控制器设计为外环控制器，以快速，准确地跟踪姿态角，从而减少了对模型精度的依赖。引入了内环的L1神经网络自适应控制器，以补偿非线性不确定性并具有良好的姿态跟踪。引入了径向基函数神经网络技术来近似估计总的非线性不确定性，并确保闭环系统的稳定性和暂态性能，而不是通过参数线性化方法将其转换为半时线性系统。仿真结果证明了所提出控制器的有效性。