Adaptive Dynamic Programming-based Fault Tolerant Control for a Quadrotor UAV Attitude Stabilization

This paper addresses an attitude stabilization control problem for a quadrotor UAV under actuator faults via adaptive dynamic programming. First, a Hamilton-Jacobi-Bellman equation is solved by utilizing an adaptive critical neural network (NN) structure in the absence of actuator faults, and an approximate optimal control policy is derived. Meanwhile, a novel NN weight update law is proposed based on the Bellman error upper bound function, which not only has simple structure but also relieve the persistent excitation condition. Then, an adaptive law-based feedforward compensator is proposed to restrain actuator faults. Moreover, the ultimately uniformly bounded stability of the closed-loop system is proved by the development of Lyapunov methodology. Finally, the numerical simulation results verify the feasibility and effectiveness of the proposed control scheme.