A Learning-Based Fault Tolerant Tracking Control of an Unmanned Quadrotor Helicopter

This paper presents a novel learning-based fault tolerant tracking control approach by using an extended Kalman filter (EKF) to optimize a Mamdani fuzzy state-feedback tracking controller. First, a robust state-feedback tracking controller is designed as the baseline controller to guarantee the expected system performance in the fault-free condition. Then, the EKF is employed to regulate the shape of membership functions and rules of fuzzy controller to adapt with the working conditions automatically after the occurrence of actuator faults. Next, based on the modified fuzzy membership functions and rules, the baseline controller is readjusted to properly compensate the adverse effects of actuator faults and asymptotically stabilize the closed-loop system. Finally, in order to verify the effectiveness of the proposed method, several groups of numerical simulations are carried out by comparing the performance of a tracking control scheme and the presented technique. Simulation results demonstrate that the proposed method is effective for optimizing the fuzzy tracking controller on-line and counteracting the side effects of actuator faults, and the control performance is significantly improved as well.