A robust and fast active fault tolerant control for quadrotor applications with actuator fault’s unknown statistical characteristics

Abstract

Fault tolerant control (FTC) methods considering known faults for quadrotor applications are presented; however, unknown statistical faults characteristics particularly for the FTC design is not discussed. This study tackles this problem and presents a robust two-stage unscented Kalman filter (RTSUKF) that is robust against unknown faults and is able to accurately estimate multi-actuator faults with unknown statistical characteristics. The estimated unknown faults are then compensated through second-order sliding mode controllers which are designed for each of the translational and rotational movements. The fast convergence speed of the proposed estimator reduces the instability time interval of the quadrotor after fault occurrence. To discuss the reliability of the proposed method, stability analysis based on the Lyapunov theory is carried out for the whole closed-loop system. Finally, the robustness of the proposed FTC structure against unknown faults with its convergence speed are validated through numerical analysis. The results confirm the effective and good performance of the proposed method compared to another FTC with a conventional two-stage unscented Kalman filter.