Nonlinear Fault Recovery of a Quadrotor Using Progressive Accommodation

Abstract

This paper proposes a nonlinear (NL) fault-tolerant control (FTC) strategy for stabilizing the attitude of a quadrotor affected by multiple rotor failures. The recovery process relies mainly on progressive accommodation (PA) strategy. The latter uses the Newton-Raphson (NR) algorithm in order to solve the state-dependent Riccati Equation (SDRE) in a recursive way. Simulations are performed using the NL model of the quadrotor in fault-free and post-fault cases. Both PA and the direct SDRE methods are employed so that the attitude control of the quadrotor can be stabilized. The results show that the PA strategy performs better than the classical method when it comes to convergence speed and to preserving the control objectives of the nominal system.

Nonlinear Accommodation of a DC-8 Aircraft Affected by a Complete Loss of a Control Surface

“…One can notice that here, the control problem differs from that of the quadrotor (Mlayeh et al, 2020), where the major concern was the stability of the drone around an equilibrium point. However, for the aircraft, one is rather concerned with the stability of a motion, i.e, whether the aircraft can keep a specific desired trajectory, if perturbed away from it.…”

“…To overcome these problems, a few existing works from the specialized literature (Ghachem, Benothman & Benrejeb, 2013;Ghachem, Benotman & Benrejeb, 2017) used the NPA, but the application was limited to firstand second-order equations and never involved a real NL system. Lately, the NPA was used for the stabilization and the recovery process of a drone (Mlayeh et al, 2020;Mlayeh et al, 2021). A comparison between the NPA and the classical accommodation was presented.…”

“…Moreover, a unique linear-likeform of NL systems was considered to allow the application of the SDRE. Additionally, the control problem considered in the previous studies concerned the stability of the system around an equilibrium point (such as the drone case (Mlayeh et al, 2020)), but never a trajectory-tracking problem.…”