Adaptive Nonsingular Terminal Sliding Mode With Event-Triggered Nonlinear Model Predictive Control for Quadrotors Under Actuator Faults and Failures
Abstract
In this article, fault-tolerant trajectory tracking problem for quadrotor unmanned aerial vehicles (UAVs) under actuator saturations and control allocations is studied in the presence of actuator faults, disturbances, and uncertainties. Traditional approaches for fault tolerant control designs either involve sliding mode control or model predictive control separately. Such designs do not ensure finite-time fault tolerance and explicit constraints along with the control allocation simultaneously. As a result, in this work, a unified fault-tolerant two-layer control design for a multi-input quadrotor UAV subject to fault, disturbances, and uncertainty is developed. In this proposed two-layer design, adaptive nonsingular terminal sliding mode control provides finite-time robust fault-tolerance capability, while event-triggered nonlinear model predictive control (ET–NMPC) manages control allocation with taking account explicit actuator constraints. Using strengths of both controllers, this design approach has achieved capabilities that cannot be done using any of the methods alone. Modeling of the quadrotor is carried out using the nonlinear control-affine model with inclusion of loss-of-effectiveness faults, complete actuator failures, additive faults, parametric uncertainties, and external disturbances. Nonsingular terminal sliding mode manifold with matched fault and parametric uncertainties estimator utilizing projections is considered in order to ensure finite-time fault tolerance and bounded estimation. Meanwhile, the ET–NMPC controller supervises and carries out the control allocation if and when needed. Thus, it is possible to save computational effort by doing only what is needed to be done. Finite-time stability analysis proves this feature mathematically in relation with energy consumption. Simulations of the proposed scheme demonstrate its superiority in comparison to some typical fault tolerant control techniques providing better accuracy and less chattering, reducing almost by 43.8% control energy consumption as well as 56 times NMPC optimization computations.