INCREMENTAL NONLINEAR DYNAMIC CONTROL WITH SPARSE NON-PARAMETRIC BAYESIAN REGRESSION OF DAMAGED AIRCRAFT

Abstract

Unlike ground vehicles, fixed-wing aircraft that experiences structural damage faces more dangerous situations, even resulting in crash landing. Especially when the control law of the aircraft is based on the prior system model knowledge of the aircraft, it can encounter loss of control when damage is inflicted to the aircraft because the structural damage changes the parameters in the system model and control effectiveness matrix, causing divergence of the closed-loop system. This structural damage causes the aircraft to change its mass, center of gravity (CG), moment of inertia (MOI), aerodynamic coefficients and these problems impose challenges to control design as it must handle high uncertainties in the system and control effectiveness matrix. Two major approaches to deal with uncertainties are to use robust controller and estimate uncertainties which are calculated online to counter them. For this problem, we present an Incremental Nonlinear Dynamic Inversion (INDI) Control with Sparse on-line Gaussian Regression (SGPR) for controlling damaged aircraft. Incremental Dynamics are used to overcome disturbances in the system by using state derivative measurements. We applied Nonlinear Dynamic Inversion control with this Incremental Dynamics to profit from the robustness properties of this form. In addition to this control law, uncertainties are estimated using SGPR to add adaptiveness to the system; a non-parametric regression method. There are various situations of structural damage. In this study, severed single main wing tip situation is considered.