Nonlinear Dynamic Inversion Control for Aircraft with Disturbance Observer Based on Gaussian Process Regression

Abstract

Fixed-wing aircraft experience disturbances known as turbulence while operating in the air. This airborne turbulence applies unwanted forces and moments to the aircraft and disturbs the movement of the aircraft, interfering with the presumed performance prescribed by the controller. A Nonlinear Dynamic Inversion (NDI) control law is especially prone to degradation, proportional to model uncertainties caused by turbulence. To solve this issue, a Nonlinear Dynamic Inversion Controller based on the Gaussian Process Regression (GPR) is proposed as a disturbance observer. With the help of the Gaussian Process regression, model uncertainties from the turbulence are estimated for the feed-forward term, to aid the performance of the NDI control law. The Dryden model is one of many turbulence models that depict turbulence, in power spectral density form. This model was implemented in this paper to simulate turbulence, and the controller performance was evaluated.