Disturbance Observer Enhanced Neural Network LPV Control for a Blended-Wing-Body Large Aircraft

Abstract

The problem of trajectory tracking control for a blended-wing-body (BWB) large aircraft with model parameter uncertainties and unknown disturbances is considered. A linear parameter-varying (LPV) model is derived from the nonlinear dynamics of the BWB aircraft from which a robust linear parameter-varying controller is designed to track a desired trajectory. Using a single quadratic Lyapunov function (SQLF) and an infinite number of linear matrix inequalities to be evaluated at all vertices, a pair of positive definite symmetric matrix solutions are determined via Lyapunov stability theory and linear matrix inequality technique. Furthermore, a disturbance-observer is designed to process the unknown disturbances. Considering the plant exists some model errors except for disturbances, a radial basis function neural network approximation is embedded into the SQLF LPV controller to improve tracking performances, and a composite disturbance-observer based neural network single quadratic Lyapunov function (NNSQLF) controller can realize desired trajectory tracking of the linear parameter-varying system through regulating performance weighting functions. The closed-loop system of trajectory tracking control is proved to be asymptotically stable by using Lyapunov theory. Simulation results of forward flight speed and altitude tracking control of the BWB aircraft show that the proposed disturbance-observer based NNSQLF control can robustly stabilize the LPV system and precisely track the desired trajectory by comparing with conventional SQLF control and parameter-dependent Lyapunov function control, even in unknown exterior disturbances and model uncertainties.