Nonlinear Dynamic Model Identification for Aircraft with Unknown Mass Properties Using Flight Data

Abstract

Development of a nonlinear dynamic model for fixed-wing aircraft generally requires accurate, labor-intensive mass properties estimation. This paper proposes a method to identify a nonlinear dynamic model for fixed-wing aircraft using flight data without requiring knowledge of the aircraft mass or moments of inertia. The standard nonlinear rigid-body aircraft equations of motion are reformulated and the nondimensional force and moment coefficients are redefined to be agnostic to vehicle mass properties. Flight experiments for system identification of a small, fixed-wing unmanned aircraft were conducted using orthogonal phase-optimized multisine inputs applied simultaneously to the aircraft control surfaces. A nonlinear aerodynamic model was then identified using the equation-error method in the frequency domain for both the proposed mass-agnostic modeling framework and a standard nonlinear modeling approach requiring mass properties information. The identified mass-agnostic nonlinear dynamic model and the traditional nonlinear dynamic model are shown to each provide a good fit to the modeling data and have good prediction capability of flight data not used for model identification in comparative flight simulations. The paper describes application of the modeling method to a small unmanned aircraft, but the method is generalizable to many aircraft configurations. The mass-agnostic dynamic modeling method will be particularly useful for modern electric aircraft that have limited mass properties changes over the course of a flight, where the identified model can be used for numerous controls and flight simulation applications.