Identification of longitudinal and lateral dynamics of an ultralight aircraft

Most aircraft can be considered as rigid bodies whose motion is determined by a set of forces due to aerodynamic effects, propulsion and gravity. Their dynamics can thus be described by computing the position and velocity of the center of gravity as well as the orientation and angular velocity of body-fixed axes with respect to a set of earth reference axes. The dynamic equations of motion are well–known so that aircraft modeling leads to very accurate results provided that an equally accurate knowledge of the aircraft parameters and of the acting forces is available. This information is, however, seldom available particularly for light or ultralight aircraft. The alternative to modeling is identification; the practical application of identification techniques is however conditioned by the choice of the class of models (often linear), the design of suitable data–gathering experiments and by the final validation of the obtained model i.e. by the evaluation of the degree of approximation of its description of the real process. This paper describes the identification of the longitudinal and lateral dynamics of an ultralight aircraft and shows that the consistency of the obtained descriptions can heavily depend on the considered class of models. In particular, it is shown that traditional equation-error approaches relying on ARX or those relying on Output Error models can prove unreliable while the less known ARX+noise models can give very consistent results.