Abstract
Abstract The requirements, the design and particularly the identification of a nonlinear continuous-time aircraft model to be used in a deterministic observer will be shown in this paper. The model is to be used in a LUENBERGER-observer that computes the aircraft velocity vector (in polar coordinates) from inertial data, provided by a laser-strap-down platform. Derived from an aircraft model using six degrees of freedom, the observer model is reduced to third order (local observer). Thereby, insufficiently known parts of the model are replaced by precisely measured data and the computing time is reduced. The identification of the model and its parameters is achieved by two main steps according to the Estimation - Before - Modelling Technique. First, the measurement errors are identified in a data - correction step by a Maximum Likelihood - identification using the kinematic part of the aircraft model. The remaining parameters (especially aerodynamic coefficients and derivatives) are estimated in a second step by a Least Squares - optimizer using the force equations. For the identification of the aircraft model and the verification of the observer, flight - test data from the twin engine aircraft Dornier DO128 are used. As shown in this paper, the identified model is a very good approximation to the real system, and the resulting observer is a good instrument for sensor fault detection and analytic redundancy.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call