|dc.description.abstract||This work employs a data-driven, system identification technique to determine the governing
dynamic equations of a nonlinear, 6-DOF aircraft model. A generalized aircraft model
is developed, and nonlinearities are introduced in the gravity model, the aerodynamic force and
moment models, and the coupled dynamic and kinematic models. As a case study, the constant
parameters within the generalized model are chosen to closely match those of the SIAI Marchetti
S-211, an Italian jet trainer. The model is trimmed and, to establish confidence, the dynamics are
excited using both nonzero initial conditions and control inputs.
The nonlinear system identification technique, SINDYc (Sparse Identification of Nonlinear
Dynamics with Control), is formally introduced, a thorough explanation is provided, and a simple
example is conducted. SINDYc is a sparse regression technique that uses a library of candidate
functions, composed of state and input variables, to determine the fewest number of terms required
to represent the set of nonlinear differential equations which govern dynamic system behavior.
Using the three control inputs -- aileron, elevator, and rudder -- the S-211 model is aggressively
excited as to sufficiently express its nonlinearities, and the resulting time histories of each
state are recorded. This data is used with the SINDYc algorithm to reconstruct the nonlinear dynamic
equations. Several iterations are performed with variations in the type of state noise and
filtering, the method of numerical differentiation, and constraints imposed upon the library of candidate
functions. In most cases, SINDYc is able to determine the terms present in the nonlinear
dynamic equations with reasonable accuracy.
Finally, the identified systems are excited using simple control inputs, and their dynamic
response is compared to that of the true system.||