System Identification of a Nonlinear Flight Dynamics Model for a Small, Fixed-Wing UAV

TR Number
Journal Title
Journal ISSN
Volume Title
Virginia Tech

This thesis describes the development of a nonlinear flight dynamics model for a small, fixed-wing unmanned aerial vehicle (UAV). Models developed for UAVs can be used for many applications including risk analysis, controls system design and flight simulators. Several challenges exist for system identification of small, low-cost aircraft including an increased sensitivity to atmospheric disturbances and decreased data quality from a cost-appropriate instrumentation system. These challenges result in difficulties in development of the model structure and parameter estimation. The small size may also limit the scope of flight test experiments and the consequent information content of the data from which the model is developed. Methods are presented to improve the accuracy of system identification which include data selection, data conditioning, incorporation of information from computational aerodynamics and synthesis of information from different flight test maneuvers. The final parameter estimation and uncertainty analysis was developed from the time domain formulation of the output-error method using the fully nonlinear aircraft equations of motion and a nonlinear aerodynamic model structure. The methods discussed increased the accuracy of parameter estimates and lowered the uncertainty in estimates compared to standard procedures for parameter estimation from flight test data. The significant contributions of this thesis are a detailed explanation of the entire system identification process tailored to the needs of a small UAV and incorporation of unique procedures to enhance identification results. This work may be used as a guide and list of recommendations for future system identification efforts of small, low-cost, minimally instrumented, fixed-wing UAVs.

Output Error Method, Unmanned Aerial Vehicle, Vortex Lattice Method, Flight Testing, Aerodynamic Modeling, Parameter Estimation