State-Space Linear Models of Aircraft From Flight Testing Using OKID and Output Error Estimation Methods

System identification estimation methods such as Observer/Kalman filter identification algorithm (OKID) and Output Error (OE) are important techniques for generating linear time-invariant (LTI) models of air vehicles. Generating accurate models is essential for determining stability of the system and identifying dynamic modes. Models in the state-space representation in particular allow prediction of how the system will react to either commanded or exogenous inputs, and is especially useful for autopilot controller design. While OE is industry standard, OKID requires no a priori knowledge about the aircraft and therefore can be used when performing system identification on an unknown aircraft. In this research, both methods are used to generate LTI state-space models of an Anaconda fixed-wing aircraft from simulated and recorded flight test data, using Juang’s System/Observer/Controller Identification Toolbox (SOCIT) and Morelli’s System IDentification Programs for AirCraft (SIDPAC) toolbox respectively. A method for estimating vehicle moments of inertia from recorded flight test data was used to reduce the amount of a priori information needed for OE. The models generated from both simulated and recorded flight data and their respective errors are compared between the two methods. Data taken from flight testing had increased noise from exogenous inputs such as wind gusts, and it was a challenge to determine when to stop data collection after a given input due to this noise. From this testing, OKID was found to have similar precision and performance as OE while needing no a priori information. Potential limitations of OKID are that known values for specific parameters cannot be accounted for. This may make the overall model less accurate, as all parameters are estimated, unlike OE where only selected parameters are estimated. Future work includes determining the applicability of these methods when generating LTI state-space models for small unmanned rotorcraft.