Time domain system identification of longitudinal dynamics of a UAV: A grey box approach

Abstract

System identification is very effective for aircraft modeling because many aircraft motions cannot be duplicated accurately using analytical methods. The identified model can represent an aircraft in all flight regimes and thus can be used for the development of flight simulators and automatic flight controllers. In this research, a unique three stage procedure is presented for time domain system identification of small scale fixed wing UAV. Flight experiment was conducted with specifically designed maneuvers for identification of the UAV's longitudinal dynamics. Initial reference model was developed using UAV's geometrical information in DATCOM. Recorded data, from flight tests, was processed in MATLAB system identification toolbox for estimating grey box aircraft models by applying Prediction Error Method. The model was iteratively improved through Adaptive Gauss Newton optimization. Model validation and error analysis were performed and the UAV's aerodynamic coefficients were determined. Excellent validation results show that the identified model can be used for various applications including the design of altitude and airspeed controllers of autopilot.