Parameter Identification and Adaptive Control of a Fixed-Wing UAV in the Longitudinal Mode

Abstract

This project aims to use parameter identification to construct state space models for an aircraft’s dynamical behavior given control input and state output histories. Improving upon traditional methods for physics-based dynamical modelling, a data-based system construction approach could circumvent modelling error, track time-variant system changes, and bypass the time and cost intensive aircraft modelling process. To this end, this project necessitates the building of a feedback control loop with low latency and high accuracy. This control loop can then be used to conduct flight tests for controllers designed using both traditional physics-based models and the novel data-based model presented in this project. Comparisons between the traditional and novel system modelling approaches reveal multiple performance benefits of the data-based parameter identification approach. Discussions of these benefits, as well as limitations and areas for future study are also presented.