Modeling the unstable DelftaCopter vertical take-off and landing tailsitter unmanned air vehicle in hover and forward flight from flight test data

Abstract

The DelftaCopter is a tilt-body tailsitter unmanned air vehicle which combines a large swashplate controlled helicopter rotor with a biplane delta-wing. Previous research has shown that the large moment of inertia of the wing and fuselage significantly interacts with the dynamics of the rotor. While this rigid rotor cylinder dynamics model has allowed initial flight testing, part of the dynamics remains unexplained. In particular, higher frequency dynamics and the forward flight dynamics were not modeled. In this work, the cylinder dynamics model is compared with the tip-path plane model, which includes the steady-state flapping dynamics of the blades. The model is then extended to include the wing and elevon dynamics during forward flight. Flight test data consisting of excitations with a large frequency content are used to identify the model parameters using grey-box modeling. Since the DelftaCopter is unstable, flight tests can only be performed while at least a rate feedback controller is active. To reduce the influence of this active controller on the identification of the dynamics, one axis is identified at a time while white noise is introduced on all other axes. The tip-path plane model is shown to be much more accurate in reproducing the high-frequency attitude dynamics of the DelftaCopter. The significant rotor–wing interaction is shown to differ greatly from what is seen in traditional helicopter models. Finally, an Linear-Quadratic Regulator (LQR) controller based on the tip-path plane model is derived and tested to validate its applicability. Modeling the attitude dynamics of the unstable DelftaCopter from flight test data has been shown to be possible even in the presence of the unavoidable baseline controller.