SYSTEM IDENTIFICATION OF A MULTI-ROTOR VEHICLE WITH ACTIVE FEEDBACK CONTROL

Abstract

As multi-rotor vehicles become integrated into the national airspace for applications such as package delivery and videography, it is important that the inner-loop control system be robust and able to meet ever-demanding performance constraints. To achieve high bandwidth control designs, it is necessary to have accurate and high bandwidth open-loop models. The material covered in this dissertation was aimed towards the flight-based identification and comprehensive understanding of the open-loop dynamics and aerodynamics of a multi-rotor vehicle despite the active feedback control system. The analytical first principles modeling and the discussion of system identification techniques informed the process for identifying the unstable multi-rotor dynamics. The inherently unstable nature of the vehicle, combined with the available measurements, necessitated a methodical approach towards identification from flight experiments. The non-negligible issue of unstable system identification was addressed by both the experiment design and the applied estimation technique. An individual propulsion system was tested on a hover stand to understand both the static and dynamic behavior of the combined propeller, motor, and electronic speed controller system. The results of the ground experiments influenced the design of the flight experiments, the postulation of the models, and the understanding of the estimated model parameters. The methodology for flight-based multi-rotor system identification involved the combined application of manual pilot inputs and automated multi-sine inputs added to the output of the controller. A series of five main flight tests were conducted for the identification of a high bandwidth open-loop linear rigid body model and two aerodynamic models of a multi-rotor vehicle. The comparison of the linear rigid body model with the two aerodynamics models emphasized the clear and substantial importance of the nonlinear rigid body coupling terms even in operations near hover. The relative improvement from the linear model to the application of the aerodynamics models to the nonlinear rigid body dynamics was considerable for the translational dynamics, yet the linear model was similarly applicable to the rotational dynamics in aggressive flight near hover.