Abstract

Developing accurate and realistic models for Unmanned Aerial Vehicles (UAVs) is a central task in effective controller design, autopilot design, and simulation model validation. System identification methods have been extensively used as reliable and less expensive alternatives for conventional analytical modeling for large-scale aircraft in the past. Yet, there is limited work on the identification of mathematical models for small-scale unmanned helicopters. This thesis focuses on development of a system identification tool for rotary-wing UAVs based on frequency-domain non-parametric and parametric identification methods. The tool, which is designed to be embedded in the computer simulation software available for a UAV platform, employs nonlinear parameter estimation and optimization techniques with the purpose of predicting dominant dynamics of the UAV from measured responses and controls. The real flight data acquired from the testbed have been used for testing and verifying the developed system identification tool. The testbed is a commercially available radio-controlled helicopter, Trex-700, equipped with MP2128G2Heli MicroPilot autopilot, and the flight tests are conducted by MicroPilot in hover regime to excite attitude dynamics of the vehicle. The identification results using the developed tool are validated with CIFER framework which is a highly reliable tool in aircraft system identification. The results demonstrate excellent prediction capability of the developed tool for model identification of the testing UAV platform.