28.02.2014 Views

The Development of Neural Network Based System Identification ...

The Development of Neural Network Based System Identification ...

The Development of Neural Network Based System Identification ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

24 CHAPTER 2 LITERATURE REVIEW<br />

Apart from high fidelity modelling theories from standard full scaled helicopter,<br />

several minimum complexity mathematical models can also be used. Several good<br />

examples <strong>of</strong> the development <strong>of</strong> minimum complexity mathematical model are given in<br />

Shim [2000], Bisgaard [2007] and Heffley and Mnich [1988]. Although such approaches<br />

can be employed to build a simulation model, the highly non-linear aerodynamics<br />

interaction between body components and the behaviour <strong>of</strong> the high order dynamics <strong>of</strong><br />

a rotorcraft is usually hard to model using the first principle approach (direct physical<br />

understanding <strong>of</strong> forces and moments balance <strong>of</strong> the vehicle) and such an approach can<br />

be inaccurate [Mettler, 2003, Budiyono et al., 2009, Deng et al., 2011]. A significant effort<br />

is necessary to validate the theoretical model and re-evaluate any potential shortcomings<br />

associated with the model.<br />

Most <strong>of</strong> the works adopting the first principle modelling approach in helicopter<br />

modelling use the developed model for control design without detailed validation against<br />

flight data [Kendoul, 2012, Bisgaard, 2007]. Among the few successful works that<br />

use first principle modelling to address the problem <strong>of</strong> physical parameter estimation<br />

with experimental validation are Nonami et al. [2010] and Gavrilets et al. [2001]. In<br />

Nonami et al. [2010], the non-linear helicopter dynamics is linearised to obtain two<br />

linear state equations that describe the helicopter’s lateral and longitudinal motion.<br />

<strong>The</strong> linear analytical model includes the rigid-body dynamics, main rotor dynamics<br />

and aerodynamics, stabiliser bar dynamics and aerodynamics effect from other main<br />

body components. <strong>The</strong> parameters <strong>of</strong> the model are determined using direct physical<br />

measurements and manufacturer specifications for different helicopter platforms. For<br />

parameters that are more difficult to obtain such as moment inertia in different axes,<br />

the values was measured roughly and manually retuned to match the flight data. <strong>The</strong><br />

linear models have been validated for three different helicopter UAV platforms and<br />

results show good fit with the flight data for operating condition around hovering flight.<br />

Gavrilets et al. [2001] describes 17 states <strong>of</strong> the non-linear dynamics model <strong>of</strong> an<br />

acrobatic helicopter UAV which includes states for longitudinal and lateral main rotor<br />

flapping, the rotor speed and an integral <strong>of</strong> the rotor-speed tracking error in addition<br />

to rigid body dynamics. Flight test experiments were used to estimate several key

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!