Neural network based feedback linearization control of an unmanned aerial vehicle

Abstract

In this thesis, a flight control design of an unmanned aerial vehicle (UAV) using neural network based feedback linearization and the output redefinition technique is presented. The UAV model we chose in this research is a nonlinear nonminimum phase system. The output redefinition technique is used in a way such that the resulting system is minimum phase and can be inverted. The nonlinear autoregressive moving average (NARMA-L2) neural network is trained off-line to identify the forward dynamics of the UAV model with the redefined output, and then inverted to force the real output to approximately track the desired trajectory. The results shows that a good tracking performance can be achieved using this control scheme.