Airship control using neural network augmented model inversion

Abstract

Airship is a highly nonlinear plant which requires gain scheduled controller for various flight conditions. Model inversion control is augmented with neural network to control the airship. The input-output feedback linearization allows one to design a linear controller from a nonlinear system. Neural network is used to compensate the nonlinearities resulting from this linearization. The control technique is applied to two different airship flight conditions. The results shows that control technique is successful in controlling both conditions.