Towards neural adaptive hovering control of helicopters

This paper presents the results of a preliminary study on the use of neural adaptive control techniques which have the potential for implementation in hover control in a helicopter. For this hover study a 'laboratory-scale helicopter simulator' was employed. This constrained the system to one degree of freedom. The neural adaptive controller is designed by combining the dynamic response from a desirable reference model with a radial basis function (RBF) neural network. The performance of the control system is compared with classical approaches of PI controllers. The proposed neural controller system is shown to be an improvement over classical PI controllers.