Helicopter adaptive flight control using neural networks

This paper summarizes the theoretical development and numerical investigation of a direct adaptive tracking control architecture using neural networks. Neural networks are proposed for use in combination with nonlinear controllers designed based on feedback linearization. Neural networks capable of online learning are required to compensate for inversion error which may arise from imperfect modeling, approximate inversion or sudden and unanticipated changes in dynamics. A stable weights adjustment rule for the online neural network is presented. Under mild assumptions on the nonlinearities representing the inversion error, the adaptation algorithm assures that all the signals in the loop are uniformly bounded and that the weights of the online neural network tend to constant values. Simulation results using an AH-64 6DOF simulation model are presented to illustrate the performance of the online neural network based adaptation algorithm