Neural network based adaptive control with Hierarchy-Structured Dynamic Inversion applied to nonlinear aircraft model

Japan Aerospace Exploration Agency plans a flight demonstration project for a super sonic transport. A typical supersonic transport has a larger flight envelope than conventional one, so that the flight controller is required to be more flexible to the flight environment. Therefore, the adaptive control method is a promising candidate for these flight control system. In this paper, we propose an adaptive control method which is comprised of neural networks and Hierarchical-Structured Dynamic Inversion. Its structure is clearly comprehensible to the users and numerical simulation results show that proposed method is effective even with model errors.