Neural network based nonlinear inverse dynamics for flight controller design

The equations of motion of modern fighter aircraft are highly nonlinear at high angles of attack and at high angular rates. Nonlinear controller design methods can therefore provide superior closed-loop performance by utilizing the nonlinear aircraft dynamics model. The paper presents a methodology for the design of a neural network based inverse dynamics controller for nonlinear aircraft dynamics. The purpose of the inverse dynamics controller is to linearize the aircraft dynamics across the flight envelope. The neural network based design methodology uses the nonlinear simulation of aircraft dynamics to design the controller.