A NEW APPROACH TO NONLINEAR MODELING OF HIGHLY MANEUVERABLE AIRCRAFT USING NEURAL NETWORKS

摘要

Artificial Neural networks offer viable solution to identification and modeling of aerospace dynamic systems. This paper proposes a new approach to the nonlinear modeling of agile aircraft which is applicable to develop flight simulators. In contrary to classical methods, neural-network-based modeling of aircraft dynamics does not require any aerodynamic or propulsion model and a few flight test measured data suffice. The obtained model is shown valid for arbitrary pilot inputs within a region of mach-altitude around the pre-specified flight condition.