Flight Control Law Synthesis Using Neural Network Theory.

摘要

A commonly used technique for advanced fighter aircraft control law development involves a lengthy process of linearizing the aircraft model and calculating many control system gains via conventional linear methods. This process must be repeated for a number of trim points within the flight envelope to achieve the aircraft stability and flying qualities mandated by various military specifications. Neural networks have been used extensively in many applications such as pattern recognition and optimization because of their ability to create nonlinear mappings of continuous valued inputs through supervised learning. This report outlines a concept which incorporates emerging neural network technology with present-day control theory to produce a system by which optimal controller gains can be automatically generated. The research completed to date and the results contained in this report are intended to provide a proof of concept by applying the neural network synthesis technique to some simplified linear and nonlinear examples.