Approximate Modelling and Multi Objective Optimisation in Aeronautic Design

Abstract

An investigation of efficient approximation methods for computationally expensive objective functions in aeronautic multi-disciplinary design and multi objective optimisation is presented. Several approximation methods based on curve fitting using polynomials and artificial neural networks are considered. A comparison of these approximation methods in terms of the achieved quality and accuracy and the required computational cost will be presented. The approximation models have been successfully applied in a preliminary design and multi objective optimisation study of a blended wing body aircraft.