Robust shape optimization using artificial neural networks based surrogate modeling for an aircraft wing

Abstract

Aerodynamic shape optimization is a very active area of research that faces the challenges of highly demanding Computational Fluid Dynamics (CFD) problems, optimization with Partial Differential Equations (PDEs) as constraints, and the appropriate treatment of uncertainties. This includes the development of robust design methodologies that are computationally efficient while maintaining the desired level of accuracy in the optimization process. This paper addresses aerodynamic shape optimization problems involving uncertain operating conditions. After a review of possible approaches to account for uncertainties, an Artificial Neural Network (ANN) model is used to approximate the aerodynamic coefficients when the operating conditions vary. Robust optimization problem-solving approaches based on deterministic measurements are used, inspired by the work of Deb [Deb K., Gupta H. Introducing robustness in multi-objective optimization. KanGAL Report 2004–2016, Kanpur Genetic Algorithms Laboratory, Indian Institute of Technology, Kanpur, India (2004)]. The first procedure is a direct extension of a technique used for single-objective optimization. The second is a more practical approach allowing a user to define the desired degree of robustness in a problem. These approaches have been tested and validated in the case of the optimization of an aircraft wing profile in the transonic regime considering two uncertain variables: the Mach number and the angle of incidence.