CUED Publications database

A multi-objective multi-fidelity framework for global optimization

Kontogiannis, SG and Savill, AM and Kipouros, T (2017) A multi-objective multi-fidelity framework for global optimization. In: UNSPECIFIED.

Full text not available from this repository.

Abstract

© 2017, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved. Advances in optimization and numerical analysis methods as well as recent developments in Multidisciplinary Design Optimization, have made aerodynamic shape optimization of a single element aerofoil a well-established problem. However, even such a simple optimization study should be performed with minimum computational cost, while providing global optimality. A modular and robust framework is needed, especially for providing mature design suggestions at an early design stage. This paper demonstrates a Multiobjective optimization framework with rapid design space exploration capabilities in which a Multifidelity approach is directly adjusted to the emerging needs of the design. The same framework is intended later to be used for Multidisciplinary Design Optimization, so an effort has been made to accelerate it as far as possible by deploying Surrogate Based Optimization methods in the form of both Radial Basis Function and Kriging models. In particular, a modification of the standard Kriging approach to account for Multifidelity data inputs has been adopted which aims to increase accuracy and reduce overall optimization costs. This uses a Multifidelity sampling which trains an error corrector metamodel and the surrogate optimization problem is solved by a Multiobjective Particle Swarm Optimization algorithm. The resulting methodology is demonstrated using 1D and 2D test cases that also serve to tune the constructed surrogates. The well-established RAE2822 transonic aerofoil test case is then examined, with Lift and Drag coefficients as objectives. Results indicate that the Kriging modification allows better design space representation while improving the infill sampling location predicted for attaining fast global convergence. Moreover, a reduction in the required elapsed time is also observed.

Item Type: Conference or Workshop Item (UNSPECIFIED)
Subjects: UNSPECIFIED
Divisions: Div C > Engineering Design
Depositing User: Cron Job
Date Deposited: 17 Jul 2017 19:26
Last Modified: 08 Aug 2017 01:52
DOI: