Detail publikace

Artificial Neural Networks for Surrogate-based Optimization in Preliminary Aerodynamic Design

Originální název

Artificial Neural Networks for Surrogate-based Optimization in Preliminary Aerodynamic Design

Anglický název

Artificial Neural Networks for Surrogate-based Optimization in Preliminary Aerodynamic Design

Jazyk

en

Originální abstrakt

A preliminary aerodynamic design often imposes requirements on global optimum search within a large, highly multimodal design space. Tools typically deployed to evaluate individual design candidates are very computationally expensive, being part of the finite volume computational fluid dynamics class. This virtually prevents deployment of traditional stochastic global optimization approaches, such as evolutionary algorithms. Hence, there has been a growing interest in metamodelling techniques, providing a reliable surrogate of the simulator response to an optimization algorithm. Efficient deployment of such techniques within preliminary aerodynamic design is of interest to Garteur Action Group 52 members. The present paper describes the involvement of Brno University of Technology, Institute of Aerospace Engineering in the AG52. The considered test case is based on the RAE2822 aerofoil constrained multipoint optimization problem. The overall problem setup is given along with selected surrogate modelling and optimization techniques. The presented approach featuring artificial neural networks is able to produce highly reliable metamodels with cutting-edge performance as documented by the AG52 performance metrics comparison.

Anglický abstrakt

A preliminary aerodynamic design often imposes requirements on global optimum search within a large, highly multimodal design space. Tools typically deployed to evaluate individual design candidates are very computationally expensive, being part of the finite volume computational fluid dynamics class. This virtually prevents deployment of traditional stochastic global optimization approaches, such as evolutionary algorithms. Hence, there has been a growing interest in metamodelling techniques, providing a reliable surrogate of the simulator response to an optimization algorithm. Efficient deployment of such techniques within preliminary aerodynamic design is of interest to Garteur Action Group 52 members. The present paper describes the involvement of Brno University of Technology, Institute of Aerospace Engineering in the AG52. The considered test case is based on the RAE2822 aerofoil constrained multipoint optimization problem. The overall problem setup is given along with selected surrogate modelling and optimization techniques. The presented approach featuring artificial neural networks is able to produce highly reliable metamodels with cutting-edge performance as documented by the AG52 performance metrics comparison.

BibTex


@inproceedings{BUT117342,
  author="Petr {Dvořák}",
  title="Artificial Neural Networks for Surrogate-based Optimization in Preliminary Aerodynamic Design",
  annote="A preliminary aerodynamic design often imposes requirements on global optimum search within a large, highly multimodal design space. Tools typically deployed to evaluate individual design candidates are very computationally expensive, being part of the finite volume computational fluid dynamics class. This virtually prevents deployment of traditional stochastic global optimization approaches, such as evolutionary algorithms. Hence, there has been a growing interest in metamodelling techniques, providing a reliable surrogate of the simulator response to an optimization algorithm. Efficient deployment of such techniques within preliminary aerodynamic design is of interest to Garteur Action Group 52 members. The present paper describes the involvement of Brno University of Technology, Institute of Aerospace Engineering in the AG52. The considered test case is based on the RAE2822 aerofoil constrained multipoint optimization problem. The overall problem setup is given along with selected surrogate modelling and optimization techniques. The presented approach featuring artificial neural networks is able to produce highly reliable metamodels with cutting-edge performance as documented by the AG52 performance metrics comparison.",
  address="University of Strathclyde",
  booktitle="Eurogen 2015 Extended Abstracts Book",
  chapter="117342",
  edition="ECCOMAS: European Community on Computational Methods in Applied Sciences",
  howpublished="print",
  institution="University of Strathclyde",
  year="2015",
  month="september",
  pages="28--34",
  publisher="University of Strathclyde",
  type="conference paper"
}