Optimization of hydraulic shapes with application of CFD, genetic algorithm and meta-modelling
Author
Janiszkiewicz, Maciej Wojciech
Term
4. term
Education
Publication year
2020
Submitted on
2020-05-31
Pages
58
Abstract
Formålet med projektet er at vurdere, hvor godt en genetisk algoritme og en bestemt type metamodel, kaldet kriging, fungerer i en ingeniøroptimeringsopgave. En genetisk algoritme er en søgemetode inspireret af naturlig selektion, og kriging er en statistisk model, der kan efterligne dyre beregninger. Vi opbygger et automatiseret miljø til CFD-analyse (Computational Fluid Dynamics), som er computersimulering af strømning, og kobler det til flere optimeringsstrategier. Hovedopgaven er at finde en hensigtsmæssig indstilling af den genetiske algoritme, så der er en god balance mellem kvaliteten af de fundne løsninger og den tid, optimeringen tager. Til sidst sammenlignes ydeevnen af flere optimeringsteknikker.
The project evaluates how well a genetic algorithm and a specific type of metamodeling, called kriging, perform in an engineering optimization task. A genetic algorithm is a search method inspired by natural selection, and kriging is a statistical surrogate model that can approximate costly simulations. We build an automated environment for CFD analysis (computational fluid dynamics), meaning computer simulations of fluid flow, and connect it to several optimization strategies. The main goal is to tune the genetic algorithm so it strikes a good balance between the quality of the results and the time the optimization takes. Finally, we compare the performance of several optimization techniques.
[This abstract was generated with the help of AI]
Keywords
Documents