Low cost PSO using metamodels and inexact pre-evaluation: Application to aerodynamic shape design

Modern optimization methods like Genetic Algorithms (GAs) and Particle Swarm Optimization (PSO) have been found to be very robust and general for solving engineering design problems. They require the use of large population size and may suffer from slow convergence. Both of these lead to large number of function evaluations which can significantly increase the computational cost. This is especially so in view of the increasing use of costly high-fidelity analysis tools like Computational Fluid Dynamics (CFD). Metamodels also known as surrogate models, are a cheaper alternative to costly analysis tools. In this work we construct radial basis function approximations and use them in conjunction with particle swarm optimization in an inexact pre-evaluation procedure for aerodynamic design. We show that the use of mixed evaluations by metamodels/CFD can significantly reduce the computational cost of PSO while yielding optimal designs as good as those obtained with the costly evaluation tool.