Integrating User-Preference Swarm Algorithm and Surrogate Modeling for Airfoil Design
Robert Carrese, Hadi Winarto, Xiaodong Li · 49th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition · 2011
Implementing evolutionary multi-objective optimization directly for high-fidelity aerodynamic shape design is computationally challenging. This is due to the excessive number of computational fluid dynamic simulations required to identify a host of Pareto-optimal solutions. Inexpensive surrogate models are therefore used co-operatively with the precise objective functions to alleviate the computational burden. Kriging metamodels predict the function values at unobserved locations, by recording precise evaluations in a training dataset. In this paper, we propose a novel Kriging-assisted multi-objective optimization algorithm (k-up-mopso). The optimizer is based on the particle swarm analogy, which utilizes an elitist archive to store the best representative Pareto front at each time-step. A user-preference module is integrated into the optimization framework, which guides the swarm towards preferred regions of the Pareto frontier, thereby focusing all computing effort on identifying only solutions of interest to the designer. The user-preferences are easily implemented within the optimization framework and can ideally be based on an existing or target design. Whilst providing a logical criterion to pre-screen candidates for precise evaluation, the additional guidance provided by user-preferences guarantees an accelerated convergence rate. The algorithm is applied to a typical transonic airfoil design scenario for robust aerodynamic performance. Data mining techniques are applied to visualize the design landscape and to illustrate the trade-offs between the final preferred designs.