Many engineering design problems need to take into account manufacturing tolerances. The goal is then to identify a solution that is robust in the sense of providing good expected performance despite possible disturbances to its design variables. In this chapter, we propose a Bayesian optimization algorithm to tackle this problem. We focus on the simple Direct Robustness Approximation idea from Le and Branke (Bayesian optimization searching for robust solutions. In: 2020 Winter Simulation Conference (WSC) (2020)) and make the following three contributions. First, we demonstrate that using Latin Hypercube Sampling to estimate solution quality improves the quality of solutions found. Second, we consider the case where there is disturbance already during the design stage. And finally, we examine the use of stochastic Kriging rather than the standard Gaussian process as the surrogate model.
Bayesian Optimization for Robust Solutions Under Uncertain Input
Space Technol.Proceedings
International Conference on Uncertainty Quantification & Optimisation ; 2020 ; Brussels, Belgium November 17, 2020 - November 20, 2020
Advances in Uncertainty Quantification and Optimization Under Uncertainty with Aerospace Applications ; Chapter : 15 ; 245-259
2021-07-16
15 pages
Article/Chapter (Book)
Electronic Resource
English
Bayesian Optimization for Robust Solutions Under Uncertain Input
TIBKAT | 2021
|Robust topology optimization of periodic structures under uncertain loading
British Library Online Contents | 2017
|