Use of data sampling, surrogate models, and numerical optimization in engineering design
Anthony Giunta · 2002
An engineering design study was performed using computational simulation software coupled with the DAKOTA (Design Analysis Kit for Optimization and Terascale Applications) software package. This study made use of the data sampling methods in DAKOTA, which included grid-based parameter studies and Latin hypercube sampling. Multidimensional surface fitting methods such as quadratic polynomial regression were used to smooth out numerical noise generated by the computational simulation software. This enabled the application of a surrogatebased optimization algorithm to solve the design problem. These results serve as a case study that demonstrates the utility of employing a combination of statistical methods and optimization methods in engineering design.