Surrogates and Approximations in Engineering Design

Surrogates and Approximations in Engineering Design

The functional purpose and design constraints are frequently approximated in engineering design by connecting the design variables and the responses of interest at a few locations in the design space. Surrogates are commonly employed in design studies since they are approximations of the underlying functionalities.

This course will introduce such surrogates, as well as how to develop, assess, and apply them in design. Polynomial regression, kriging, and radial basis function will be discussed as surrogates, whereas Latin hypercube sampling and Hammersley sequence will be discussed as the design of experiments.

Surrogates and Approximations Course Outline

Week 1: Overview, physical vs. computational experiments, engineering optimization, and the use of surrogates in optimization

Week 2: Sampling plans, Latin squares, Latin hypercubes, stratification, orthogonal arrays, and Hammersley sequencesSurrogates in

Week 3: Polynomial Regression, Radial Basis Function, and Kriging

Week 4: Infill criteria, adaptive sampling, and the use of surrogates in design space exploration and exploitation