Kriging Algorithms

Kriging postulates a combination of a polynomial model plus departures of the form given by:

where is the unknown function of interest, is a polynomial function of , and is the realization of a normally distributed Gaussian random process with mean zero, variance , and non-zero covariance. The term in is similar to the polynomial model in a response surface and provides a "global" model of the design space.

While "globally" approximates the design space, creates "localized" deviations so that the Kriging model interpolates the sample data points. The covariance matrix of is given by:

where is the correlation matrix and is the spatial correlation of the function between any two of the sample points and . is an symmetric, positive definite matrix with ones along the diagonal. The correlation function is a Gaussian correlation function:

The in are the unknown parameters used to fit the model, is the number of design variables, and and are the components of sample points and . In some cases, using a single correlation parameter gives sufficiently good results. You can specify the use of a single correlation parameter or one correlation parameter for each design variable in the Options dialog box. Select Tools > Options > Design Exploration > Response Surface and then make the desired section for Kernal Variation Type in the Kriging Options category.

can be written:


Release 18.2 - © ANSYS, Inc. All rights reserved.