Surrogate model
From Wikipedia, the free encyclopedia
Most engineering design problems require experiments and/or simulations to evaluate design objective and constraint functions as function of design variables. Design optimization some times requires thousands or even millions of evaluations, and often the associated high cost and time requirements render this infeasible. One can alleviate the computational cost of optimization by constructing algebraic expressions, known as surrogate models, which approximate the actual function(s), based on a limited amount of data. When only a single design variable is involved, the process is known as curve fitting as illustrated in the Figure. The process comprises of three major steps: sampling, construction of the surrogate model, and appraisal of the accuracy of surrogate.
The accuracy of the surrogate depends on the number and location of samples (expensive experiments or simulations) in design space. Various design of experiments (DOE) techniques cater to different sources of errors, in particular errors due to noise in data or errors due to improper surrogate model.
Selection of appropriate surrogate model to represent the actual function has significant bearing on the accuracy of predictions. The most popular surrogate models are polynomial response surfaces, Kriging and radial basis function. For most problems, the nature of true function is not known a priori so it is not clear which surrogate model will be most accurate. In addition, there is no consensus on how to obtain the most reliable estimates of the accuracy of a given surrogate.
[edit] See also
[edit] References
- Queipo, N.V., Haftka, R.T., Shyy, W., Goel, T., Vaidyanathan, R., Tucker, P.K. (2005), “Surrogate-based analysis and optimization,” Progress in Aerospace Sciences, 41, 1-28.
