Using genetic algorithms to generate D s -optimal response surface designs
This research introduces a new approach to the generation of D s -optimal designs for subsets of parameters of the second-order response surface model for 2, 3, and 4-dimensional hypercube design spaces while simultaneously satisfying a specified minimum value for the D-efficiency for the full model. Specifically, 1-point and 2-point exchange algorithms and a genetic algorithmwere developed to generate these designs. The results indicate that the D s -criterion values of the GA designs were greater than or equal to those of designs generated by exchange algorithms and by the Optex procedure in SAS. Thus, in general, the GA is superior to the exchange algorithms for generating subset optimal designs.
Keywords and phrasesExchange algorithm genetic algorithm optimal design response surface design subset optimality
Unable to display preview. Download preview PDF.
- 3.J. J. Borkowski, J. of Probability and Statistical Science 1, 65–88 (2003).Google Scholar
- 7.A. Heredia-Langner, W. M. Carlyle, D. C. Montgomery, C. M. Borror, and G. C. Runger, J. of Quality Technology 35, 28–46 (2003).Google Scholar
- 9.J. H. Holland, Adaptation in Natural and Artificial Systems (The University of Michigan Press, Ann Arbor, MI, 1975).Google Scholar
- 11.L. P. Khoo and C. H. Chen, Advanced Manufacturing Technology 1, 298–325 (2001).Google Scholar