Abstract
The use of statistical techniques to build approximations of expensive computer analysis codes pervades much of today's engineering design. These statistical approximations, or metamodels, are used to replace the actual expensive computer analyses, facilitating multidisciplinary, multiobjective optimization and concept exploration. In this paper, we review several of these techniques, including design of experiments, response surface methodology, Taguchi methods, neural networks, inductive learning and kriging. We survey their existing application in engineering design, and then address the dangers of applying traditional statistical techniques to approximate deterministic computer analysis codes. We conclude with recommendations for the appropriate use of statistical approximation techniques in given situations, and how common pitfalls can be avoided.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 129-150 |
| Number of pages | 22 |
| Journal | Engineering with Computers |
| Volume | 17 |
| Issue number | 2 |
| DOIs | |
| State | Published - Jan 1 2001 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Software
- Modeling and Simulation
- Engineering(all)
- Computer Science Applications
Cite this
}
Metamodels for computer-based engineering design : Survey and recommendations. / Simpson, T. W.; Peplinski, J. D.; Koch, P. N.; Allen, J. K.
In: Engineering with Computers, Vol. 17, No. 2, 01.01.2001, p. 129-150.Research output: Contribution to journal › Review article
TY - JOUR
T1 - Metamodels for computer-based engineering design
T2 - Survey and recommendations
AU - Simpson, T. W.
AU - Peplinski, J. D.
AU - Koch, P. N.
AU - Allen, J. K.
PY - 2001/1/1
Y1 - 2001/1/1
N2 - The use of statistical techniques to build approximations of expensive computer analysis codes pervades much of today's engineering design. These statistical approximations, or metamodels, are used to replace the actual expensive computer analyses, facilitating multidisciplinary, multiobjective optimization and concept exploration. In this paper, we review several of these techniques, including design of experiments, response surface methodology, Taguchi methods, neural networks, inductive learning and kriging. We survey their existing application in engineering design, and then address the dangers of applying traditional statistical techniques to approximate deterministic computer analysis codes. We conclude with recommendations for the appropriate use of statistical approximation techniques in given situations, and how common pitfalls can be avoided.
AB - The use of statistical techniques to build approximations of expensive computer analysis codes pervades much of today's engineering design. These statistical approximations, or metamodels, are used to replace the actual expensive computer analyses, facilitating multidisciplinary, multiobjective optimization and concept exploration. In this paper, we review several of these techniques, including design of experiments, response surface methodology, Taguchi methods, neural networks, inductive learning and kriging. We survey their existing application in engineering design, and then address the dangers of applying traditional statistical techniques to approximate deterministic computer analysis codes. We conclude with recommendations for the appropriate use of statistical approximation techniques in given situations, and how common pitfalls can be avoided.
UR - http://www.scopus.com/inward/record.url?scp=0034933620&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0034933620&partnerID=8YFLogxK
U2 - 10.1007/PL00007198
DO - 10.1007/PL00007198
M3 - Review article
AN - SCOPUS:0034933620
VL - 17
SP - 129
EP - 150
JO - Engineering with Computers
JF - Engineering with Computers
SN - 0177-0667
IS - 2
ER -