## The Use of Metamodeling Techniques for Optimization Under Uncertainty (2001)

Venue: | 2001 ASME Design Automation Conference, Paper No. DAC21039 |

Citations: | 10 - 2 self |

### BibTeX

@INPROCEEDINGS{Jin01theuse,

author = {Ruichen Jin and Xiaoping Du and Wei Chen},

title = {The Use of Metamodeling Techniques for Optimization Under Uncertainty},

booktitle = {2001 ASME Design Automation Conference, Paper No. DAC21039},

year = {2001},

pages = {99--116}

}

### OpenURL

### Abstract

Metamodeling techniques have been widely used in engineering design to improve the efficiency in simulation and optimization of design systems that involve computationally expensive simulation programs. Many existing applications are restricted to deterministic optimization. Very few studies have been conducted on studying the accuracy of using metamodels for optimization under uncertainty. In this paper, using a twobar structure system design as an example, various metamodeling techniques are tested for different formulations of optimization under uncertainty. Observations are made on the applicability and accuracy of these techniques, the impact of sample size, and the optimization performance when different formulations are used to incorporate uncertainty. Some important issues for applying metamodels to optimization under uncertainty are discussed.

### Citations

417 |
A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code, Technometrics
- McKay, Beckman, et al.
- 1979
(Show Context)
Citation Context ...s +8 corner points +1 center point, totally 73 points. The corner points are created using 2-level full factorial design for three variables. The inner points are chosen using Latin Hypercube design (=-=McKay, et al., 1979-=-) where the points are so selected that they are uniformly distributed for each dimension. Once the metamodels are created using polynomial regression, Kriging, and radial basis function, additional c... |

320 |
Design and Analysis of Computer Experiments
- Sacks, Weltch, et al.
(Show Context)
Citation Context ...s metamodeling techniques for optimization under uncertainty. Three representative techniques, the polynomial regression (PR) (Box, et al.; 1978; Myers and Montgomery, 1995), the Kriging method (KG) (=-=Sacks, et al., 1989-=-; Booker, et al., 1999), and the method of radial basis function (RBF) (Hardy, 1971; Dyn, et al., 1986) are investigated. These methods are tested for different types of feasibility and objective func... |

312 |
Response Surface Methodology: Process and Product Optimization Using Designed Experiments
- Montgomery, Raymond
- 2002
(Show Context)
Citation Context ...r is to study the applicability and accuracy of various metamodeling techniques for optimization under uncertainty. Three representative techniques, the polynomial regression (PR) (Box, et al.; 1978; =-=Myers and Montgomery, 1995-=-), the Kriging method (KG) (Sacks, et al., 1989; Booker, et al., 1999), and the method of radial basis function (RBF) (Hardy, 1971; Dyn, et al., 1986) are investigated. These methods are tested for di... |

309 | Statistics for experimenters - Box, Hunter, et al. - 1978 |

263 |
Radial basis functions for multivariable interpolation: A review. IMA Conference on Algorithms for the approximation of Functions and Data (pp
- Powell
- 1985
(Show Context)
Citation Context ... the expression and x0i is the observed input. Radial basis function approximations have been shown to produce good fits to arbitrary contours of both deterministic and stochastic response functions (=-=Powell, 1987-=-). 3. THE EXAMPLE PROBLEM 3.1 Problem Description A two-bar structure design problem is used as an example in our study. The two-bar design example is adopted from Geilen (Geilen, 1986). The purpose o... |

186 |
Quality Engineering using Robust Design
- Phadke
- 1989
(Show Context)
Citation Context ...y proposed by Japanese engineer G. Taguchi (Taguchi, 1993), is a method for improving the quality of a product through minimizing the effect of the causes of variation without eliminating the causes (=-=Phadke, 1989-=-). When using a nonlinear programming formulation for robust design (see, e.g., Parkinson et al., 1993; Sundaresan et al., 1993; Chen et al.1996; Su and Renaud, 1997), a robust design solution is obta... |

163 |
Multiquadric equations of topography and other irregular surfaces
- Hardy
- 1971
(Show Context)
Citation Context ...es, the polynomial regression (PR) (Box, et al.; 1978; Myers and Montgomery, 1995), the Kriging method (KG) (Sacks, et al., 1989; Booker, et al., 1999), and the method of radial basis function (RBF) (=-=Hardy, 1971-=-; Dyn, et al., 1986) are investigated. These methods are tested for different types of feasibility and objective function formulations in optimization models with the consideration of uncertainty. Thr... |

112 |
Statistical tools for simulation practitioners
- Kleijnen
- 1987
(Show Context)
Citation Context ... on simulation codes for the purpose of design optimization. A preferable strategy is to utilize approximation models which are often referred to as metamodels as they provide a “model of the model” (=-=Kleijnen, 1987-=-) to replace the expensive simulation model. A comprehensive review of metamodeling applications in mechanical and aerospace systems can be found in (Simpson, et al., 1997). A comparative study of var... |

52 | Comparative studies of metamodelling techniques under multiple modelling criteria - Jin, Chen, et al. - 2001 |

46 | A Framework For Decision-Based Engineering Design - Hazelrigg - 1988 |

44 |
Numerical procedures for surface fitting of scattered data by radial functions
- Dyn, Levin, et al.
- 1986
(Show Context)
Citation Context ...omial regression (PR) (Box, et al.; 1978; Myers and Montgomery, 1995), the Kriging method (KG) (Sacks, et al., 1989; Booker, et al., 1999), and the method of radial basis function (RBF) (Hardy, 1971; =-=Dyn, et al., 1986-=-) are investigated. These methods are tested for different types of feasibility and objective function formulations in optimization models with the consideration of uncertainty. Through our investigat... |

40 |
A General Approach for Robust Optimal Design
- Parkinson, Sorensen, et al.
- 1993
(Show Context)
Citation Context ...ality of a product through minimizing the effect of the causes of variation without eliminating the causes (Phadke, 1989). When using a nonlinear programming formulation for robust design (see, e.g., =-=Parkinson et al., 1993-=-; Sundaresan et al., 1993; Chen et al.1996; Su and Renaud, 1997), a robust design solution is obtained by making the tradeoff between "optimizing the mean performance µy" and "minimizing the performan... |

32 | Towards a better understanding of modeling feasibility robustness in engineering design - P, Chen - 2000 |

29 |
Taguchi on robust technology development, bringing quality engineering upstream
- Taguchi
- 1992
(Show Context)
Citation Context ...may be infeasible or over-conservative. Many approaches exist in the literature to incorporate uncertainty in a design formulation. Robust design, originally proposed by Japanese engineer G. Taguchi (=-=Taguchi, 1993-=-), is a method for improving the quality of a product through minimizing the effect of the causes of variation without eliminating the causes (Phadke, 1989). When using a nonlinear programming formula... |

20 |
A Robust Optimization
- Sunderasan, Ishii, et al.
- 1993
(Show Context)
Citation Context ...gh minimizing the effect of the causes of variation without eliminating the causes (Phadke, 1989). When using a nonlinear programming formulation for robust design (see, e.g., Parkinson et al., 1993; =-=Sundaresan et al., 1993-=-; Chen et al.1996; Su and Renaud, 1997), a robust design solution is obtained by making the tradeoff between "optimizing the mean performance µy" and "minimizing the performance variance σy 2 ". To ca... |

15 |
A procedure for Robust Design
- Chen, Allen, et al.
- 1996
(Show Context)
Citation Context ..., most of these applications are restricted to deterministic optimization. Some researchers have documented the use of metamodels in robust design applications where system variations are considered (=-=Chen et al. 1996-=-; Padmanabhan and Batill, 2000). Nevertheless, very few studies have been conducted on studying the accuracy of this approach for optimization under uncertainty. As one can predict that the accuracy f... |

15 |
Automatic Differentiation in Robust Optimization
- Su, Renaud
- 1997
(Show Context)
Citation Context ...variation without eliminating the causes (Phadke, 1989). When using a nonlinear programming formulation for robust design (see, e.g., Parkinson et al., 1993; Sundaresan et al., 1993; Chen et al.1996; =-=Su and Renaud, 1997-=-), a robust design solution is obtained by making the tradeoff between "optimizing the mean performance µy" and "minimizing the performance variance σy 2 ". To capture the risk attitude of designers u... |

10 | Reliability-Based Design
- Rao
- 1992
(Show Context)
Citation Context ...uncertainty is ensured by making the probability of the constraints being satisfied being greater than the user specified probability, a very similar concept as that used in reliability-based design (=-=Rao 1992-=-). In the work of Du and Chen (2000), several feasibility-modeling techniques are examined. 2.2 Optimization Formulations Under Uncertainty By examining the optimization formulations under uncertainty... |

3 | An Iterative Concurrent Subspace Robust Design Framework
- Padmanabhan, Batill
- 2000
(Show Context)
Citation Context ...plications are restricted to deterministic optimization. Some researchers have documented the use of metamodels in robust design applications where system variations are considered (Chen et al. 1996; =-=Padmanabhan and Batill, 2000-=-). Nevertheless, very few studies have been conducted on studying the accuracy of this approach for optimization under uncertainty. As one can predict that the accuracy for deterministic design optimi... |

1 |
Sensitivity Analysis at Optimal Point," Studienarbeit, UNJ-GH Siegen, Institut fur Mechanik und Regelungstechnik Hazelrigg
- Geilen
- 1986
(Show Context)
Citation Context ...onse functions (Powell, 1987). 3. THE EXAMPLE PROBLEM 3.1 Problem Description A two-bar structure design problem is used as an example in our study. The two-bar design example is adopted from Geilen (=-=Geilen, 1986-=-). The purpose of this design problem is to find the values of two variables, i.e., d, nominal diameter of the cross section, and H, the height of the two bar structure, which can minimize the total v... |