Paper
Global Approximation using Adaptive Regressive Polynomial Response Surfaces
- Authors:
- Xiaoling Luo
- Abstract
- Global approximation substitute for the original model is also called response surface (RS), surrogate or meta-model. The key aspects should be considered when using the RS approximation method: the accuracy of the RS, the number of original model evaluations, the time consuming when updating the RS, the memory space occupied by the RS, and the speed of evaluation for a given point using the RS. This paper analyzes the drawbacks of the existing response surface methods, and then proposes an adaptive regressive polynomial response surfaces method using quadratic functions with domain decomposition. The test cases and applications effectively support the proposed method of this paper. As it is shown, the proposed method is far more effective and accurate.
- Keywords
- global approximation; polynomial response surface; domain decomposition; regressive; quadratic function; domain combination
- StartPage
- 52
- EndPage
- 60
- Doi