| 학회 |
한국화학공학회 |
| 학술대회 |
2002년 봄 (04/26 ~ 04/27, 강원대학교) |
| 권호 |
8권 1호, p.365 |
| 발표분야 |
공정시스템 |
| 제목 |
오목한 함수의 차를 과소평가 함수로 이용한 새로운 전역 최적화 방법 |
| 초록 |
A new global optimization algorithm is proposed which is applicable to NLPs consist of twice differentiable objective functions. The key idea of the algorithm is the generation of difference of convex underestimator of objective function iteratively. In the case of global minimization problem, the lower bound of objective function is given by the global optimum of underestimator. There are many researchers who utilize a convex function as underestimator and they use a local optimization technique to obtain the lower bound of objective function. In this paper the difference of convex underestimator is always represented by a continuous piecewise concave function whose global minimum can be easily found. And the lower bound can be updated by only one calculation of objective function and gradient of that. The upper bound is additionally obtained during updating lower bound. The novel algorithm is successfully applied to several unconstrained problems and linearly constrained problems.
|
| 저자 |
장민호, 박영철, 이태용
|
| 소속 |
한국과학기술원 생명화학공학과 |
| 키워드 |
Global Optimization; Difference of Convex Underestimator; Branch-and-Bound Algorithm
|
| E-Mail |
, , |
| VOD |
VOD 보기 |
| 원문파일 |
초록 보기 |
