Back to Search
Start Over
An effective subinterval analysis method for uncertain problems with large uncertainty based on positive and negative gradients.
- Source :
-
Advances in Engineering Software (1992) . Jul2024, Vol. 193, pN.PAG-N.PAG. 1p. - Publication Year :
- 2024
-
Abstract
- • The effective subinterval analysis is conducted to handle large uncertainty. • The positive and negative gradients are used simultaneously. • The method can avoid falling into the local optimal solution. • The method presents good computational accuracy and efficiency. An effective subinterval analysis method is proposed to predict the response boundaries of uncertain problems with large uncertainty based on the positive and negative gradients. In the proposed method, the uncertain parameters with large uncertainty are described as interval variables. The variation intervals of interval variables are firstly divided into a series of uniform subintervals, and the original uncertainty domain with large uncertainty is thereby transformed into a series of subdomains with small uncertainty. Then, based on the gradient information of response function, the positive and negative gradients are employed as the search directions to identify the potential uncertainty subdomains which may contribute to the response boundaries. Next, the first-order Taylor expansion is used to precisely calculate the response boundaries of the uncertain problem on the potential uncertainty subdomains. Subsequently, the response boundaries on the original large uncertainty domain are further obtained through comparing the responses obtained on the uncertainty subdomains. Finally, several numerical examples and application examples are used to verify the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]
- Subjects :
- *INTERVAL analysis
*TAYLOR'S series
Subjects
Details
- Language :
- English
- ISSN :
- 09659978
- Volume :
- 193
- Database :
- Academic Search Index
- Journal :
- Advances in Engineering Software (1992)
- Publication Type :
- Academic Journal
- Accession number :
- 177315755
- Full Text :
- https://doi.org/10.1016/j.advengsoft.2024.103650