Details
| Originalsprache | Englisch |
|---|---|
| Seitenumfang | 7 |
| Publikationsstatus | Veröffentlicht - 26 Mai 2019 |
| Veranstaltung | 13th International Conference on Applications of Statistics and Probability in Civil Engineering - Seoul, South Korea, Seoul, Südkorea Dauer: 26 Mai 2019 → 30 Mai 2019 Konferenznummer: 13 |
Konferenz
| Konferenz | 13th International Conference on Applications of Statistics and Probability in Civil Engineering |
|---|---|
| Kurztitel | ICASP13 |
| Land/Gebiet | Südkorea |
| Ort | Seoul |
| Zeitraum | 26 Mai 2019 → 30 Mai 2019 |
Abstract
Two challenges may exist in the reliability analysis of highly reliable structures in, e.g., aerospace engineering. The first one is that, the failure probability may be extremely small (typically, smaller than 1e-6), which commonly prevents us from generating accurate estimation with acceptable computational costs by using the available methods. The second one is that, the available information for the input variables may be subject to incompleteness (e.g., sparse data) and/or imprecision (e.g., measuring error), which, makes it impossible to generate precise probability models for the input variables. To address the above two challenges, this work proposes two effective algorithms based on combining the sampling techniques (i.e., extended Monte Carlo simulation and subset simulation), active learning techniques and high-dimensional model representation decomposition. The proposed methods can effectively estimate the failure probability function w.r.t. the uncertain distribution parameters of the input variables with small number of training samples even when the failure event is extremely rare. A numerical test example is introduced to illustrate the proposed methods.
ASJC Scopus Sachgebiete
- Ingenieurwesen (insg.)
- Tief- und Ingenieurbau
- Mathematik (insg.)
- Statistik und Wahrscheinlichkeit
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
2019. Beitrag in 13th International Conference on Applications of Statistics and Probability in Civil Engineering, Seoul, Südkorea.
Publikation: Konferenzbeitrag › Paper › Forschung › Peer-Review
}
TY - CONF
T1 - Rare failure event analysis of structures under mixed uncertainties
AU - Wei, Pengfei
AU - Bi, Sifeng
AU - Zhang, Yi
AU - Beer, Michael
N1 - Conference code: 13
PY - 2019/5/26
Y1 - 2019/5/26
N2 - Two challenges may exist in the reliability analysis of highly reliable structures in, e.g., aerospace engineering. The first one is that, the failure probability may be extremely small (typically, smaller than 1e-6), which commonly prevents us from generating accurate estimation with acceptable computational costs by using the available methods. The second one is that, the available information for the input variables may be subject to incompleteness (e.g., sparse data) and/or imprecision (e.g., measuring error), which, makes it impossible to generate precise probability models for the input variables. To address the above two challenges, this work proposes two effective algorithms based on combining the sampling techniques (i.e., extended Monte Carlo simulation and subset simulation), active learning techniques and high-dimensional model representation decomposition. The proposed methods can effectively estimate the failure probability function w.r.t. the uncertain distribution parameters of the input variables with small number of training samples even when the failure event is extremely rare. A numerical test example is introduced to illustrate the proposed methods.
AB - Two challenges may exist in the reliability analysis of highly reliable structures in, e.g., aerospace engineering. The first one is that, the failure probability may be extremely small (typically, smaller than 1e-6), which commonly prevents us from generating accurate estimation with acceptable computational costs by using the available methods. The second one is that, the available information for the input variables may be subject to incompleteness (e.g., sparse data) and/or imprecision (e.g., measuring error), which, makes it impossible to generate precise probability models for the input variables. To address the above two challenges, this work proposes two effective algorithms based on combining the sampling techniques (i.e., extended Monte Carlo simulation and subset simulation), active learning techniques and high-dimensional model representation decomposition. The proposed methods can effectively estimate the failure probability function w.r.t. the uncertain distribution parameters of the input variables with small number of training samples even when the failure event is extremely rare. A numerical test example is introduced to illustrate the proposed methods.
UR - http://www.scopus.com/inward/record.url?scp=85083952727&partnerID=8YFLogxK
U2 - 10.22725/ICASP13.040
DO - 10.22725/ICASP13.040
M3 - Paper
T2 - 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP 2019
Y2 - 26 May 2019 through 30 May 2019
ER -