TY - GEN

T1 - A PDEM-COM framework for quantification of epistemic uncertainty

AU - Wan, Zhiqiang

AU - Chen, Jianbing

AU - Li, Jie

AU - Beer, Michael

N1 - Funding information: Financial supports from the National Science Fund for Distinguished Young Scholars of China (Grant No.51725804), National Science Fund of China (Grant No.51538010) and the Funds for International Cooperation and Exchange of the National Natural Science Foundation of China (Grant No.11761131014) are gratefully appreciated.

PY - 2020

Y1 - 2020

N2 - In the uncertainty quantification and structural reliability evaluation, epistemic uncertainty usually exists simultaneously with the aleatory uncertainty. To characterize these two types of uncertainty is crucial for decision making in structural design. Though extensive investigations have been conducted, resulting in various approaches, including, e.g., the fuzzy analysis method, the imprecise probability and empirically based probabilistic method, etc., a compatible framework with efficient implementation is still in need. Actually, due to limited available data, the distribution parameters (e.g. mean value and standard deviation) carry epistemic uncertainty, thus the basic random variables should be characterized by a family of probability distributions, rather than an uniquely specified probability distribution. This set of distribution parameters can be determined by the bootstrap method. Such problem is addressed in the present paper. Moreover, to improve the computational efficiency, a newly proposed method called PDEM-COM is adopted, without compromising numerical accuracy. Numerical applications are illustrated to indicate the feasibility of PDEM-COM framework for quantification of epistemic uncertainty. Moreover, this basic idea can also be extended to quantification of epistemic uncertainty due to other sources.

AB - In the uncertainty quantification and structural reliability evaluation, epistemic uncertainty usually exists simultaneously with the aleatory uncertainty. To characterize these two types of uncertainty is crucial for decision making in structural design. Though extensive investigations have been conducted, resulting in various approaches, including, e.g., the fuzzy analysis method, the imprecise probability and empirically based probabilistic method, etc., a compatible framework with efficient implementation is still in need. Actually, due to limited available data, the distribution parameters (e.g. mean value and standard deviation) carry epistemic uncertainty, thus the basic random variables should be characterized by a family of probability distributions, rather than an uniquely specified probability distribution. This set of distribution parameters can be determined by the bootstrap method. Such problem is addressed in the present paper. Moreover, to improve the computational efficiency, a newly proposed method called PDEM-COM is adopted, without compromising numerical accuracy. Numerical applications are illustrated to indicate the feasibility of PDEM-COM framework for quantification of epistemic uncertainty. Moreover, this basic idea can also be extended to quantification of epistemic uncertainty due to other sources.

KW - Change of probability measure

KW - Epistemic uncertainty

KW - Nonlinear structures

KW - PDEM

UR - http://www.scopus.com/inward/record.url?scp=85089175302&partnerID=8YFLogxK

U2 - 10.3850/978-981-11-2724-3_0969-cd

DO - 10.3850/978-981-11-2724-3_0969-cd

M3 - Conference contribution

AN - SCOPUS:85089175302

SP - 2622

EP - 2627

BT - Proceedings of the 29th European Safety and Reliability Conference, ESREL 2019

A2 - Beer, Michael

A2 - Zio, Enrico

CY - Singapur

T2 - 29th European Safety and Reliability Conference, ESREL 2019

Y2 - 22 September 2019 through 26 September 2019

ER -