Publication detail

Estimation of Sampled Domain Probability using Convex Hull Approximation

GERASIMOV, A. VOŘECHOVSKÝ, M.

Original Title

Estimation of Sampled Domain Probability using Convex Hull Approximation

Type

conference paper

Language

English

Original Abstract

In various sampling methods developed for estimation of failure probability, there are stages at which it is useful to estimate the extent of the domain of basic variables which has already been covered by sampling points. This information is helpful for adaptive update of the sampling strategy. We propose using a geometrical representation of the explored part of the domain by a convex hull. Once performed, we may target at exploration sampling, as well as using convex hull for estimation measure of the (un)explored domain. Whereas in problems which feature just a few random variables we are able to set up convex hull bounding the given set of points precisely, in problems of high dimensions we only can assemble convex hull from a limited number of hyperplanes bounding a set of points. This paper compares various approximation to convex hull in small to moderate domain dimensions.

Keywords

Structural reliability, Spatial approach, Convex hull approximation, Failure probability, Monte carlo, Exploration sampling

Authors

GERASIMOV, A.; VOŘECHOVSKÝ, M.

Released

19. 9. 2021

Publisher

Research Publishing (S) Pte. Ltd.

ISBN

978-981-18-2016-8

Book

Proceedings of the 31st European Safety and Reliability Conference (ESREL 2021)

Pages from

2770

Pages to

2776

Pages count

7

URL

BibTex

@inproceedings{BUT172818,
  author="Aleksei {Gerasimov} and Miroslav {Vořechovský}",
  title="Estimation of Sampled Domain Probability using Convex Hull Approximation",
  booktitle="Proceedings of the 
31st European Safety and Reliability Conference (ESREL 2021)",
  year="2021",
  pages="2770--2776",
  publisher="Research Publishing (S) Pte. Ltd.",
  doi="10.3850/978-981-18-2016-8\{_}683-cd",
  isbn="978-981-18-2016-8",
  url="https://www.rpsonline.com.sg/proceedings/9789811820168/html/683.xml"
}