Determination of the best distribution and effective interval using statistical characterization of uncertain variables

Minho Joo, Jaehyeok Doh, Jongsoo Lee

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In this paper, an algorithm for estimating the best distribution about data containing uncertainties is proposed. The proposed algorithm combines sequence statistical modeling (SSM) and a method for determining the minimum experimental data. SSM is a method for selecting the best distribution about data using a goodness of fit (GoF) test and a comparison of the likelihood. The method used to determine the minimum experimental data determines the minimum data needed to an estimate the best distribution. The SSM presented herein is a method for selecting a suitable data distribution when considering only a parametric distribution. Thus, in this paper, the SSM was improved in order to select the correct distribution of both parametric and non-parametric distributions simultaneously. In addition, with the existing method for determining the minimum data, the data should be added based on actual experiments when the results data show an insufficient number, and there is a limitation in that the designers cannot broadly identify the data required. To overcome this limitation, SSM and random sampling are applied to the method to determine the minimum data, and thereby, ensure that the designer knows the approximate minimum data needed. To verify the validity of the proposed algorithm, it was applied to a real world case study on determining multiple statistical parameters in the bolt fastening problem. The sequence of verification methods used is as follows: First, the best distribution of the bearing surface and thread friction coefficient estimated by the proposed algorithm and based on a normal distribution are selected as comparison targets. Second, the bearing surface and thread friction coefficient data are sampled within the 95% confidence interval of the two distributions. Third, the reliability of the sampled friction coefficient data are compared using a Monte-Carlo simulation and an equation to calculate the bolt fastening force. In this study, the effectiveness of the proposed algorithm is validated.

Original languageEnglish
Pages (from-to)358-367
Number of pages10
JournalJournal of Computational Design and Engineering
Volume5
Issue number3
DOIs
Publication statusPublished - 2018 Jul

Fingerprint

Bearings (structural)
Interval
Statistical Modeling
Bolts
Friction
Friction Coefficient
Normal distribution
Thread
Sampling
Experimental Data
Random Sampling
Goodness of Fit Test
Data Distribution
Confidence interval
Gaussian distribution
Likelihood
Experiments
Monte Carlo Simulation
Verify
Uncertainty

All Science Journal Classification (ASJC) codes

  • Computational Mechanics
  • Modelling and Simulation
  • Engineering (miscellaneous)
  • Human-Computer Interaction
  • Computer Graphics and Computer-Aided Design
  • Computational Mathematics

Cite this

@article{f11e6bf9a4004cf6859ac23b75844b76,
title = "Determination of the best distribution and effective interval using statistical characterization of uncertain variables",
abstract = "In this paper, an algorithm for estimating the best distribution about data containing uncertainties is proposed. The proposed algorithm combines sequence statistical modeling (SSM) and a method for determining the minimum experimental data. SSM is a method for selecting the best distribution about data using a goodness of fit (GoF) test and a comparison of the likelihood. The method used to determine the minimum experimental data determines the minimum data needed to an estimate the best distribution. The SSM presented herein is a method for selecting a suitable data distribution when considering only a parametric distribution. Thus, in this paper, the SSM was improved in order to select the correct distribution of both parametric and non-parametric distributions simultaneously. In addition, with the existing method for determining the minimum data, the data should be added based on actual experiments when the results data show an insufficient number, and there is a limitation in that the designers cannot broadly identify the data required. To overcome this limitation, SSM and random sampling are applied to the method to determine the minimum data, and thereby, ensure that the designer knows the approximate minimum data needed. To verify the validity of the proposed algorithm, it was applied to a real world case study on determining multiple statistical parameters in the bolt fastening problem. The sequence of verification methods used is as follows: First, the best distribution of the bearing surface and thread friction coefficient estimated by the proposed algorithm and based on a normal distribution are selected as comparison targets. Second, the bearing surface and thread friction coefficient data are sampled within the 95{\%} confidence interval of the two distributions. Third, the reliability of the sampled friction coefficient data are compared using a Monte-Carlo simulation and an equation to calculate the bolt fastening force. In this study, the effectiveness of the proposed algorithm is validated.",
author = "Minho Joo and Jaehyeok Doh and Jongsoo Lee",
year = "2018",
month = "7",
doi = "10.1016/j.jcde.2017.11.007",
language = "English",
volume = "5",
pages = "358--367",
journal = "Journal of Computational Design and Engineering",
issn = "2288-4300",
publisher = "Society for Computational Design and Engineering",
number = "3",

}

Determination of the best distribution and effective interval using statistical characterization of uncertain variables. / Joo, Minho; Doh, Jaehyeok; Lee, Jongsoo.

In: Journal of Computational Design and Engineering, Vol. 5, No. 3, 07.2018, p. 358-367.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Determination of the best distribution and effective interval using statistical characterization of uncertain variables

AU - Joo, Minho

AU - Doh, Jaehyeok

AU - Lee, Jongsoo

PY - 2018/7

Y1 - 2018/7

N2 - In this paper, an algorithm for estimating the best distribution about data containing uncertainties is proposed. The proposed algorithm combines sequence statistical modeling (SSM) and a method for determining the minimum experimental data. SSM is a method for selecting the best distribution about data using a goodness of fit (GoF) test and a comparison of the likelihood. The method used to determine the minimum experimental data determines the minimum data needed to an estimate the best distribution. The SSM presented herein is a method for selecting a suitable data distribution when considering only a parametric distribution. Thus, in this paper, the SSM was improved in order to select the correct distribution of both parametric and non-parametric distributions simultaneously. In addition, with the existing method for determining the minimum data, the data should be added based on actual experiments when the results data show an insufficient number, and there is a limitation in that the designers cannot broadly identify the data required. To overcome this limitation, SSM and random sampling are applied to the method to determine the minimum data, and thereby, ensure that the designer knows the approximate minimum data needed. To verify the validity of the proposed algorithm, it was applied to a real world case study on determining multiple statistical parameters in the bolt fastening problem. The sequence of verification methods used is as follows: First, the best distribution of the bearing surface and thread friction coefficient estimated by the proposed algorithm and based on a normal distribution are selected as comparison targets. Second, the bearing surface and thread friction coefficient data are sampled within the 95% confidence interval of the two distributions. Third, the reliability of the sampled friction coefficient data are compared using a Monte-Carlo simulation and an equation to calculate the bolt fastening force. In this study, the effectiveness of the proposed algorithm is validated.

AB - In this paper, an algorithm for estimating the best distribution about data containing uncertainties is proposed. The proposed algorithm combines sequence statistical modeling (SSM) and a method for determining the minimum experimental data. SSM is a method for selecting the best distribution about data using a goodness of fit (GoF) test and a comparison of the likelihood. The method used to determine the minimum experimental data determines the minimum data needed to an estimate the best distribution. The SSM presented herein is a method for selecting a suitable data distribution when considering only a parametric distribution. Thus, in this paper, the SSM was improved in order to select the correct distribution of both parametric and non-parametric distributions simultaneously. In addition, with the existing method for determining the minimum data, the data should be added based on actual experiments when the results data show an insufficient number, and there is a limitation in that the designers cannot broadly identify the data required. To overcome this limitation, SSM and random sampling are applied to the method to determine the minimum data, and thereby, ensure that the designer knows the approximate minimum data needed. To verify the validity of the proposed algorithm, it was applied to a real world case study on determining multiple statistical parameters in the bolt fastening problem. The sequence of verification methods used is as follows: First, the best distribution of the bearing surface and thread friction coefficient estimated by the proposed algorithm and based on a normal distribution are selected as comparison targets. Second, the bearing surface and thread friction coefficient data are sampled within the 95% confidence interval of the two distributions. Third, the reliability of the sampled friction coefficient data are compared using a Monte-Carlo simulation and an equation to calculate the bolt fastening force. In this study, the effectiveness of the proposed algorithm is validated.

UR - http://www.scopus.com/inward/record.url?scp=85048145721&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048145721&partnerID=8YFLogxK

U2 - 10.1016/j.jcde.2017.11.007

DO - 10.1016/j.jcde.2017.11.007

M3 - Article

AN - SCOPUS:85048145721

VL - 5

SP - 358

EP - 367

JO - Journal of Computational Design and Engineering

JF - Journal of Computational Design and Engineering

SN - 2288-4300

IS - 3

ER -