Masters Theses

Abstract

"Existing techniques for reliability analysis are computationally expensive. This thesis presents a new technique that requires significantly less computational effort. This technique relies on the First Order Reliability Method (FORM), most probable point based (MPP)-based importance sampling, and support vector machines (SVM). These methods are used to calculate the probability of failure using a small number of samples. First, the MPP is located, and then samples are generated around this point. These samples are used to approximate the limit-state function in the SVM. The MPP of the approximated function is then shifted to the MPP of the given limit-state function. Finally, the approximated function is evaluated by calculating the probability of failure.

The small sample size required by this method reduces the computational cost for linear as well as nonlinear problems. The results have proved remarkably accurate in comparison with those obtained from Monte Carlo simulation with a large sample size"--Abstract, page iii.

Advisor(s)

Du, Xiaoping

Committee Member(s)

Takai, Shun
Chandrashekhara, K.

Department(s)

Mechanical and Aerospace Engineering

Degree Name

M.S. in Mechanical Engineering

Publisher

Missouri University of Science and Technology

Publication Date

Spring 2009

Pagination

ix, 44 pages

Note about bibliography

Includes bibliographical references (pages 41-43).

Rights

© 2009 Venkata Naga Praveen Thadigadapa, All rights reserved.

Document Type

Thesis - Restricted Access

File Type

text

Language

English

Library of Congress Subject Headings

First-order logic
Reliability (Engineering)
Reliability -- Analysis

Thesis Number

T 9497

Print OCLC #

436233577

Electronic OCLC #

906033328

Link to Catalog Record

Electronic access to the full-text of this document is restricted to Missouri S&T users. Otherwise, request this publication directly from Missouri S&T Library or contact your local library.

http://laurel.lso.missouri.edu/record=b7077218~S5

Share

 
COinS