Login | Register

Intelligent one-point damage localization of an isotropic surface pipeline using Guassian Process regression

Title:

Intelligent one-point damage localization of an isotropic surface pipeline using Guassian Process regression

khazaeli, shida (2018) Intelligent one-point damage localization of an isotropic surface pipeline using Guassian Process regression. Masters thesis, Concordia University.

[thumbnail of Khazaeli_MASc_F2018.pdf]
Preview
Text (application/pdf)
Khazaeli_MASc_F2018.pdf - Accepted Version
35MB

Abstract

Pipelines are subjected to many damaging agents, such as, earthquake, ground movement, and
aging which are responsible for important financial expenses. Structural Health Monitoring (SHM)
of civil structures using arrays of sensors is promising such that data form the monitoring systems
enable us to trace the structural anomalies and performance for early treatments. The need for
introducing faster and intelligent methods has helped researchers propose novel approaches for such
monitoring procedures. In this study a new method is introduced for monitoring of surface pipelines
used primarily for oil and gas. The framework takes the advantage of Gaussian Process Regression
Method (GPRM) to create a probabilistic predictive model for damage detection and the subsequent
localization of the defect. To this end, an isotropic pipeline is modeled numerically and validated with
an experimental setup. Afterwards, the model is extended to the real-life application to establish
a meta model. Damages are introduced as small holes at different locations (one at each time).
The GPRM is used to map the system responses to the selected statistical features which are
utilized as indicators for the existence of the damages and their locations. GPRM reveals more
promising results compared with conventional regression analysis. It considers the uncertainties due
to lack of observation. In addition, it is an updatable approach with having local effects on the
model. In another words, it affects the model in the vicinity of new observations. Moreover, among
selected statistical features, number of peaks greater than or equal to 20% and 60% of the maximum
peak values show better results corresponding to damage localization. Also the curve length and
correlation coefficient of the system response (induced signal) are found to be efficient for damage
detection. The novel method has been validated with filed measurements and experimental data
and found to work efficiently.

Divisions:Concordia University > Gina Cody School of Engineering and Computer Science > Building, Civil and Environmental Engineering
Item Type:Thesis (Masters)
Authors:khazaeli, shida
Institution:Concordia University
Degree Name:M.A. Sc.
Program:Civil Engineering
Date:6 August 2018
Thesis Supervisor(s):Bagchi, Ashutoush and khazaeli, shida
ID Code:984542
Deposited By: Shida Khazaeli
Deposited On:16 Nov 2018 15:55
Last Modified:01 Sep 2020 00:00
All items in Spectrum are protected by copyright, with all rights reserved. The use of items is governed by Spectrum's terms of access.

Repository Staff Only: item control page

Downloads per month over past year

Research related to the current document (at the CORE website)
- Research related to the current document (at the CORE website)
Back to top Back to top