Toggle Main Menu Toggle Search

Open Access padlockePrints

A novel vibration based non-destructive testing for predicting glass fibre/matrix volume fraction in composites using a neural network model

Lookup NU author(s): Professor Geoff Gibson

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


Abstract

This study proposes a novel approach to determine the fibre volume fraction in composites using vibration based non-destructive technique with a neural network. Currently, the volume fraction of a glass fibre/matrix based composite material is assessed using destructive techniques. Instead of changing or destroying the structure, a new non-destructive approach based on vibration analysis is proposed. Complete experimental protocols were developed to capture the vibration pattern. An auto-regressive model was developed as a feature extraction tool to classify the fibre volume fractions and as a pole tracking algorithm. The classification performances were within the range of 90-98%. For NDT method to be efficient, the classification results were then compared with destructive burn-out technique. The results of non-destructive test showed good agreement with those obtained through destructive test suggesting that the proposed method is an alternative to ASTM D2584-11 for determining the volume fraction of a glass fibre/matrix composite. (C) 2016 Elsevier Ltd. All rights reserved.


Publication metadata

Author(s): Farhana NIE, Majid MSA, Paulraj MP, Ahmadhilmi E, Fakhzan MN, Gibson AG

Publication type: Article

Publication status: Published

Journal: Composite Structures

Year: 2016

Volume: 144

Pages: 96-107

Print publication date: 01/06/2016

Online publication date: 27/02/2016

Acceptance date: 01/01/1900

ISSN (print): 0263-8223

ISSN (electronic): 1879-1085

Publisher: Elsevier

URL: http://dx.doi.org/10.1016/j.compstruct.2016.02.066

DOI: 10.1016/j.compstruct.2016.02.066


Altmetrics

Altmetrics provided by Altmetric


Share