Toggle Main Menu Toggle Search

Open Access padlockePrints

System identification-based frequency domain feature extraction for defect detection and characterization

Lookup NU author(s): Professor Gui Yun TianORCiD, Professor Jeffrey Neasham, Dr David Graham



This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND).


© 2018 Feature extraction is the key step for defect detection in Non-Destructive Evaluation (NDE) techniques. Conventionally, feature extraction is performed using only the response or output signals from a monitoring device. In the approach proposed in this paper, the NDE device together with the material or structure under investigation are viewed as a dynamic system and the system identification techniques are used to build a parametric dynamic model for the system using the measured system input and output data. The features for defect detection and characterization are then selected and extracted from the frequency response function (FRF) derived from the identified dynamic model of the system. The new approach is validated by experimental studies with two different types of NDE techniques and the results demonstrate the advantage and potential of using control engineering-based approach for feature extraction and quantitative NDE. The proposed approach offers a general framework for selection and extraction of the dynamic property-related features of structures for defect detection and characterization, and provides a useful alternative to the existing methods with a potential of improving NDE performance.

Publication metadata

Author(s): Li P, Lang Z-Q, Zhao L, Tian G, Neasham JA, Zhang J, Graham DJ

Publication type: Article

Publication status: Published

Journal: NDT and E International

Year: 2018

Volume: 98

Pages: 70-79

Print publication date: 01/09/2018

Online publication date: 19/04/2018

Acceptance date: 02/04/2018

Date deposited: 12/06/2018

ISSN (print): 0963-8695

ISSN (electronic): 1879-1174

Publisher: Elsevier Ltd


DOI: 10.1016/j.ndteint.2018.04.008


Altmetrics provided by Altmetric


Funder referenceFunder name