0
RESEARCH PAPERS

Defect Classification in Pipes by Neural Networks Using Multiple Guided Ultrasonic Wave Features Extracted After Wavelet Processing

[+] Author and Article Information
Piervincenzo Rizzo

NDE & Structural Health Monitoring Laboratory, Department of Structural Engineering,  University of California, San Diego, 9500 Gilman Drive, M.C. 0085, La Jolla, CA 92093-0085prizzo@soe.ucsd.edu

Ivan Bartoli

NDE & Structural Health Monitoring Laboratory, Department of Structural Engineering,  University of California, San Diego, 9500 Gilman Drive, M.C. 0085, La Jolla, CA 92093-0085ibartoli@ucsd.edu

Alessandro Marzani

Dip. di Ingegneria delle Strutture, dei Transporti, delle Acque, del Rilevamento, del Territorio (DISTART),  Universitá degli Studi di Bologna, Viale Risorgimento 2, Bologna 40136, Italyalessandro.marzani@mail.ing.unibo.it

Francesco Lanza di Scalea1

NDE & Structural Health Monitoring Laboratory, Department of Structural Engineering,  University of California, San Diego, 9500 Gilman Drive, M.C. 0085, La Jolla, CA 92093-0085flanza@ucsd.edu

1

To whom correspondence should be addressed.

J. Pressure Vessel Technol 127(3), 294-303 (Jan 27, 2005) (10 pages) doi:10.1115/1.1990213 History: Received January 25, 2005; Revised January 27, 2005

This paper casts pipe inspection by ultrasonic guided waves in a feature extraction and automatic classification framework. The specific defect under investigation is a small notch cut in an ASTM-A53-F steel pipe at depths ranging from 1% to 17% of the pipe cross-sectional area. A semi-analytical finite element method is first used to model wave propagation in the pipe. In the experiment, reflection measurements are taken and six features are extracted from the discrete wavelet decomposition of the raw signals and from the Hilbert and Fourier transforms of the reconstructed signals. A six-dimensional damage index is then constructed, and it is fed to an artificial neural network that classifies the size and the location of the notch. Overall, the wavelet-based multifeature analysis demonstrates good classification performance and robustness against noise and changes in some of the operating parameters.

FIGURES IN THIS ARTICLE
<>
Copyright © 2005 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 3

Raw waveforms recorded for (a) defect size 5 at a 500mm notch-receiver distance; (b) defect size 7 at a 500mm notch-receiver distance; (c) defect size 7 at a 1300mm notch-receiver distance.

Grahic Jump Location
Figure 2

(a) Experimental setup; (b) the different defect sizes examined.

Grahic Jump Location
Figure 1

(a) Notation and mesh for the SAFE model of the pipe; (b) phase velocity dispersion curves; (c) energy velocity dispersion curves; (d) attenuation curves.

Grahic Jump Location
Figure 8

Defect classification performance of the four-combination problem as a function of (a) the excitation voltage to the MsS transmitter; (b) the number of signal averages; (c) the notch-receiver distance, and (d) the defect size.

Grahic Jump Location
Figure 4

Discrete wavelet analysis of signal in Fig. 3. (a) Gated direct signal; (b) gated defect reflection; (c) cD5 wavelet coefficient vector for direct signal (10% threshold); (d) cD5 wavelet coefficient vector for defect reflection (70% threshold); (e) reconstructed direct signal; (f) reconstructed defect reflection.

Grahic Jump Location
Figure 5

Damage index computed by (a) the peak amplitude of the Hilbert transform; (b) the variance of the wavelet coefficient vector cD5, and (c) the area of the FFT amplitude spectrum of the reconstructed signal. Data for a 100mm notch-receiver distance. (d) Pipe area reduction versus defect size.

Grahic Jump Location
Figure 6

Damage index computed by (a) the peak amplitude of the Hilbert transform; (b) the variance of the wavelet coefficient vector cD5, and (c) the area of the FFT amplitude spectrum of the reconstructed signal. Data for a 900mm notch-receiver distance. (d) Pipe area reduction versus defect size.

Grahic Jump Location
Figure 7

Defect classification performance of the nine-combination problem as a function of (a) the excitation voltage to the MsS transmitter; (b) the number of signal averages; (c) the notch-receiver distance, and (d) the defect size.

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In