ISSN:
1013-9826
Source:
Scientific.Net: Materials Science & Technology / Trans Tech Publications Archiv 1984-2008
Topics:
Mechanical Engineering, Materials Science, Production Engineering, Mining and Metallurgy, Traffic Engineering, Precision Mechanics
Notes:
A key requirement in most ultrasonic weld inspection systems is the ability for rapidautomated analysis to identify the type of flaw. Incorporation of spatial correlation information fromadjacent A-scans can improve performance of the analysis system. This paper describes two neuralnetwork based classification techniques that use correlation of adjacent A-scans. The first methodrelies on differences in individual A-scans to classify signals using a trained neural network, with apost-processing mechanism to incorporate spatial correlation information. The second techniquetransforms a group of spatially localized signals using a 2-dimensional transform, and principalcomponent analysis is applied to the transform coefficients to generate a reduced dimensional featurevectors for classification. Results of applying the proposed techniques to data obtained from weldinspection are presented, and the performances of the two approaches are compared
Type of Medium:
Electronic Resource
URL:
http://www.tib-hannover.de/fulltexts/2011/0528/01/51/transtech_doi~10.4028%252Fwww.scientific.net%252FKEM.321-323.1266.pdf
Permalink