Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 Jul;69:1-10.
doi: 10.1016/j.ultras.2016.03.004. Epub 2016 Mar 10.

Material Grain Size Characterization Method Based on Energy Attenuation Coefficient Spectrum and Support Vector Regression

Affiliations

Material Grain Size Characterization Method Based on Energy Attenuation Coefficient Spectrum and Support Vector Regression

Min Li et al. Ultrasonics. .

Abstract

A grain size characterization method based on energy attenuation coefficient spectrum and support vector regression (SVR) is proposed. First, the spectra of the first and second back-wall echoes are cut into several frequency bands to calculate the energy attenuation coefficient spectrum. Second, the frequency band that is sensitive to grain size variation is determined. Finally, a statistical model between the energy attenuation coefficient in the sensitive frequency band and average grain size is established through SVR. Experimental verification is conducted on austenitic stainless steel. The average relative error of the predicted grain size is 5.65%, which is better than that of conventional methods.

Keywords: Austenitic stainless steel; Energy attenuation coefficient spectrum; Grain size characterization; Support vector regression; Ultrasonic test.

Similar articles

See all similar articles

Publication types

LinkOut - more resources

Feedback