Online infrared detection of inclusions and lack of penetration during welding

Abstract

Thermographic imaging using infrared sensors has been a natural choice for sensing and weld process monitoring. The basis for using infrared imaging lies in the fact that an ideal welding condition would produce surface temperature distributions that show a regular and repeatable pattern. Any variation or perturbation would result in discernible change in the thermal profiles. This paper focuses on the application of infrared imaging for online detection of lack of penetration and tungsten inclusions formed during manual metal arc welding. Both lack of penetration and tungsten inclusions could be distinctly detected with good confidence levels. Wavelet transform based image processing tools could be successfully applied for the extraction of the required features and determination of the dimensions of the discontinuities from the thermal images and radiographs. These studies form the basis for online discontinuity detection and control.