In-process detection of weld defects using laser ultrasonics

Abstract

Laser ultrasonics shows great promise for on-line monitoring of weld quality in a number of laser welding applications. Laser ultrasonics uses a pulsed laser to generate an ultrasonic wave and a continuous-wave laser interferometer to detect the small surface displacement when this wave arrives at the point of detection. Laser ultrasonics is ideal for in-line measurements because there is no sensor in contact or near-contact with the workpiece. In addition, measurements can be performed on curved surfaces and in areas that are difficult to access.In this study, we have used laser-generated plate (Lamb) waves to interrogate laser butt welds used in the manufacture of tailor-welded blanks. We have used wavelet analysis to separate the temporal signals and select the distinguishing features that are most sensitive to weld integrity. We have detected a number of defect types with good reliability. We anticipate that, in an in-line industrial application, these measurements can be implemented just downstream from the weld process. The weld quality data can then be fed back to control critical weld parameters or alert the operator of a problem requiring maintenance. Weld defects can then be corrected before defective parts are produced.Laser ultrasonics shows great promise for on-line monitoring of weld quality in a number of laser welding applications. Laser ultrasonics uses a pulsed laser to generate an ultrasonic wave and a continuous-wave laser interferometer to detect the small surface displacement when this wave arrives at the point of detection. Laser ultrasonics is ideal for in-line measurements because there is no sensor in contact or near-contact with the workpiece. In addition, measurements can be performed on curved surfaces and in areas that are difficult to access.In this study, we have used laser-generated plate (Lamb) waves to interrogate laser butt welds used in the manufacture of tailor-welded blanks. We have used wavelet analysis to separate the temporal signals and select the distinguishing features that are most sensitive to weld integrity. We have detected a number of defect types with good reliability. We anticipate that, in an in-line industrial application, these measurements can be implemented just downstream f...