We describe a Compensated Laser Ultrasonic Evaluation (CLUE™) system which combines a wavefront scrubbed laser receiver with a pulsed laser ultrasonic generator. The receiver uses nonlinear optics to compensate for the speckle produced by the surfaces of the machined parts, thereby allowing the extraction of the signal from background noise.
A prototype CLUE™ system designed for the manufacturing environment has been constructed and tested. The system is ruggedized and fully automated, requiring only software installation to suit the particular needs of the end user. We have used the system to evaluate laser lap welds made in 1010 steel. A neural network was developed to classify the welds according to their fusion widths by training on features extracted from A-scans which were calibrated using destructive test measurements.