Detection of Gaps and Overlaps in Laser Line Triangulation Data of Dry Fibre Tape Layups Using Image Projection

Abstract

A major cost driver in the production of carbon composite parts is the quality inspection, which to this day relies on manual investigation by a trained worker. Gaps and overlaps in the layups are to be detected because they are proven to be detrimental towards the mechanical properties of the final part. In recent works, laser line triangulation sensors have been applied to inspect layups of prepreg tapes and non-crimp fabric material. These sensors create a 3D point cloud of the specimens surface. This is then evaluated by conversion into a grey-scale image and a subsequent image processing algorithm. However, the most commonly used algorithms fail to differentiate between defects and small, acceptable irregularities, such as welding spots, slits and single fibres which stick out.The aim of this research is to develop a reliable evaluation method for scans of dry fibre tape layups. An overview over the different groups of algorithms is provided, image projection is selected and compared to algorithms which have been proven to work best on pre-pregs. While the common algorithms fail to classify a test set of dry fibre specimen, image projection can reach a true positive rate of 100% and a false positive rate of 19%. The proposed setup can be a centrepiece of a future in-line quality inspection system for dry fibre layups which has potential for a significant decrease of the manufacturing costs.