Machine vision platform for non-destructive testing methods of fibre reinforced plastics

Non-destructive testing (NDT) of fibre reinforced plastics (FRP) is a challenge due to their complex inner structures. Different NDT technologies like ultrasound, thermography or computer tomography (CT) are currently used for the inspection of FRP. Each of these testing methods has its pros and cons. CT offers detailed 3D information about the inner structures of the materials, but is quite time-consuming. Ultrasound and thermography are fast testing technologies and therefore could be also used for inline inspection. However they have the disadvantage of less detailed information compared to CT. In order to combine the advantages and overcome the disadvantages, CT is used as reference technology for the inline testing methods. This means that ultrasonic or thermography systems are qualified by the referencing process to detect defects so that they can deliver reliable inspection results within the production process. For this purpose it is absolutely necessary to easily compare the different kind of NDT methods. This has been achieved by developing a NDT software platform for loading, visualising and analysing ultrasonic, thermographic and CT data. This platform enables the user to analyse data sets of the different technologies in one single software package and to easily compare and also combine the results. For CT data evaluation advanced and adapted algorithms have been developed and integrated in the software platform, which allow automated defect detection and porosity analysis. In addition the fibre orientation can be determined based on a new 3D texture analysis algorithm. For 2D evaluation of ultrasound and thermography data advanced and automated defect detection algorithms have been implemented and included into the software platform. As some defects are only visible in one of the two testing technologies, a data fusion algorithm has been developed which combines the two analysis results. Through this procedure extensive failure detection is achieved. The new machine vision platform for NDT has been tested on selected data sets generated from real as well as special test parts. The evaluation results are promising and show clearly the possibilities offered for visualisation, analysis and data fusion for CT, thermography and ultrasound inspection.