Machine vision industrialising the textile-based FRP production

Producing textile preforms for fibre-reinforced plastics (FRP) still involves many manually performed steps. In consequence, the build-up process is characterised by low reproducibility and high costs without quality traceability. This paper illustrates how inline metrology enables the automation of the preform production through machine vision, in spite of the difficult optical properties of the fibre structures. The main focus is on the algorithms for an accurate and robust detection of the fibre orientation under industrial conditions. A structure tensor method and a spectral analysis method are presented in detail. Thus, a control variable is provided to automate the preform build-up.