Metrology is the key to an economically feasible production of fiber-reinforced composites in the field of automated tape
laying, applying a novel laser light-section sensor system (LLSS) to measure process quality and feed back the results to
close control loops of the production system. The developed method derives 3D measurements from height profiles
through an in-process surface scan by the integrated LLSS. Gaps, overlaps, misalignment and defects of the composite
tapes are detected during their lay-up and consolidation by comparing the measurement results with a CAD/CAM model
of the lay-up. The height profiles are processed with a novel algorithm based on a non-linear least-square fitting to a set
of sigmoid functions to ensure sub-pixel accuracy.
Due to their advanced weight-specific mechanical properties, the application of fibre-reinforced plastics (FRP) has been
established as a key technology in several engineering areas. Textile-based reinforcement structures (Preform) in
particular achieve a high structural integrity due to the multi-dimensional build-up of dry-fibre layers combined with 3D-sewing
and further textile processes. The final composite parts provide enhanced damage tolerances through excellent
crash-energy absorbing characteristics. For these reasons, structural parts (e.g. frame) will be integrated in next
generation airplanes. However, many manufacturing processes for FRP are still involving manual production steps
without integrated quality control. The non-automated production implies considerable process dispersion and a high
rework rate. Before the final inspection there is no reliable information about the production status.
This work sets metrology as the key to automation and thus an economically feasible production, applying a laser light-section
sensor system (LLSS) to measure process quality and feed back the results to close control loops of the
The developed method derives 3D-measurements from height profiles acquired by the LLSS. To assure the textile's
quality a full surface scan is conducted, detecting defects or misalignment by comparing the measurement results with a
CAD model of the lay-up. The method focuses on signal processing of the height profiles to ensure a sub-pixel accuracy
using a novel algorithm based on a non-linear least-square fitting to a set of sigmoid functions. To compare the measured
surface points to the CAD model, material characteristics are incorporated into the method. This ensures that only the
fibre layer of the textile's surface is included and gaps between the fibres or overlaying seams are neglected. Finally,
determining the uncertainty in measurement according to the GUM-standard proofed the sensor system's accuracy.
First tests under industrial conditions showed that applying this sensor after the drapery of each textile layer reduces the
scrap quota by approximately 30%.
Fibre-reinforced plastics (FRP) are particularly suitable for components where light-weight structures with advanced
mechanical properties are required, e.g. for aerospace parts. Nevertheless, many manufacturing processes for FRP
include manual production steps without an integrated quality control. A vital step in the process chain is the lay-up of
the textile preform, as it greatly affects the geometry and the mechanical performance of the final part. In order to
automate the FRP production, an inline machine vision system is needed for a closed-loop control of the preform lay-up.
This work describes the development of a novel laser light-section sensor for optical inspection of textile preforms and
its integration and validation in a machine vision prototype. The proposed method aims at the determination of the
contour position of each textile layer through edge scanning. The scanning route is automatically derived by using
texture analysis algorithms in a preliminary step. As sensor output a distinct stage profile is computed from the acquired
greyscale image. The contour position is determined with sub-pixel accuracy using a novel algorithm based on a non-linear
least-square fitting to a sigmoid function. The whole contour position is generated through data fusion of the
measured edge points.
The proposed method provides robust process automation for the FRP production improving the process quality and
reducing the scrap quota. Hence, the range of economically feasible FRP products can be increased and new market
segments with cost sensitive products can be addressed.