Title: "Autonomous Optical Inspection of Large Scale Freeform Surfaces"
Friday 26 October at 13:30, The Technical University of Denmark, Richard Petersens Plads, Building 303A, Aud. 41
Supervisor:
Professor MSO Anders B. Dahl, DTU Compute
Co-supervisors:
Henrik Aanæs, Software Developer, 3Shape
Ewa Nielsen, Head of Quality Execution, Siemens
Science summary:
Quality inspection of wind turbine blades is very important for both the performance and the life time of
the blades, and thus their respective turbines. It has been assessed, that about 80% of all blade failures,
which leads to turbine loss, are introduce
d during production. This Ph.D. thesis studies how the surface
of blades can be inspected automatically by a robot system, without human interference. A new
surface
measurement
system is introduced
with a robot holding
a scanner, which measures the surface
up
-
close
and
moves the scanner along the blade surface. The surface is however
difficult to follow
because
the blades are flexible and bend and twist, why their exact location is not known when the inspection
starts. Consequently, the robot must adapt to
the surface position. It does so by using the scanner’s
measurements to “see” the surface, and then change its movement accordingly. It is the first time such
an automatic inspection system has been demonstrated on wind turbine blades.
We need to know how
precise the measurements are. The scanner uses a camera to do its
measurements and therefore it is susceptible to the optical behavior of the scanned surface. This effect
is studied through the development of a separate reflection measurement system, whic
h also uses a
robot. From reflection measurements of the blade surface, it was found, that the blade surface is well
suited for scanning. It is not easy to estimate the measurement precision for a measurement system
capable of measuring large and flexible
objects such as wind turbine blades. The thesis studies the
measurement precision through an extensive pipeline, which involved cutting out a small section of a
blade. It was found that the measurements are accurate to within 1 millimeter, when measuring a
s much
as 20 meters of a 55 meter blade, even when taking the bending and twisting effects into account.
READ MORE at orbit.dtu.dk
Examiners:
Associate Professor Allan Aasbjerg Nielsen, DTU Compute
Professor Robert Jessen, Tromsø University
Professor Ole Madsen, Aalborg University
Chairman at defence:
Associate Professor Jeppe Revall Frisvad, DTU Compute
All are welcome