Niklas Svanberg; Irina Gertsovich , pp. 132. TEK/avd. för signalbehandling, 2008.
Since 1987 Uddcomb Engineering has repaired pulps by
their own developed overlay welding method even called
Uddcomb method. Currently each welding machine is operated
by two persons. To increase Uddcomb Engineering
competitiveness the reduced number of operators is desired.
An installation of a monitoring system which can
aid humans in the welding quality control also helps to improve
company’s position. A future goal would be to make
this monitoring system automatic without a human operator
in the loop.
In this thesis, arc voltage, weld current and audio signals
were collected and analyzed with aim on finding algorithms
to monitor the quality of the welding process. The
use of statistics tools is the basis for detecting variations in
the voltage and current data, associated with welding process.
It has been shown that voltage signal can be used
as a part of the welding quality control. The audio signal
from welding at low frequencies varies with the speed
of the process. The signal can also be incorporated in the
monitoring of the process.
The use of filters, growing sums and statistics are key
elements in the algorithms presented in this report.