Page 115 - Welding Robots Technology, System Issues, and Applications
P. 115
burn-off rate, arc stability and arc rooting. It is therefore desirable that the mean
current is constant and stable. Sensors for Welding Robots 101
By selecting suitable statistical detection algorithms a change in the weld condition
can be detected that produces decreased quality. This can be combined with a
decision function that accumulates over time and produces a weld error alarm after
a preset threshold value, defined case by case.
3.10 Discussion
In general, applying sensors to measure parameters at the welding process is a
difficult task as the process environment is a challenge to apply sensors in with
high temperature, liquid metal, high current, spatter, etc. Due to this, it is common
that parameters which we can observe by sensors are not the same as we want to
control. Thus, a simple feed-back control is often not adequate or possible to
obtain. Instead, models must be developed that map the observable parameters to
proper actions within a model based control concept that focuses on relevant issues
defined within a welding specification procedure. Here, productivity and quality
measures should be defined together with nominal data to produce the weld (weld
joint data, weld data for the welding power source and for the robot). Therefore, to
utilize fully the information from sensors, a model based control should be applied
together with welding procedure specifications that are defined with robotized
welding and model based control in mind.
Seam tracking sensors are the most commonly used sensor systems in robotic
welding. In general, two types of seam trackers exist, laser triangulation based
scanners and through arc trackers which need a weaving motion of the weld torch
during welding. Although laser based sensors have a better performance with
respect to accuracy and resolution, the choice is not obvious in a real case. The
through arc sensor works only during welding and can only measure anything
through the arc and hence, the operating “scan” is equal to the weaving amplitude.
This is critical for the run-in sequence of a weld using the sensor. However, the
sensor detects the weld right at the weld arc with reasonable accuracy, although not
all joints can be detected in a proper way compared to a laser sensor which is
mounted in front of the welding torch and thus takes up some space and also
requires being in line with the weld joint during welding. In addition to this rather
scattered picture, the price of the two most frequently used seam-tracking sensors
are quite different; the through-arc sensor is a low cost sensor while the laser
sensor is to be considered as rather expensive. Thus, in reality the choice is
dependent on many interacting factors within the specification of the task to be
performed.
Monitoring systems are for the most part used in production where high quality is
required. They provide in general the necessary means to work in real time with
quality assurance and, if needed, stop the welding operation if the weld is found to