requirement for system design.Many thick wall-welding systems are used to manufacture
high or low volume parts. This leads to a third main require-
ment for these systems which is that they must be able to deal
with a large range of joint profiles without significant amounts of
setup or programming.
This paper describes a system design and implementation
which addresses the main requirements listed above. After ini-
tial concept development, extensive discussions with the fabri-
cation department helped to refine the system specification and
design. Following the first implementations of the system, prod-
uct trials led to further improvements. Getting extensive practical
experience was a vital stage in achieving good levels of perform-
ance [2, 3]. This process of refining the system has taken place
over a period of some three years, and has led to a very flexible,
high performance system which is nevertheless easy to use.
2 Image processing
This system is designed for thick wall submerged arc welding
(SAW) applications where the welding operation can involve
many individual welding passes. Typical joint depths range from
less than 25 to 90 mm or greater. Typical joint preparations in-
clude “traditional” “V” grooves and semi narrow gap and narrow
gap U-type joint profiles. Joint profiles may be produced by ther-
mal processes or by machining. These two different approaches
introduce significantly different issues for the system. Obviously,
the system itself must be able to handle both methods of making
joint profiles with equal ease.
The overall welding operation is normally divided into a se-
ries of phases, usually:
• Root phase,
• Hot phase,
• Fill phase,
• Cap phase.
During the root and hot pass welds, the prime concern is to
achieve accurate seam tracking with tightly controlled welding
parameters. This addresses the issue of consistent root pene-
tration and reduces the likelihood of these early weld passes
burning through the root face.
The fill welding phase covers most of the welding cycle time.
The sensor system must address the following issues:
• Accurate placement of each individual weld pass, especially
on joint side walls.
• Automatic calculation of the number of passes required per
fill layer.
• Smooth automatic transitions between passes to minimize
unevenness of the weld bead height.
• Automatic compensation for variations in the joint area
across the joint, i.e., putting more weld material in one side
of the joint than the other.
• Automatic compensation for variations in the joint area along
the length of the joint, i.e., puttingmore weldmaterial in areas
where the joint is larger than in areas where it is smaller.
• Automatic detection of the end of the fill welding phase.An exact automatic transition function is required to minimize
unevenness of the weld bead height which may happen during
multi-pass transition welding phase. Figure 2 shows the transi-
tion distance for each pass change.
During the root pass, hot passes and for most of the fill weld-
ing phase, the joint remains physically well defined. The top
edges of the weld preparation can be used as a reference for
at least the top of the joint, and the bottom edges of the joint
sidewall can be detected reliably and provide useful information
about the current state of the partly welded joint.
Near the end of fill welding, and during cap welding, the
top edges of the weld joint may no longer be well defined. This
means that the laser sensor alone cannot be used as the sole in-
formation source. Two other information sources then come into 多传感器全自动焊接系统英文参考文献和翻译(2):http://www.youerw.com/fanyi/lunwen_370.html