You are on page 1of 9

Control for Weld Penetration in VPPAW

of Aluminum Alloys Using the Front

Weld Pool Image Signal
The study shows the feasibility of implementing weld formation control of VPPAW
of aluminum in real time


Abstract. This paper presents a technique Compared with other welding processes, welding (PAW): inaccessibility of the
for real-time, closed-loop feedback con- VPPAW can generate high weld quality weld pool because of the limited torch
trol of weld penetration based on the and high productivity at relatively low stand-off distance and the interference of
front image signal of the weld pool in cost. These attractive features are attrib- the arc radiation. PAW technology in the
variable polarity plasma arc welding uted mainly to a fully penetrated one-keyhole-per-pulse mode can be
(VPPAW) of aluminum alloys. The for- keyhole-mode weld pool, inside which fairly well applied to steels, so the arc
mation of an image can be acquired hydrogen cannot be trapped, and to the sound or the arc efflux light from the back
when the arc light reflects off the con- removal of tenacious oxide film on the side of the workpiece can be used to de-
cave, mirror-like surface of the depressed workpiece surface, which guarantees tect the size of the keyhole. Based on this
keyhole weld pool and passes through a better fluidity of the metal in the weld principle, a full-penetration weld bead
band-pass filter onto the image sensor. pool. However, keyhole collapse and has been guaranteed in real-time feed-
The image of the visual keyhole (nominal melt-through may occur during a weld- back control (Refs. 6–9). However, this
keyhole) is a two-dimensional projected ing process if disturbances such as type of technology is not applicable to
picture of the actual keyhole weld pool. abruptly varying thermal conditions aluminum alloys. Also, detecting the
The determination of the geometrical exist, especially when welding plates keyhole from the back side of the work-
size of the nominal keyhole is also de- ranging from 4.0 to 25.4 mm. Thus, se- piece is not feasible in some cases, such
scribed according to the consecutive lecting process parameters and providing as in the welding of pressure vessels. To
frames of the image. The variation in size control of the stability of weld formation date, constant parameter open-loop con-
of the nominal keyhole is closely corre- during welding to produce a satisfactory trol of weld formation is still being used
lated to the bottom diameter of a key- weld remains a challenge. in PAW of aluminum alloys by the key-
hole. A model of the relationship be- One effective approach is to monitor hole mode.
tween the bottom diameter of a keyhole the keyhole weld pool. Recently, it was In recent years, welding researchers
and the geometrical size of the nominal found the presence or absence of a key- have focused on using machine vision
keyhole weld pool in the image is estab- hole could be determined by measuring systems to sense the weld pool for con-
lished and examined using the BP artifi- the ratio of hydrogen to argon in the trolling the full penetration state in gas
cial neural network theory. A cutting or a plasma arc column with an optical spec- tungsten arc welding (GTAW) and gas
keyhole collapse phenomenon is suc- trometer (Ref. 5). However, the size of the metal arc welding (GMAW) (Refs.
cessfully avoided and uniform weld for- keyhole cannot be determined and the 10–15). The arc light filtering solution has
mation is obtained in a welding process welding process cannot be distinguished been investigated through coaxial view-
using the model to control both the wire from the cutting process according to the ing of the weld pool in GTAW (Ref. 16).
feed and the welding current when the signal. At present, two difficult problems These approaches are based on the prin-
thermal conditions of the butt-jointed are associated with front-face sensing of ciple the diffuse reflection of arc light
workpieces are changed. The results the keyhole weld pool in plasma arc from the mirror-like weld pool surface is
achieved show a feasible way to imple- weaker than that from the surrounding
ment the real-time weld formation con- area. Thus, in the image, the weld pool
trol into the aluminum VPPAW. produces a dark area, while the solid part
of the workpiece appears as a bright area.
Introduction Key Words Some researchers used a pulsating laser
synchronized with a high-shutter-speed
Variable polarity plasma arc welding Aluminum Alloys camera to overcome the arc light inter-
(VPPAW) of aluminum alloys in the key- VPPAW ference in GTAW of stainless steel (Refs.
hole mode has been used successfully in Front Image Sensing 17–19). With this approach, a clear
production, such as in fabricating the Weld Pool image of the weld pool is captured and
space shuttle external tanks (Refs. 1–4). Keyhole Diameter the weld pool boundary is calculated in
Neural Networks Model real time using a developed image-
Penetration Control processing algorithm. The geometrical
are with Southern Methodist University, Dal-
Weld Formation appearance of the weld pool is charac-
las, Tex. Q. l. WANG is with Harbin Institute terized by the rear angles and the length
of Technology, Harbin, P. R. China.


aluminum alloys is the transition from a
Travelling CCD
mechanism Camaera Power source
keyhole welding process to a cutting
Image processing
for wire feeder software
process or to a melt-in-mode welding
process without a keyhole. Although lots

Wire of factors influence the stability of the

weld formation, variation in thermal con-
Computer system for Frame grabber ditions of the workpiece (which is an un-
weld monitoring and
controllable factor in application) is
mainly responsible for the transition if the
Plasma torch
optimized welding parameters are ap-
Plasma and Video recorder plied. Therefore, to avoid this transition,
shielding gases
VPPAW power and monitor the guarantee of the dynamic presence of
source system a keyhole weld pool in a plasma arc
Cooling water
system welding process with a varied thermal
Storage for data
condition is one of the biggest challenges
Motion control in the control of a quality weld.
Currently, the real-time feedback con-
trol of the weld formation using a keyhole
signal in PAW of aluminum alloys is still
Fig. 1 — Schematic of the monitoring and control system in VPPAW. not available, despite its applications to
some key products. Therefore, explo-
ration of this research issue is crucial to
achieving a quality weld. This paper will
of the weld pool. For the full-penetration focus on front imaging the weld pool to
case, the back weld bead width could be achieve the required signal of a keyhole;
related to the geometrical features of the establishing the model of the relationship
front side of the weld pool. Infrared ther- between the bottom diameter of a key-
mography has also been extensively in- hole and the geometrical size of a weld
vestigated (Refs. 20–22). It was found the pool in the image and implementing the
interference of the bright arc light could real-time feedback control of the weld
be prevented, and the depth of the joint penetration.
penetration could be determined using
the characteristics of the temperature Experimental Procedure
profiles of steel in GTAW. The instanta-
neous decrease of the welding current Experimental System
Fig. 2 — Schematic of workpieces. A — Or- has also been used to weaken the inten-
dinary conditions; B — varied thermal con- sity of the arc light in order to capture a The image sensing system used is
ditions with a slot. clear weld pool image during welding shown in Fig. 1. It consists of a commer-
(Ref. 23). However, in some cases, this cial CCD camera (15 mm in diameter
approach may not be acceptable be- and 100 mm in length) with a band-pass
cause of the poor weld bead that occurs, filter, a monitor, an image processing
such as in PAW of aluminum alloys using card, a PC-Pentium computer and a
the keyhole mode. video recorder. The parameters of the fil-
Besides the inaccessibility of the weld ter are the following: center wave length
pool and the arc radiation interference in 658 nm, half wave width 10 nm, trans-
PAW, a common problem existing in the missivity 27% and background depth of
welding of aluminum alloys is that nearly field 1/1000. These specific filter para-
no color change occurs when the plate is meters are selected because the intensity
melted. Consequently, the weld pool of the arc light when using argon as the
cannot be easily distinguished from the shielding gas is much weaker within the
solid base metal using the visible spec- above spectrum. The welding system
trum range. Thus, processing a weld pool also shown in Fig. 1 consists of a vari-
Fig. 3 — Three-dimensional schematic of a able-polarity welding power source,
image in the case of aluminum alloys is
much more difficult than in the case of programmable sequence controller,
steel welding. plasma gas controller, CNC positioning
Convex Top Face Weld Bead Another problem in the welding of system, computer-controlled wire feeder
aluminum alloys is the tenacious oxide and plasma arc torch. The camera is at-
film on aluminum strongly impedes the tached to and positioned in front of the
flow of the weld pool in the welding plasma arc torch with its axis at an angle
process so that a very poor weld forma- φ (43 deg) from the workpiece surface
tion with oxide inclusion is easily gener- plane.
Convex Bottom Weld ated. A cutting process with an oxidizing
Bead with Wire Feeding
Concave-downwards Bottom Weld Bead
surface cut may also occur if the plates Experimental Conditions
without Wire Feeding are melted through in a plasma arc weld-
ing process by the keyhole mode. Actu- Variations in the keyhole size can be
Fig. 4 — Schematic of a weld bead cross sec- ally, the main representation of an unsta- effected by changing some of the weld-
tion. ble weld formation in VPPAW of ing conditions such as the welding cur-

364-s | DECEMBER 2000

rent, the wire feed speed and the thermal
conditions of the plates (the shape of the
workpiece, for instance). Since variation
in the thermal conditions is crucial to

stability of weld formation in the weld-
ing of aluminum products, all the exper-
iments were conducted using work-
pieces with varied thermal conditions,
except the experiments done for com-
parison. Butt-joint welds were made
with wire feeding. Two shapes of 5-mm-
thick 2024 aluminum plates are shown Fig. 5 — Schematic of the normal keyhole weld pool viewed from the top. (S is the front pe-
in Fig. 2. The welding current in the start- riphery of the keyhole weld pool; R is the right periphery of the nominal keyhole; H is the
up segment reaches 80–90 A within 5 s, length of the nominal keyhole; L is the left periphery of the nominal keyhole; W is the width
which increases up to 90–120 A within of the nominal keyhole; and α is the angle between L and R.)
5–10 s from the termination of this seg-
ment to the beginning of the main body
segment. Other parameters have the fol-
lowing values: a direct current electrode
negative (DCEN) to direct current elec- A B
trode positive (DCEP) time ratio of 22 to
3.0 ms, DCEN current 80–100 A, DCEP
current 110–120 A, pilot arc current 15
A, plasma gas flow rate 4.5 L/min,
shielding gas flow rate 6 L/min, welding
speed 100 mm/min, torch stand-off dis-
tance 5 mm, angle of workpiece surface
plane to the horizontal plane equal to 85
deg, orifice diameter 3.2 mm, orifice
length 3.5 mm and a tungsten electrode Fig. 6 — Periphery tracing for the nominal keyhole. A — Without wire feed; B — with wire
setback of 3.2 mm. feed.

Image Features of the

Keyhole Weld Pool
Image Features curves (L, R and S) are not in the same
Keyhole Profile spatial plane (L and R are not parallel to
To better understand a keyhole image, A schematic of a keyhole weld pool the workpiece surface). The area
a three-dimensional schematic of a key- without wire feeding and with wire feed- bounded by lines L and R and the pe-
hole is shown in Fig. 3. The keyhole is ac- ing is shown in Fig. 5. The main features riphery S is a type of keyhole area
of the image are as follows: through which the plasma arc passes.
tually a cavity through the weld pool,
1) The images are two-dimensional This area is defined in this paper as the
with a profile resembling a trumpet. The
profiles of the keyhole weld pools pro- nominal keyhole. With wire feeding, the
cross section of this cavity varies both in
jected onto the target plane of the cam- lines L and R become a smooth curve,
diameter and shape (irregular ellipse).
era, which is defined in this paper as the and the middle segment of the periphery
The maximum cross-sectional diameter
visual (nominal) keyhole weld pool. This S is blocked by the cold wire.
occurs at the top surface of the workpiece means only a single viewpoint of the key- 4) A highlighted area behind the nom-
(top section diameter). The cross section hole weld pool can be captured by the inal keyhole exists in the image during a
with the smallest diameter is located camera, and the curves in the image are normal welding process. The highlighted
about 1.5–2.5 mm up from the bottom generally in different spatial planes. The area is not the image of the bright arc, but
surface of the workpiece (throat section pattern of the keyhole weld pool in the the reflected light from the arc on the rear
diameter). The diameter of the cross sec- image is deformed with respect to the surface of the weld pool. The area of the
tion on the bottom surface of the work- pattern of the actual keyhole weld pool. nominal keyhole will increase with an in-
piece is larger than the throat diameter 2) The front periphery (S) of the nom- crease in the welding current, but the
but less than the top section diameter. In inal keyhole weld pool is clear, but the highlighted area will decrease with an in-
uphill welding without wire feeding, the portion of it at the location with the crease in welding current. This is because
shape of a back-side weld bead is largest weld bead width cannot be the size of the actual keyhole increases.
concave-downward along the entire clearly distinguished from the solid area The area of the nominal keyhole in-
weld path — Fig. 4. Therefore, the key- of the base metal. The rear periphery of creases to a maximum until the high-
hole weld pool is a deformed weld pool, the nominal keyhole weld pool cannot lighted area vanishes (because there is no
and is much different from a weld pool in be seen in the image. surface to reflect the arc light) when the
GTAW and GMAW (Refs. 17–19). How- 3) The lines L and R are projected lines welding process becomes the cutting
ever, with wire feeding, the concave- of the actual periphery of the keyhole. process.
downward shape in the back side of a They may be the periphery of the keyhole 5) The bright arc cone cannot be seen
weld bead will be filled into the convex throat (at the throat cross section [Fig. 3]). directly from the image because of the
shape that is a normal full-penetration However, the periphery S is on the work- shielding of the torch gas cup.
weld bead. piece surface. So, actually, the three 6) Besides other dimensions of the



Fig. 7 — Variation of keyhole parameters. A — Bottom diameter of a keyhole; B — nominal keyhole area; C — nominal keyhole width; D — nom-
inal keyhole length.


Fig. 8 — Back side photographs of weld beads in butt joints with wire feed. A — Ordinary workpiece; B — varied heat sink workpiece.

nominal keyhole, there are three charac- tracted, as shown in Fig. 6A. fied algorithm is taken to acquire the area
teristic points with coordinates (XL, YL), The periphery of the nominal keyhole of a nominal keyhole:
(XR, YR) and (XO, YO) in the image. is determined as follows: 1) Within the predefined window, an
7) The quantity of the molten metal in • Find the vertical centerline of the image frame is divided into two areas
the weld pool with wire feeding is much highlighted area in the image. separated by the dotted line ab, which is
more than in the case of no wire feeding, • Determine the starting point for shown in Fig. 6B. The data points to the
so the bottom weld bead profile is con- tracing the periphery of the nominal key- left of line ab are processed in a sequence
vex. This will make the highlighted area hole. from left to right and from top to bottom.
in the image become larger than that in • Trace the periphery of the nominal In buffer 1, if the gray of a point (x, y) is
the case with wire feeding. However, the keyhole. greater than fifty and the gray difference
image is not as clear as in the case with- • Determine the front periphery of the between this point and the other point
out wire feeding. nominal weld pool. (x+5, y+5) is between zero and ten, the
gray of the point (x, y) is changed to zero.
Extracting the Nominal With Wire Feeding Otherwise, the original gray of the point
Keyhole Periphery (x, y) is saved. The same operation for the
To process the image of a nominal data points to the right of line ab is done,
Without Wire Feeding keyhole weld pool, one buffered area of but the sequence is from right to left and
the RAM in the computer is preserved for from top to bottom.
Because the image of the nominal storing the data of a whole frame of the 2) The first point with a gray value of
keyhole weld pool is clear and the dif- image sampled per second. The same zero in each column is searched by scan-
ference in grayness between the periph- size area of RAM is preserved for a copy ning the image frame from top to bottom.
ery of the nominal keyhole and the other of the data stored in the area of buffer 1. The left intersecting point between curve
area is large, no image preprocessing is During the data processing of a new sam- S and the nominal keyhole periphery arc
needed in the case of no wire feed. The pled image, the data in buffer 1 are L can be located according to the fol-
characteristic points and the lines of the changed, but the data in buffer 2 are kept lowing: to the left of line ab in the image
nominal keyhole can be directly ex- as the initial value. The following simpli- frame, if the ordinates y of the points with

366-s | DECEMBER 2000

Fig. 9 — Architecture of the BP neural network model of the key- Fig. 10 — Error comparison of the BP neural networks.
hole diameter.

the gray value of zero consecutively de- of the prerequisites for

crease and the last point among these achieving a uniform
points has the same abscissa as that of and stable weld forma-
line ab, the first point (xL, yL) among these tion.
points is the left terminal point of the In essence, the rea-
nominal keyhole periphery L. The right son for generating a
terminal point (xR, yR) of the nominal key- cutting process is that
hole periphery R can also be acquired by not enough metal ex-
the same method. ists in the weld pool
3) A straight line YL, with ordinate yL and too much energy is
to the left of line ab and perpendicular to deposited in the work-
the Y axis, is drawn. Another straight line, piece. The experiments
YR, can be drawn in a similar way. The reported in this paper
area I, surrounded by lines ab, L and YL, show that under con-
and the area II, surrounded by lines ab, R stant energy input,
and YR, are added together to represent when the thermal con-
the nominal keyhole area. dition of a workpiece is
The other geometrical size (width and changed due to the
length) of a nominal keyhole and the co- buildup of heat, the
ordinates of the points (XL, YL), (XR, YR) keyhole size becomes
and (XO, XO) can also be calculated ac- larger as the weld pool
cording to their number of pixels and becomes wider. When
their corresponding gray values in the the keyhole size is
image frame. larger than a certain
value, a cutting
Variation of a Keyhole process will be gener-
ated. To avoid this
Many factors — arc stability, energy melt-through mode
input of a workpiece, flow rates of gases, cutting process, the
position and speed stability of the wire quantity of the filler
feed, thermal condition variations gen- metal should be in-
erated by the workpiece structure and creased, which will de-
heat buildup because of the continuous crease the size of the
welding process itself — may influence keyhole even though
the transition from a normal keyhole the width of the weld
welding process to a cutting process or a bead generated will
melt-in mode welding process. The vari- probably be larger than
ation in thermal conditions is the most that of a normal key-
crucial and complex among these fac- hole weld bead. Figure
tors because of its variability, immeasur- 7 shows the relation-
ability and uncontrollability. In keyhole ship of the bottom di-
welding, the cutting process usually oc- ameter of a keyhole
curs more easily than in the melting-in and the nominal key-
mode welding process. Once a cutting hole size to the time
process occurs, the weld bead is not ac- using the workpieces
ceptable and must be repaired. Thus, shown in Fig. 2, in Fig. 11 — Bottom diameters of a keyhole from neural net output and
avoiding a cutting process resulting which the origin of the actual measurement. A — Constant welding parameters; B — vari-
mainly from a variation in thermal con- time axis represents the able welding current; C — variable wire feed speed.
ditions is a high priority. This is also one instant when the work-


Table 1 — Welding Parameters for BP Neural Network Modeling of the Bottom Diameter of a
Keyhole with Wire Feeding

Number of Slot length Welding current Wire feed speed


experiment (mm) (A) (mm/s)

1 0 97 14
2 0 97 15
3 6 97 15
4 6 97–87–97 15
5 6 97–92–97 15
6 6 97 15–17–15
7 6 97 15–19–15
8 6 97 15–21–15
9 6 97–92–97 15–19–15
10 6 97–92–97 15–21–15
11 6 97–87–97 15–17–15
12 6 97–87–97 15–19–15 Fig. 12 — Control model of the bottom di-
ameter of a keyhole.

piece movement is initiated after the key- the cutting process.

hole size reaches a certain value. Be- 2) The geometrical size (the
cause it is not currently possible to syn- area, width and height are shown,
chronously measure the bottom diameter respectively, in Fig. 7B–D) of the
of a keyhole in a welding process, the off- nominal keyhole in an image has
line measured width between the bottom the same trend as the bottom di-
oxide films along a weld bead, which is ameter of the keyhole, for both the
the trace of the arc efflux, is used for rep- ordinary and the varied heat sink
resenting the bottom diameter of a key- workpiece, even though there are
hole. To accurately acquire the data, some slight differences between
marks with an interval of 1.0 mm are them. The change in the area
made on the top and bottom faces along shows the largest variation among
the butt-jointed workpieces before weld- these characteristic sizes. How-
ing. The cycle of image sampling is 1.0 s ever, all this size data cannot rep-
and the welding speed is 6.48 m/h. resent the variation in the bottom
Fig. 13 — Training error of the BP neural network.
Therefore, the measuring interval is 1.8 diameter of a keyhole, which is ac-
mm. Associated with the corresponding tually a measurement of the full-
photos given in Fig. 8, the following re- penetration state. Thus, to avoid
sults were obtained: the cutting process, the bottom di-
1) The bottom diameter of the keyhole ameter of a keyhole should be cor-
on the ordinary workpiece increases with related with the geometrical size
time during the first 20 s — Fig. 7A. How- of the nominal keyhole.
ever, at the nearby origin, it is about 3) The allowable magnitude of
4.5–5.0 mm because of the arc preheat at the variation in the bottom diame-
the start location, and then decreases to ter of a keyhole is not large (± 2.0
3.8 mm because of the better heat trans- mm). The time interval required for
fer conditions of the workpiece after the welding process to become the
torch motion is initiated. From that in- cutting process for nominal weld-
stant, the bottom is nearly constant in the ing velocities of 100 mm/min Fig. 14 — Output of the controlling model of the wire
range of 3.8 to 4.0 mm up until 90 s. From shows the response time for both feed speed. (Note: curve 1 is the output of the model
about 90 s, it increases to 4.5 mm be- acquiring and processing the using the data from open-loop control experiments;
cause of the heat buildup and maintains image and for controlling the full curve 2 is the output of the model using the data from
4.5 mm after 110 s. Therefore, the key- weld penetration should be less closed-loop control experiments.)
hole welding process can enter a stable than 20 s.
state after 20 s from the start, during
which the variation of the keyhole diam- Modeling the Keyhole 4) Varied heat sink workpiece with a
eter is 0.2 mm up to 100 s. For the varied Diameter variation in welding current.
heat sink workpiece, it can also be seen 5) Varied heat sink workpiece with
the keyhole diameter has nearly the same The following experiments were variations in both the welding current
variation as that of the ordinary work- made for mapping the relationship of the and the wire feed speed.
piece up until 30 s, even though there ex- bottom diameter of a keyhole with the The purpose of the variations in the
ists a small difference during the first 10 geometrical size of the corresponding welding current and/or the wire feed
s. Because the thermal condition is get- nominal keyhole in an image: speed is to maintain a continuous weld-
ting worse after 30 s (the arc is reaching 1) Ordinary workpiece with constant ing process and prevent an occurrence of
the area with a slot machined on each welding parameters. a cutting process. The parameters ap-
workpiece), the variation of the keyhole 2) Varied heat sink workpiece with plied are shown in Table 1. As an exam-
diameter rapidly increases to more than constant welding parameters. ple of experiment 4, the welding current
6.8 mm at the instant of 72 s. After that, 3) Varied heat sink workpiece with a is maintained at an initial value of 97 A
the keyhole welding process becomes variation in wire feed speed. until the area with the varied heat sink is

368-s | DECEMBER 2000

reached. Then the welding current is
stepped down to 87 A. When the area
with the varied heat sink is passed, the
welding current is stepped up again to 97

A. In experiment 6, the initial wire feed
speed is maintained at a value of 15 cm/s
until the area with the varied heat sink is
reached. Then it is increased to a maxi-
mum value of 17 cm/s according to a
slope at the middle location of the area
with the varied heat sink. After that, it is
decreased to the initial value of 15 cm/s
according to a slope. Every experiment is
repeated four times. The first three of the
same four experiments are used for
model recognition, and the last experi-
ment is for verifying the accuracy of the
model established based on the three ex-
periments. Every group of parameters as-
sociated with the results of the image pro-
cessing is recorded once a second, 20
seconds after the weld initiation. From
the three groups of experiments in Table
1 (from 1 to 4, 5 to 8 and 9 to 12), 3855
groups of data were acquired.
An artificial BP neural network of
three layers with nine parameters as in-
puts and the bottom diameter of a key- Fig. 15 — Block diagram of the program for closed loop control of full penetration in butt-jointed
hole as an output, which can implement plates.
any map from m dimensions to n dimen-
sions, was used to model the system —
Fig. 9. As the number of hidden layers is guarantee full penetra-
not easily determined using current the- tion control and
ories, the BP models with one hidden achieve a uniform
layer consisting of elements of 3, 6, 9, 12, weld bead, the quan-
15 and 18 are respectively examined tity of the weld pool
using the Gauss initiation procedure and metal should be con-
the Delta rule (Ref. 24). Figure 10 shows trolled. Based on the
the error comparison of the models with fact the keyhole size
different layers. A comparison of the may be regulated by
neural network output with the measured adjusting the wire feed
results is given in Fig. 11. It can be seen speed and the heat
the training error of the model is the input may be con- Fig. 16 — Bottom diameter of a keyhole and wire feed speed vs. the
smallest value of 6.77%, and the corre- trolled by changing the time in the full penetration control of a weld bead in a butt joint
sponding error of the experiment exam- welding current, the using wire feed speed as a control variable.
ined is 6.64% when the number of the wire feeding speed
hidden layer element is selected as six. was selected as the first
Also, the corresponding curves measured controlling variable for
and output from the model agree well full weld penetration. A
with each other. Thus, the model with a
six-element hidden layer was selected as Controlling Model
the mapping model between the bottom
diameter of a keyhole and the geometri- To model the rela-
cal size of a corresponding nominal key- tionship between the
hole, as well as the three pairs of coordi- wire feed speed and
nates in the image schematically shown the bottom diameter of
in Fig. 5. However, the model should be a keyhole by the black- B
retrained using new experimental data if box method, the artifi-
the thickness of a workpiece and differ- cial BP neural network
ent welding parameters are applied. theory was applied.
Because the welding
Real-time Feedback Control of the process is a time delay
Full Weld Penetration process (the response
of the weld pool status
As stated above, the reason for the oc- to the adjustment of Fig. 17 — A weld bead with closed loop control of full penetration
currence of a cutting phenomenon is not the controlling para- in a butt joint using wire feed speed as a controlling variable. A —
enough metal is inside the weld pool. To meters takes a longer Top face; B — bottom face.



Fig. 18 — Schematic of the welding current regu- Fig. 19 — Bottom diameter of a keyhole, welding current and wire feed speed vs.
lation. the time in the full penetration control of a weld bead in a butt joint using wire feed
speed and welding current as controlling variables.

time), the previous data values for the first four experiments
input should be utilized in the establish- are for constant weld- A
ment of a model using current data value. ing current; the second
A model of three layered BP neural net- four experiments are
works was established, with the current for a decrease in weld-
wire feed speed as the output and the ing current by 5 A
current and previous bottom diameters of when the area with a
the keyholes as the input. The number of slot is reached; the last
hidden layer elements is selected from four experiments are
the experiments. In the beginning, the for a decrease in weld-
training sample data are acquired from ing current by 10 A B
the experiments in Table 1 when the wire when the heat sink of
feed speed was changed. So, two cate- the workpieces is in-
gories of data (associated with the varied creased.
heat sink workpiece and the ordinary According to the
workpiece, respectively) are selected ac- feedback control the-
cording to the standard: there is no ap- ory, it can be seen the
parent concave or undercut in a weld regulator in Fig. 12
bead. The bottom diameter of a keyhole, can regulate the wire Fig. 20 — A weld bead in a butt joint with the closed-loop control
which is the input for the controlling feed speed to change of weld formation using both wire feed and welding current as con-
model shown in Fig. 12, is the output in the keyhole size de- trolling variables. A — Top face; B — bottom face.
Fig. 9. From the comparison of the output pending on the varia-
results of the neural networks model with tion in the keyhole di-
the training sample data, illustrated in ameter, so an occurrence of the cutting etration when the thermal condition of
Fig. 13, it can be seen the model with process may be avoided. the workpiece changes.
three elements in the hidden layer has the • In the areas with a poorer heat sink
smallest error. Also, the difference be- Full Penetration Control because of the slots, the width and height
tween the output results of the neural net- of the bottom weld bead are apparently
works model and the training sample The block diagram of the program for wider and higher than those in other
data is large. The reason for this is the controlling full penetration of a weld areas with a better heat sink. The corre-
changes in the wire feed speed are lim- bead for the butt-joint plates is shown in sponding width of the top bead is a little
ited and the standard for an acceptable Fig. 15. Because a welding process needs wider and the corresponding height of
weld bead is not very strict. Conse- about 20 s to enter the stable main body the top bead is a little reduced. However,
quently, contradictory data are included segment from the start-up segment, the there are no concave or undercut defects
in the training data sample collection. control process of the full penetration be- in the weld bead.
In Fig. 14, curve 1 is the model output gins at the 20th cycle of the image sam- • The weld formation is not uniform.
of the wire feed speed regulator based on pling (i.e., the control strategy is applied
the above data sample collection used after the variable I equals 19 s in Fig. 15). Optimization of the
for training the model. The correspond- The wire feed speed and the bottom Weld Formation
ing initial and terminal data values, as diameter of a keyhole are recorded, mea-
well as the increment value of the bottom sured and calculated, the relationships of To achieve a uniform weld formation
diameter of a keyhole are respectively which, versus time, are shown in Fig. 16. during the full weld penetration control
3.0, 8.0 and 0.1 mm. Curve 2 is the out- The corresponding photograph of a weld process, the welding current should be
put of the same regulator trained using bead is given in Fig. 17. It shows the fol- controlled for regulating the heat input to
the new data sample collection acquired lowing: the workpiece. Although the experiments
from the feedback experiments done • The cutting process can be avoided show the cutting process can also be
under the condition of the varied heat reliably using the wire feed speed as a avoided by adjusting only the welding
sink according to curve 1. These experi- controlling variable in the real-time current, there often exist undercut and
ments are divided into three groups: the closed loop control of the full weld pen- concave defects in the top weld bead.

370-s | DECEMBER 2000

A solution that includes the regulation 3) The model, established between the Study on arc sound in TIG and plasma
of the wire feed speed is shown in Fig. 18. bottom diameter of a keyhole and the geo- processes. IIW Doc. 212-610-85.
Three thresholds of the bottom keyhole metrical size of a nominal keyhole weld 9. Hu, B. X. 1980. The study of controlling
diameters of 4.3, 4.6 and 5.1 mm are re- pool in the image, effectively and accu- system of welding quality at all-positions by

spectively matched with the welding cur- rately reflects the keyhole variation. The using the arc sound in pulsed plasma arc weld-
rent of 97, 92 and 87 A. The changing bottom diameter of a keyhole can be used ing of steel. Welding 3: 17–20
mode of the welding current is designed as a characteristic parameter to monitor 10. Agapakis, J. E., and Bolstad, J. 1991. Vi-
into the following: and control the full weld penetration. sion sensing and processing system for moni-
1) When the bottom diameter of a key- 4) The model for controlling the full toring and control of welding and other high
hole increases to a threshold, the welding weld penetration can be applied to a luminosity processes. Proc. of the Interna-
current is decreased according to the top welding process using the wire feed tional Robots & Vision Automation Confer-
downstairs-shaped curve in Fig. 18. speed as a control variable to reliably ence, pp. 23–29.
2) When the bottom diameter of a avoid an occurrence of a cutting process 11. Nakata, S, Huang, J., and Tsuruha, Y.
keyhole decreases to a threshold, the or a melt-in mode. However, a uniform 1988. Visual sensing system for in-process
welding current is increased according to weld bead may not be obtained by using control of arc welding process. Welding Inter-
the bottom upstairs-shaped curve. only the wire feed speed as a single con- national 12: 1086–1090.
3) When the bottom diameter of a trolling variable. 12. Hoffman, T. 1991. Real-time imaging
keyhole changes between the two neigh- 5) Uniform weld formation can be for process control. Advanced Materials &
boring thresholds and does not reach achieved using a combination of the wire Processes 9: 37 to 43.
those thresholds, the welding current is feed speed regulation with welding cur- 13. Guu, A. C., and Rokhlin, S. I. 1989.
held constant. To weaken the influence rent adjustment according to the varia- Computerized radiographic weld penetration
of any noise during an on-line control tion in the bottom diameter of a keyhole control with feedback on weld pool depres-
process, an average-weighted filtering in the full weld penetration control of a sion. Materials Evaluation 10: 1204–1210.
method is applied to the determination of weld bead. 14. Guu, A. C., and Rokhlin, S. I. 1992. Arc
the bottom diameter of a keyhole: the 6) The approach used in this paper weld process control using radiographic sens-
weighted coefficients are 0.5, 0.3 and should be further explored when thicker ing. Materials Evaluation 11: 1344– 1348.
0.2, respectively, for the current diame- materials are applied. The influence of 15. Rokhlin, S. I., and Guu, A. C. 1990.
ter, the former diameter and the diameter the arc light intensity should be a greater Computerized radiographic sensing and con-
before the former diameter. condition when designing a filtering sys- trol of an arc welding process. Welding Jour-
Figure 19 shows the responses of the tem under the condition of a high weld- nal 69(3): 83-s to 95-s.
welding current and the wire feed speed ing current. 16. Richardson, R. W., and Gutow, D. A.
to the variation in the bottom diameter of 1984. Coaxial arc weld pool viewing for
a keyhole during a real-time feedback Acknowledgments process monitoring and control. Welding Jour-
control process. Compared with the re- nal 63(3): 43-s to 50-s.
sults in Fig. 16, it can be seen the bottom The authors wish to acknowledge the 17. Kovacevic, R., and Zhang, Y. M. 1993.
diameter becomes significantly smaller financial support from the National Key Three-dimensional measurement of weld pool
in area with a poorer heat sink because Laboratory of Advanced Welding Pro- surface. Proc. of the International Conference
of the slots and the quantity of wire feed duction Technology at Harbin Institute of on Modeling and Control of Welding
is decreased to avoid a cutting process. Technology, P. R. China, and from the Processes, Fla.
The corresponding photograph of the National Science Foundation (Project 18. Kovacevic, R., and Zhang, Y. M. 1995.
weld bead (Fig. 20) also shows the bead Nos. DMI-9900011 and DMI-9700102). Vision sensing of 3D weld pool surface. Proc.
is more uniform than that in Fig. 17. of the 4th International Conference on Trends
Therefore, a combination of the wire feed References in Welding Research, Gatlinburg, Tenn.
regulation with welding current adjust- 19. Kovacevic, R., and Zhang, Y. M. 1996.
ment can further improve the weld for- 1. Nunes, A. C. 1984. Variable polarity Monitoring of weld penetration based on weld
mation in VPPAW of aluminum alloys by plasma arc welding on space shuttle external pool geometrical appearance. Welding Jour-
the keyhole mode. tank. Welding Journal 63(4): 27-s to 35-s. nal 75(10): 317-s to 329-s.
2. Woodward, H. A. 1996. U.S. contractor 20. Nagarajan, S., Chen, W. H., and Chin
Conclusions for the international space station. Welding B. A. 1989. Infrared sensing for adaptive arc
Journal 75(3): 35-s to 40-s. welding. Welding Journal 68(11): 462-s to
The following conclusions can be 3. Torres, M. R. 1992. Gas contamination 466-s.
made based on the results of this study: effects in variable polarity plasma arc welded 21. Nagarajan, S., Banerjee, P., Chen, W.
1) A clear image of the keyhole weld aluminum. Welding Journal 71(4): 123-s to H., and Chin, B. A. 1990. Weld pool size and
pool can be obtained using a band-pass 130-s. position control using IR sensors. Proceedings
filtering method. The nominal keyhole in 4. Martinez, L. F. 1994. Effect of weld gases of NSF Design and Manufacturing Systems
the image is a part of the actual keyhole, on melt zone size in VPPA welding of Al 2219. Conference. Arizona State University.
the geometrical features of which can re- Welding Journal 73(10): 50-s to 55-s. 22. Chen, W., and Chin, B. A. 1990. Mon-
flect the variation in the bottom diameter 5. Martinez, L. F. 1992. Front side keyhole itoring joint penetration using infrared sensing
of a keyhole regardless of whether there detection in aluminum alloys. Welding Journal techniques. Welding Journal 69(4): 181-s to
is wire feed or not. 71(5): 49-s to 52-s. 185-s
2) The main reason for an occurrence 6. Steffens, H. D. 1972. Automatic control 22. Oshima, K., and Morita, M. 1992.
of a cutting phenomenon is not enough for plasma arc welding with constant keyhole. Sensing and digital control of weld pool in
metal is appropriately added to the weld Welding Journal 51(6): 40-s to 45-s. pulsed MIG welding. Transactions of the Japan
pool during a welding process. The devel- 7. Metcalfe, J. C., and Quigley, M. B. C. Welding Society 23(4): 36–42.
opment of a cutting process from a keyhole 1975. Keyhole stability in plasma arc welding. 23. Hagan, M. T., Demuth, H. B., and
welding process takes some time, which Welding Journal 54(11): 401-s to 404-s. Beale, M. 1996. Neural Network Design. PWS
allows full weld penetration control. 8. Zhang, J. H., and Wang, Q. L. 1985. Publishing Co., pp. 1–43.