——ONLINE FIRST(NOT Peer-Reviewed)——

Title：An Overview of Arc Welding Control Systems

Author：Primož Podržaj
Institute/Affiliation ： Faculty of Mechanical
Engineering, University of Ljubljana, Askerceva 6,
1000 Ljubljana, Slovenia
Received：2018-10-25
Accepted：2018-10-30
Process：1、First trial(Field and check)
2、Peer review
3、Editing and three trials
4、Published online

Whioce Publishing Pte. Ltd. Singapore

Company Registration No.: 201427293E

7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
 An Overview of Arc Welding Control Systems
Primož Podržaj

Abstract—In this paper an overview of arc welding control enable the bonding of the surfaces. An alternative is the so
systems is given. At first some basic physical background of called fusion welding, where the heat input must be intense
the process is given. Then the two most important subtypes enough to locally melt the two weldpieces. In the case of a
Gas tungsten arc welding (GTAW) and Gas metal arc welding
(GMAW) are presented in some detail. In order to understand surface heat source the minimum rate of energy release per
the logic of feedback control systems, the most essential control unit area P required to maintain a molten weld pool of radius
theory is outlined shortly. In the overview of control systems r is approximately [3]:
a feedback signal is used a means of division. The analysis of
recent research papers in the area has shown that recently image AkTm
P = (1)
processing based control systems seem to be the most popular r
ones. where k is thermal conductivity, Tm is melting temperature
Index Terms—arc welding, control, overview, GTAW, GMAW, and A is a factor dependent on welding speed, weld size and
TIG, MIG/MAG. thermal diffusivity. The weld pool size is of course limited
by practical considerations. The typical values are within 10-
I. I NTRODUCTION 20mm range, which for steels results in the required power
W
density of around 107 m 2 . As shown in Fig. 2 this is a typical

I N general welding can be defined as a localized coalescence

of metals or nonmetals (usually called weldpieces) pro-
duced either by heating them to the welding temperature, with
value of power density provided by arc welding sources. This

Vaporisation
or without the application of pressure, or by the application No welding possible
dominated
of pressure alone, with or without the use of additional filler 1013
Vaporisation
metal. A schematic overview of the better known welding 1012 conduction
Electron and melting
processes is shown in Fig. 1. Although there are some 40 or so 1011
beam
Power density [W/m2]

(keyholing)
1010
CO2 laser Radial
9
Welding Process
10 Plasma conduction
108 dominated
with
107 Arc processes welding
Resistance High Beam Solid State
Gas Welding Arc Welding
Welding Energy Welding Welding 6
10
No welding possible Negligible
105 melting
4
- Butt - Carbon arc - Friction 10
- Oxy-acetylene - Metal arc - Ultrasonic
- Spot - Plasma arc - Electron beam - Diffusion
- Air-acetylene - Explosive
- Seam - Gas Metal Arc - Laser - Magnetically
Fig. 2. Power density for various welding processes [3]
(MIG) impelled Arc
- Oxy-hydrogen
- Projection - Gas Tungsten Butt Welding
Arc (TIG) (MIAB)
- Cold Metal
- Induction Pressure
Welding (IPW)
is one of the reasons for its widespread usage.
Transfer (CMT) - Flash Butt welding
(FBW)
An arc (see Fig. 3) refers to strong electric fields that cause
- Magnetic Pulse -
Welding (MPW)
cold welding process
Gas plasma Protection nozzle
Fig. 1. An overview of welding processes [1] Electrode
Nozzle constriction
welding processes, only a few processes are really important
Electric arc
[2]. One of the most important ones, if not the most important
Shielding gas
one, is arc welding. The term arc welding refers to a broad
group of welding processes that employ an electric arc as the
source of heat to melt and join metals. It is believed that in
the entire metal fabrication industry, arc welding is the third Fig. 3. Arc plasma [1], [4]
largest group of processes behind assembly and machining in
metal fabrication industry [2]. gas to induce an electric breakdown and continuous discharge
From the historical perspective the first type of welding was of plasma [5]. Thus, the current passes through insulation
forge welding, where the weldpieces are first heated and then media in a manner similar to air. A typical arc welding setup
hammered or pressed together to remove slag and oxides and is shown in Fig. 4. The welding current can be DC or AC
and the electrode can be consumable or non-consumable. In
P. Podržaj is with the Faculty of Mechanical Engineering, Uni-
versity of Ljubljana, Askerceva 6, 1000 Ljubljana, Slovenia e-mail: many cases the welding region is protected by a shielding
primoz.podrzaj@fs.uni-lj.si gas. A power source is also needed in order to get the voltage
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
 electrode holder characterized by a much higher arc energy density and higher
electrode plasma gas velocity due to the arc plasma being forced through
a constricting nozzle [11].
arc
B. Gas metal arc welding
power
source weldpiece Gas metal arc welding (GMAW) (sometimes also referred to
by its subtypes metal inert gas (MIG) welding or metal active
gas (MAG) welding) is widely applied in various industries
because of its high productivity, flexibility, and low cost. It
can be operated in semiautomatic and automatic modes and
Fig. 4. Schematic representation of arc welding
can be used particularly well in a high-volume production
environment [12]. A typical setup is shown in Fig. 6. In
down to appropriate value. Welding processes in general need
Solid
lower voltages and higher currents. Advanced power sources electrode
are also capable of providing signals of various waveforms, wire
Shielding gas
which might be adjusted in real time based on some feedback input Current
signal. Power source related research is still very active [6], conductor

[5], [7]. There are several subtypes of arc welding, but only Wire guide and
the two most common ones will be described in some detail. Direction contact tube
of travel

Gas nozzle
A. Gas tungsten arc welding
Gasseous
Gas tungsten arc welding (GTAW), also known as tungsten Consumable electrode
shield
inert gas (TIG) welding is a widely used high-quality, high- Arc
precision welding process. In TIG welding, an arc is estab-
lished between a nonconsumable tungsten electrode and the
metal that is heated and melted (see Fig. 5). Filler wire may Weld
Base
metal
metal

Fig. 6. Schematic representation of GMAW (MIG/MAG) [13]

Current conductor
Welding GMAW, there are three major modes of metal transfer from
Torch
the electrode wire to the weld pool: globular transfer, spray
Shielding gas transfer, and pulse transfer as shown in Fig. 7 (some authors
hose
also define streaming transfer [14]). By using a pulse current,

Nonconsumable
tungsten electrode Gas nozzle

Shielding gas
Power source Direction of
Arc welding

Base metal
Solidifield Liquid weld
weld metal pool Fig. 7. Metal transfer modes (globular, spray, pulse) [15]

Fig. 5. Schematic representation of GTAW (TIG) welding [8] it is for example possible to achieve a stable one drop per
pulse mode and therefore a controllable metal transfer to the
be added for the weld. The inert gas [9]: weldpiece [16]. A feedback control system for this type of
1) shields the welding area from the air, preventing oxida- welding have already been proposed [17].
tion, Despite being one of the most common arc welding pro-
2) transfers heat from the electrode to metal, and cesses, GMAW also has shortcomings. One is the fact that
3) helps to start and maintain a stable arc (due to the low the same current is used to melt the wire and heat the
ionization potential material. That means, that by increasing the melting speed
Although pure argon is the most common choice, shielding for higher productivity we also increases the base metal heat
gas composition can also have effect on the resulting weld input resulting in increased distortion. To resolve this problem
in some cases [10]. The process can be used for many a modified approach called double-electrode GMAW or DE-
types of materials in all welding positions, and it provides GMAW has been proposed [18]. As each arc’s magnetic field
a concentrated, stable arc and a high-quality and neat weld influences the other, the process is still being researched [19],
deposit. Plasma arc welding (PAW) is an extension of GTAW [20].
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
 II. CONTOL SYSTEM BACKGROUND made. Because of that, a lot of research is made based on
some more or less easily measured signal, which however is
In control system theory the system is usually presented by a
correlated with good quality weld. Some of the most common
block diagram as shown in Fig. 8. In the case of arc welding
systems will be discussed.
d d

x y ydes x y
A. Seam tracking
object controller object
Especially in robot arc welding applications it is very impor-
tant for the welding system to follow the seam accurately. This
type of arc welding control system do not exactly fit into the
Fig. 8. Open loop and close loop control system
idea presented in the following section, despite being a proper
the object is the welding process under consideration. The feedback control system. The main goal of seam tracking
input vector x is composed of all the parameters that influence control system is the elimination of one disturbance (improper
the process. These are welding voltage, distance between positioning of the electrode in relation to the weldpieces.
the electrode and the weldpiece, electrode geometry, wire There are various approaches in this group. Some researchers
feed rate (in GMAW), weldpiece thickness and composition, have for example used ultrasound [31], but the most common
shielding gas flow and composition, etc. The components of approach are image processing based control systems. An early
the output vector y are all the characteristics of the resulting example of such a system is shown in Fig. 9 [32]. The system
weld (weld strength, corrosion resistance, visual appearance,
Interupt signal
possible deformations etc.). The relationship between input
Video signal
and output can therefore at least in theory be given by some Driving wheel
Laser beam
vector function f Pre-processor
projector
y = f (x) (2) Rotary
encoder
Although we cannot write down this function, experience tells Image processor
us, that an appropriate combination of input parameters x, will Image sensor

usually (but not always) result in an acceptable output y. The Welding torch
problem lies in the fact that in some cases we lack experience
or there are some signals, which we cannot monitor or control
(the so called disturbances d. So, Eq. 2 can be rewritten into Servomechanism
the following form: x
y = f (x, d) (3)
L
The solution is the modification of the system from the so Welding
direction
called open-loop form (left side of Fig. 8) into the so called
close-loop form (right side of Fig. 8). In close loop form
Fig. 9. Seam tracking system [32]
the output y is monitored (measured) and fed back to the
controller. The controller compares the actual values y with uses structured laser light which reflects from the weld seam
the desired ones ydes . Based on this difference the input to the and forms an image in the image sensor. A typical image
object is calculated in such a way that the difference between obtained by this method is shown in the right part of Fig. 9.
the actual values y and the desired ones ydes (the so called The lowest point in the image is very simple to determine
error signal) decreases or ideally equals zero. So the control by the most basic image processing algorithms. This setup is
algorithm programmed in the controller can be written in the used often in recent research [33], [34]. A very similar seam
following way: tracking system on a mobile robot welding platform has also
x = g(ydes − y) (4) been developed [35]. Some recent solutions within this set
Within the domain of arc welding related control systems by don’t even need a source of structured laser light anymore [36],
far the most common control algorithm is the so called PID [23]. Another approach, presented in [37], uses an asymmetry
control algorithm [21], [22], [23], [24]. The other commonly in the weld pool image in order to track the seam.
used control algorithms are based on fuzzy logic [25], [26], A system presented in [38] is also given in this subsection
neural networks [27], [28], [29], sliding mode control [59] and despite not being a seam tracking system. It uses a linear CCD
others [30]. sensor to determine the type of weld joint.

III. A RC W ELDING C ONTROL S YSTEMS B. Temperature measurement

From control theory point, the main problem, when design- The quality of a weld joint is strongly related to both
ing a control system is the fact, that most of the components the spatial distribution and temporal history of temperature
of the output vector y in Eq. 3 cannot be measured. Weld in the weld region during the welding process. Therefore
strength, which is the most important, can for example only a control system which would be able to control the tem-
be determined by destructive testing after the weld is already perature would clearly be beneficial. Such a system, where
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
temperature is being measured using a commercially available resolution images of 256 x 256 pixel size. A typical image
infrared pyrometer with a 3-mm target diameter focused on obtained in this way is shown in Fig. 12. As camera is fixed
the trailing weld bead, 2.5 cm from the tip of the electrode X
is for example presented in [39] and shown in Fig. 10. The
Bead

(X1,Y1) (X2,Y1)
Pyrometer

08H
(30, Y2)

Torch Y

Torch
Temperature
Measurement Electrode
Point 2.5 cm

Welding direction
Seam

Workpiece

Fig. 12. Image obtained from a system shown in Fig. 11

Fig. 10. A possible temperature measurement setup [39]
to the torch, the position of the arc in the image tells us if the
manipulated variable is the welding voltage. A camera based torch is either too high or too low. The system is also capable
system used for that purpose have also already been developed of determining the weld pool size. This is done by counting the
[40]. Temperature measurements obtained with this system number of pixels (and therefore area) with a certain brightness.
are within 10% of those measured by a calibrated two-color When the brightness reaches the lowest level of the weld pool,
optical pyrometer despite the fact that an inexpensive and the area typically increases more than 1.5 times. Therefore, it
readily available video camera was used. is possible to determine the weld pool area by counting the
number of all the pixels that are brighter than the lowest level
C. Image processing based control systems of the weld pool. Other (later) systems use similar setup for
control purposes.
The idea of image processing based arc welding control A system using a camera viewing the weld pool region at
system lies in the fact that an experienced welder can judge the 45◦ angle is presented in [42]. Contrary to the previous
the quality of welding process quite well. His main sense used system, here a pulsed laser with a peak intensity that is
in this task is vision. A typical setup of an early version of much higher than the intensity of the arc light is used for
image processing based arc welding control system is shown illumination. The laser is projected on to the weld pool and
schematically in Fig. 11 [41]. In this specific case the process its surrounding area. Because the weld pool surface is mirror-
like, the projected laser light is specularly reflected from the
TIMER
surface of the weld pool. The corresponding image of the weld
pool is therefore dark. But from the solid area surrounding
S/Š the weld pool, the projected laser is diffusely reflected. The
PROGRAMMABLE
PERSONAL
COMPUTER INTERFACE
FUNCTION
CURRENT
SOURCE
resultant image is bright. Thus, the weld pool can be very
GENERATOR
clearly viewed and estimated. Further development of this
I.M.SET WELDING
approach is given in [43]. A problem of too bright images
CURRENT
and therefore a need for a laser illumination can be eliminated
I.M.RESET
D/C DC MOTOR if images at the right moment on the current wavefrom are
taken [28]. A more advanced approach using a camera and
filters with the application of active contours is presented in
[44]. Another interesting approach is outlined in [45]. It is well
TIMING
PULSE
known that the spectroscopic analysis of the plasma spectra
GENERATOR
produced during the welding process is a possible technique to
IMAGE
CAMERA
monitor the quality of the resulting weld seams. The analysis
MEMORY
of specific emission lines and the subsequent estimation of
WELD POOL ARC
the electronic temperature profile offers a direct correlation
between this parameter and the corresponding weld seams.
Systems using a structured laser light reflection (similar
Fig. 11. A typical setup of an early version of image processing based arc
welding control system [41] as in seam tracking) in order to determine the weld pool
parameters are presented in [46] and [47]. A more advanced
under consideration is the GMAW. For image acquisition a system using a set of laser lines is shown in Fig. 13 [48].
monochrome CCD camera which is capable of obtaining 16bit Due to the change of the welding pool surface, the projected
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
 Z Camera C1 in [57]. There are also systems based on acoustic signals. A
d
real-time control of welding penetration during robotic GTAW
Laser
dynamical process by audio sensing of arc length is presented
Camera C2 in [24].

Weld pool
Diffusive X
E. Other systems
plane P1

Laser lines There are also some other interesting control systems. When
arc welding is used in additive manufacturing, wire twisting
Horizontal
plane P0 is an important disturbance. An image processing system
Y addressing this problem is presented in [58]. A complex
Fig. 13. Weld pool observed by the reflection of structured light
control system used for an arc welding based shaped metal
deposition (SMD) system that is an innovative method for the
manufacturing of metal objects is outlined in [60]. Both visual
straight laser lines will be distorted differently. Accurately and acoustic signals were taken in consideration for control
computing the equations of these laser lines is necessary for purposes.
the subsequent weld pool shape estimation. A similar approach
using a 20mW laser with variable focus in order to project IV. C ONCLUSION
a 19-by-19 dot matrix structured light pattern on the weld Arc welding might be considered as a mature joining
pool region is presented in [8]. The measured characteristic process. The paper however showed that there is still an
parameters are then used to estimate the backside bead width. enormous amount of research going on in relation with arc
An adaptive neuro-fuzzy inference system (ANFIS) is used for welding control. This might be attributed to the fact that it is
this purpose. An even more advanced system is presented in probably by far the most important welding process.
[49]. It uses three cameras and structured laser pattern to get The main conclusion based on the conducted overview
an accurate reconstruction of weld pool. could easily be that machine vision / image processing based
control systems are becoming the most hot topic. In arc
D. Arc Length based control systems welding control where process feedback is needed images may
For most welding situations, it is desirable to keep the be required of the arc, weld pool, underbead, surface bead, or
arc length constant to produce a high-quality weld [50]. The weld preparation. Observation of the arc length and electrode
common method used is to keep the arc voltage constant. In stickout can also give important information about the process.
order to explain the rationale behind the arc length control the Arc intensity is also affected by various process parameters
power supply is converted into the equivalent RL circuit as and disturbances. The weld pool also emits radiation in the
[51] visual part of the EM spectrum. As intensity levels are much
di lower than those emitted by the arc itself, it is often used
Up = L + Ri + Ua (5) to asses the weld quality. This is also attributed to the fact
dt
that this is where the weld strength comes from. So, it can
where Up is the voltage of the power source, L the inductance,
be concluded that there are still many possibilities for future
R the resistance and i the welding current and Ua the arc
research in the area of arc welding control, especially control
voltage. The arc voltage is determined by equation [52]:
system based on visual information, because the equipment
Ua = ka la + kp i + Uc (6) needed is getting more an more inexpensive.

where la is the arc length, and ka , kp and Uc the parameters

R EFERENCES
of the arc characteristics. This equation is an approximation
as the voltage drop in the arc is still being researched [53]. [1] S. A. Vendan, L. Gao, A. Garg, P. Kavitha, G. Dhivyasri, and S. Rahul,
Interdisciplinary Treatment to Arc Welding Power Sources. Springer,
The basic conclusion that is however always valid is that arc 2018.
length can be controlled by arc voltage. This fact was for [2] S. Ozcelik and K. Moore, Modeling, sensing and control of gas metal
example used in a control system presented in [50]. A special arc welding. Elsevier, 2003.
[3] J. F. Lancaster, “The physics of welding,” Physics in technology, vol. 15,
care is taken of arc ignition and arc termination. A GTAW no. 2, p. 73, 1984.
feedback system in which the actual arc voltage is compared [4] J. Zhang and B. L. Walcott, “Adaptive interval model control of arc
to the reference voltage and then the welding torch position welding process,” IEEE transactions on control systems technology,
vol. 14, no. 6, pp. 1127–1134, 2006.
is being controlled is presented in [54]. An improved version [5] J.-M. Wang and S.-T. Wu, “A novel inverter for arc welding machines,”
of this approach using gain scheduling is detailed in [55]. As IEEE Transactions on Industrial Electronics, vol. 62, no. 3, pp. 1431–
parameter ka in Eq. 6 is not really constant, an adaptive voltage 1439, 2015.
[6] A. Navarro-Crespin, R. Casanueva, and F. J. Azcondo, “Performance
control system is proposed in [56] and a predictive functional improvements in an arc-welding power supply based on resonant in-
controller based on ARMarkov model structure in [30]. verters,” IEEE Transactions on Industry Applications, vol. 48, no. 3, pp.
Optical systems used to determine the arc length have also 888–894, 2012.
[7] S. Narula, B. Singh, and G. Bhuvaneswari, “Improved power-quality-
been introduced. A precise system for arc length sensing in based welding power supply with overcurrent handling capability,” IEEE
GTAW based on arc ligth spectrum is for example presented Trans. Power Electron, vol. 31, no. 4, pp. 2850–2859, 2016.
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
 [8] Y. K. Liu and Y. M. Zhang, “Model-based predictive control of weld [31] C. Umeagukwu, B. Maqueira, and R. Lambert, “Robotic acoustic seam
penetration in gas tungsten arc welding,” IEEE Transactions on Control tracking: System development and application,” IEEE Transactions on
Systems Technology, vol. 22, no. 3, pp. 955–966, 2014. Industrial Electronics, vol. 36, no. 3, pp. 338–348, 1989.
[9] R. Casanueva, F. J. Azcondo, F. J. Dı́az, and C. Brañas, “Tig welding [32] M. Kawahara, “Tracking control system using image sensor for arc
machines,” IEEE Industry Applications Magazine, vol. 17, no. 5, pp. welding,” Automatica, vol. 19, no. 4, pp. 357–363, 1983.
53–58, 2011. [33] Z. Fang, D. Xu, and M. Tan, “A vision-based self-tuning fuzzy controller
[10] H.-Y. Huang, “Effects of shielding gas composition and activating flux for fillet weld seam tracking,” IEEE/ASME Transactions on mechatron-
on gtaw weldments,” Materials & Design, vol. 30, no. 7, pp. 2404–2409, ics, vol. 16, no. 3, pp. 540–550, 2011.
2009. [34] A. F. Villán, R. G. Acevedo, E. A. Alvarez, A. C. López, D. F. Garcı́a,
[11] G. E. Cook, “Robotic arc welding: Research in sensory feedback R. U. Fernández, M. J. Meana, and J. M. G. Sánchez, “Low-cost system
control,” IEEE Transactions on Industrial Electronics, no. 3, pp. 252– for weld tracking based on artificial vision,” IEEE transactions on
268, 1983. industry applications, vol. 47, no. 3, pp. 1159–1167, 2011.
[12] Y. Chu, S. Hu, W. Hou, P. Wang, and S. Marin, “Signature analysis for [35] L. Zhang, Q. Ye, W. Yang, and J. Jiao, “Weld line detection and tracking
quality monitoring in short-circuit gmaw,” Weld. J, vol. 83, no. 12, pp. via spatial-temporal cascaded hidden markov models and cross struc-
336–343, 2004. tured light,” IEEE Trans. Instrumentation and Measurement, vol. 63,
[13] M. I. Khan, Welding science and technology. New Age International, no. 4, pp. 742–753, 2014.
2007. [36] X. Gao, D. Ding, T. Bai, and S. Katayama, “Weld-pool image centroid
[14] J. Johnson, N. Carlson, H. Smartt, and D. Clark, “Process control of algorithm for seam-tracking vision model in arc-welding process,” IET
gmaw: sensing of metal transfer mode,” Welding Journal, vol. 70, no. 4, image processing, vol. 5, no. 5, pp. 410–419, 2011.
p. 91, 1991. [37] K.-Y. Bae, T.-H. Lee, and K.-C. Ahn, “An optical sensing system for
[15] P. Praveen, P. Yarlagadda, and M.-J. Kang, “Advancements in pulse gas seam tracking and weld pool control in gas metal arc welding of steel
metal arc welding,” Journal of Materials Processing Technology, vol. pipe,” Journal of Materials Processing Technology, vol. 120, no. 1-3,
164, pp. 1113–1119, 2005. pp. 458–465, 2002.
[16] M. Rouffet, M. Wendt, G. Goett, R. Kozakov, H. Schoepp, K. Weltmann, [38] F. He, Y. Su, K. Wang, and J. Pan, “Pattern recognition of weld joint
and D. Uhrlandt, “Spectroscopic investigation of the high-current phase with linear ccd sensor,” Chinese Journal of Mechanical Engineering,
of a pulsed gmaw process,” Journal of Physics D: Applied Physics, vol. 5, no. 4, pp. 290–296, 1992.
vol. 43, no. 43, p. 434003, 2010. [39] T. Santos, R. Caetano, J. M. Lemos, and F. J. Coito, “Multipredictive
[17] Q. L. Wang, P. J. LI, and M. Naka, “Arc light sensing of droplet transfer adaptive control of arc welding trailing centerline temperature,” IEEE
and its analysis in pulsed gmaw process,” Quarterly Journal of the Japan Transactions on Control systems technology, vol. 8, no. 1, pp. 159–169,
Welding Society, vol. 15, no. 3, pp. 415–424, 1997. 2000.
[18] K. Li and Y. Zhang, “Interval model control of consumable double- [40] D. Nishar, J. Schiano, W. Perkins, and R. Weber, “Adaptive control of
electrode gas metal arc welding process,” IEEE Transactions on Au- temperature in arc welding,” IEEE Control Systems, vol. 14, no. 4, pp.
tomation Science and Engineering, vol. 7, no. 4, pp. 826–839, 2010. 4–12, 1994.
[19] R. P. Reis, A. Scotti, J. Norrish, and D. Cuiuri, “Investigation on welding [41] K. Ohshima, M. Yamamoto, T. Tanii, and S. Yamane, “Digital control
arc interruptions in the presence of magnetic fields: welding current of torch position and weld pool in mig welding using image processing
influence,” IEEE Transactions on Plasma Science, vol. 40, no. 3, pp. device,” IEEE transactions on industry applications, vol. 28, no. 3, pp.
870–876, 2012. 607–612, 1992.
[20] ——, “Investigation on welding arc interruptions in the presence of
[42] Y. Zhang, R. Kovacevic, and L. Li, “Characterization and real-time
magnetic fields: arc length, torch angle and current pulsing frequency
measurement of geometrical appearance of the weld pool,” International
influence,” IEEE Transactions on Plasma Science, vol. 41, no. 1, pp.
Journal of Machine Tools and Manufacture, vol. 36, no. 7, pp. 799–816,
133–139, 2013. 1996.
[21] D. E. Henderson, P. Kokotovic, J. Schiano, and D. Rhode, “Adaptive
[43] R. Kovacevic and Y. Zhang, “Real-time image processing for monitoring
control of an arc welding process,” IEEE Control Systems, vol. 13, no. 1,
of free weld pool surface,” Journal of Manufacturing Science and
pp. 49–53, 1993.
Engineering, vol. 119, no. 2, pp. 161–169, 1997.
[22] S.-B. Chen, Y. Zhang, T. Lin, T. Qiu, and L. Wu, “Welding robotic
[44] J. Liu, Z. Fan, S. I. Olsen, K. H. Christensen, and J. K. Kristensen,
systems with visual sensing and real-time control of dynamic weld pool
“Boosting active contours for weld pool visual tracking in automatic arc
during pulsed gtaw,” International Journal of Robotics and Automation,
welding,” IEEE Transactions on Automation science and engineering,
vol. 19, no. 1, pp. 28–35, 2004.
vol. 14, no. 2, pp. 1096–1108, 2017.
[23] Y. Xu, H. Yu, J. Zhong, T. Lin, and S. Chen, “Real-time seam tracking
control technology during welding robot gtaw process based on passive [45] J. Mirapeix, P. Garcı́a-Allende, A. Cobo, O. Conde, and J. López-
vision sensor,” Journal of Materials Processing Technology, vol. 212, Higuera, “Real-time arc-welding defect detection and classification with
no. 8, pp. 1654–1662, 2012. principal component analysis and artificial neural networks,” NDT & e
International, vol. 40, no. 4, pp. 315–323, 2007.
[24] N. Lv, J. Zhong, H. Chen, T. Lin, and S. Chen, “Real-time control
of welding penetration during robotic gtaw dynamical process by [46] Y. M. Zhang, R. Kovacevic, and L. Li, “Adaptive control of full
audio sensing of arc length,” The International Journal of Advanced penetration gas tungsten arc welding,” IEEE Transactions on Control
Manufacturing Technology, vol. 74, no. 1-4, pp. 235–249, 2014. Systems Technology, vol. 4, no. 4, pp. 394–403, 1996.
[25] H. Shisheng, H. Jianfeng, and S. Yonglun, “Design of fuzzy logic [47] Y. Li, Y. F. Li, Q. L. Wang, D. Xu, and M. Tan, “Measurement and
controller for bead width of gtaw [j],” Transactions of the China Welding defect detection of the weld bead based on online vision inspection,”
Institution, vol. 2, 1996. IEEE Transactions on Instrumentation and Measurement, vol. 59, no. 7,
[26] D. Zhao, S. Chen, L. Wu, M. Dai, and Q. Chen, “Intelligent control for pp. 1841–1849, 2010.
the shape of the weld pool in pulsed gtaw with filler metal,” Welding [48] Z. Wang, “Monitoring of gmaw weld pool from the reflected laser
Journal(USA), vol. 80, no. 11, p. 253, 2001. lines for real-time control,” IEEE Transactions on industrial informatics,
[27] C. Wu, T. Polte, and D. Rehfeldt, “Gas metal arc welding process vol. 10, no. 4, pp. 2073–2083, 2014.
monitoring and quality evaluation using neural networks,” Science and [49] ——, “An imaging and measurement system for robust reconstruction
technology of welding and joining, vol. 5, no. 5, pp. 324–328, 2000. of weld pool during arc welding,” IEEE Transactions on industrial
[28] S.-B. Chen, Y. Zhang, T. Qiu, and T. Lin, “Robotic welding systems with electronics, vol. 62, no. 8, pp. 5109–5118, 2015.
vision-sensing and self-learning neuron control of arc welding dynamic [50] Y. Liu, G. E. Cook, R. J. Barnett, and J. F. Springfield, “Pc-based arc
process,” Journal of Intelligent and Robotic Systems, vol. 36, no. 2, pp. ignition and arc length control system for gas tungsten arc welding,”
191–208, 2003. IEEE transactions on industry applications, vol. 28, no. 5, pp. 1160–
[29] J. Xiong, G. Zhang, J. Hu, and L. Wu, “Bead geometry prediction for 1165, 1992.
robotic gmaw-based rapid manufacturing through a neural network and a [51] J. Choi, J. Lee, and C. Yoo, “Simulation of dynamic behavior in a gmaw
second-order regression analysis,” Journal of Intelligent Manufacturing, system,” Welding Journal, vol. 80, no. 10, pp. 239S–246S, 2001.
vol. 25, no. 1, pp. 157–163, 2014. [52] P. Jiluan, Arc welding control. Elsevier, 2003.
[30] M. M. Anzehaee and M. Haeri, “Welding current and arc voltage control [53] M. L. Rossi, V. Ponomarev, and A. Scotti, “Heat exchange and voltage
in a gmaw process using armarkov based mpc,” Control Engineering drop in welding arc column,” IEEE Transactions on Plasma Science,
Practice, vol. 19, no. 12, pp. 1408–1422, 2011. vol. 44, no. 10, pp. 2446–2454, 2016.
Company Registration No.: 201427293E
7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com
[54] J. B. Bjorgvinsson, G. E. Cook, and K. Andersen, “Microprocessor-
based arc voltage control for gas tungsten arc welding using gain
scheduling,” IEEE transactions on industry applications, vol. 29, no. 2,
pp. 250–255, 1993.
[55] D. J. Leith and W. E. Leithead, “On microprocessor-based arc voltage
control for gas tungsten arc welding using gain scheduling,” IEEE
Transactions on Control Systems Technology, vol. 7, no. 6, pp. 718–
723, 1999.
[56] P. Koseeyaporn, G. E. Cook, and A. M. Strauss, “Adaptive voltage con-
trol in fusion arc welding,” IEEE Transactions on Industry Applications,
vol. 36, no. 5, pp. 1300–1307, 2000.
[57] P. Li and Y. Zhang, “Precision sensing of arc length in gtaw based on
arc light spectrum,” Journal of Manufacturing Science and Engineering,
vol. 123, no. 1, pp. 62–65, 2001.
[58] Q. Zhan, Y. Liang, J. Ding, and S. Williams, “A wire deflection
detection method based on image processing in wire+ arc additive
manufacturing,” The International Journal of Advanced Manufacturing
Technology, vol. 89, no. 1-4, pp. 755–763, 2017.
[59] A. K. Paul, “Robust product design using sosm for control of shielded
metal arc-welding (smaw) process,” IEEE Transactions on Industrial
Electronics, vol. 63, no. 6, pp. 3717–3724, 2016.
[60] F. Bonaccorso, L. Cantelli, and G. Muscato, “An arc welding robot
control for a shaped metal deposition plant: Modular software interface
and sensors,” IEEE Transactions on Industrial Electronics, vol. 58, no. 8,
pp. 3126–3132, 2011.

Company Registration No.: 201427293E

7030 Ang Mo Kio Avenue 5#04-155
Northstar @ AMK Singapore (569880)
www.Whioce.com