Professional Documents
Culture Documents
1 Introduction
Welding processes have been developed into an auto- energy domain process that is essential to many types of
mated operation over the past decades. In the beginning manufacturing. Weld quality features such as final metal-
stages of automation, the welding machines were success- lurgy and joint mechanics are typically not measurable on-
fully mechanized. Control systems for the welding torch line for control; thus, some indirect way of controlling the
motion (e.g., joint tracking and robotic welding) are now weld quality is necessary. A comprehensive approach to in-
commercially available. However, control of the welding process control of welding includes both geometric features
process itself is also very important so that full automation of the bead (such as the cross section features width, depth
of the process and, possibly, unmanned operation may be and height) and thermal characteristics (such as the heat-
achieved. Unfortunately, process control systems have not affected zone width and cooling rate). The definitions of
been fully developed for many reasons such as complexity these features are illustrated in Fig. 1. Geometric features
of the welding process and lack of reliable sensors; therefore, can be related to the basic mechanics of the joint, while the
much research is underway in order to solve these problems. thermal features can indicate the final microstructure of the
Gas Metal Arc Welding (GMAW) is a complex, multi- material. A general multi-input, multi-output approach to
control is illustrated in Fig. 2.
Contributed by the Dynamic Systems and Control Division for publication in Nonlinearity of welding has been confirmed by several
the JOURNAL OF DYNAMIC SYSTEMS, MEASUREMENT, AND CONTROL. Manuscript
received by the DSCD August 26, 1993; revised manuscript received July 15, process modeling investigations. For Gas Tungsten Arc
1993. Associate Technical Editor: J. Stein. Welding (GTAW) process, Hardt et al. (1985) developed a
Journal of Dynamic Systems, Measurement, and Control SEPTEMBER 1994, Vol. 116/405
>i(z)\
J2U).
Welding .-
(bn + *i2Z )z » (bl3 + bHz'l)z~ai2-i
Depth
Table
\+auZX + a\2z2 \+anzx + a\2Z2 u\{z)
Temperatures
(b2l + &22Z )z 21 (b23 + b24z~i\„-rfi
l
)z-"22 U2(z)J
Fig. 3 Schematic for welding setup and measurement devices 1 + a2\Z~l +a22z2 1 + a2\z x
+ a22z .
(1)
cases. Some guidelines for designing the adaptive control where yi(z) and y2(z) are the width and depth, u\(z) and u2(z)
system are also suggested. are the wire feedrate (/) and inverse travel speed or inverse
velocity (v_1), and dy is the time delay between the ith output
2 Process Modeling for Control and j'th input. Note that the inverse travel speed is used
Among many possible outputs shown in Fig. 2, bead width instead of the travel speed in Eq. (1). This is because the
and depth, are chosen as process outputs in this research. leading coefficients b 13 and b23 corresponding to u2(z) will
Because the process inputs must be able to regulate the have positive sign with the inverse travel speed as input;
chosen outputs, wire feedrate (f) and travel speed of the both width and depth increase as the inverse travel speed
torch (v) (i.e., torch velocity) are selected as inputs. Even increases (i.e., the travel speed decreases). This convention
though these two inputs have similar effects on the outputs will turn out to be very useful in controller design later.
and cause strong coupling of I/O pairs, they are considered It is noted that each output has been modeled to have
the best choices under the current configuration of GMAW common denominator dynamics; two process inputs have
setup. Note that the wire feedrate regulates the welding similar effects on each output from a physical point of view
current, and thus the heat input into the weld. because both inputs are regulating the outputs through the
In order to investigate the dynamics of the process, a amount of the heat input to the weldment. This common
series of open-loop welding experiments were performed dynamics also simplifies the subsequent analysis and design;
using the welding setup in Fig. 3. The welding table with if each I/O pair had different second-order dynamics, the
the workpiece clamped to it was set up to move in the resulting transfer function matrix would include a fourth-
direction opposite to the welding direction, while the torch order polynomial. This idea of common dynamics was exam-
and measurement devices remain stationary. The experi- ined in a series of open-loop welding experiments using a
ments were conducted based on a bead-on-plate welding. pseudo-random binary sequence (PRBS) input. The PRBS's
The primary metals were low carbon steels which are 6.35 for the feedrate and travel speed were designed such that
mm ('/j in) thick. The top surface of the work was ground the minimum pulse duration is chosen as twice the sampling
to remove oxides, while the back side was cleaned chemi- time so as to put more energy on the low frequency range
cally in order to improve temperature measurement. An Ar and the levels of pulses are chosen as a compromise between
+ 2 percent oxygen gas mixture was used as a shielding gas the linearity consideration and a high signal-to-noise ratio.
for this experiment. These signals combined with a recursive least-squares
The weld bead depth, which is the key geometric attribute method were used to identify the process parameters for
of a major class of welds, is very difficult to directly measure, both SISO and MIMO cases. With 20 cm/s of nominal
but a robust method to estimate the depth using temperature feedrate and 7 mm/s of nominal travel speed, the following
measurement was developed (Song and Hardt, 1992). The two TFM models were identified:
estimator employs a 3-D analytical heat conduction relation-
ship with a Gaussian distributed heat source. The algorithm is
designed such that the temperature measurement, nonlinear
parameter identification based on a combined Gauss-Newton y2(z)J
and Steepest Descent method, and estimation of the depth (0.13-0.10z -1l\,-2
)z (2.24-1.85z_1)z""2
are performed in real-time. Destructive measurement of the l
\-\.\2z- +Q36z- z
l-1.15z _ 1 + 0.38z"i! «i(z)
weld profile indicates that the method can provide depth 1 1
estimates of acceptable accuracy and speed for real-time (0.03-O.Olz-V (0.82-0.38Z-V «2(z).
control. In addition to the depth estimator, a weld pool width 1 - 1.05z~' + 0.35z"2 1 + 0.98z~' + 0.29z~
estimator using a CCD camera and image analysis was also
developed for real-time control (Song, 1992). (2)
Extensive open-loop step tests have shown that the
GMAW process can be adequately modeled by the second- vi(z)]
order dynamics (Hale and Hardt, 1990a). Since the process V2(Z)
Journal of Dynamic Systems, Measurement, and Control SEPTEMBER 1994, Vol. 116/407
Journal of Dynamic Systems, Measurement, and Control SEPTEMBER 1994, Vol. 116/409
Controller
Parameters Parameter
Desired Estimator
Outputs
Multivariate
Adaptive Welding
Wire Feedrate Process
Controller Width
Travel Speed
7 Depth
10 15 20
Time {sec) ponding to the portions before t = 2 s and after t = 22 s in
Fig. 7(c). As a result, tracking performance for the portion
Fig. 7 Comparison of OSA adaptive controller to nonadaptive control-
lers (a) OSA adaptive controller, (b) PI controller, and (c) nonadaptive between t = 2 and t = 22 s was very poor. This indicates
OSA controller (input; feedrate, output; width, travel speed = 7 mm/s, that the performance of the OSA controller depends strongly
T, = 0.5 s)
upon the accuracy of the parameters; thus, it cannot be
effectively used in a system with time-varying parameters
unless some adaptation mechanism (e.g., on-line parameter
in Fig. 6(b); inputs and outputs are normalized so that they estimation scheme) is also used.
are of about the same magnitude (typically, =1). The over- Disturbance Rejection. In arc welding, various forms of
shoot observed in Fig. 6(a) is significantly reduced and the disturbances may exist (e.g., thickness change, change of
output maintains good command-following. Additionally, material properties, parameter drifts of the welding ma-
the control signal shows a reasonable variation. chines, etc.). The disturbances need to be rejected in most
Command Following. The command following per- cases since they adversely affect the performance of control
formance by the adaptive controller is compared to two systems. Figure 8 shows the disturbance rejection perform-
nonadaptive controllers in Fig. 7 for feedrate-width pair. ance of the OSA adaptive controller for travel speed-width
Figure 1(b) shows the response of the digital PI controller pair. Step changes in the feedrate were used as the distur-
in the following form bance. At the time t = 5 s, the feedrate decreases from 22
to 18 cm/s, thus resulting in a quick decrease in the width
K, output. The control parameters then adapt to the new parame-
u(z) = KP + - e(z) (16)
l-z~ ters; the controller decreases the travel speed to compensate
for the decreased width. A similar trend is observed in the
where Kp and Kj are the proportional and integral gains, disturbance introduced at time t =25 s.
respectively, and e(z) is the error signal defined as y*(z) -
y(z). It is observed that the width response cannot follow 4.2 MIMO Cases. The same control experiments as
the step change in reference command quickly enough, in the SISO case were extended to the two-input, two-output
though the steady-state performance is relatively good. This case. In the actual implementation with the sampling period
slow response is because the PI controller is operated based of 0.5 s, the width (yt) and depth (y2) outputs have 2 units
on the error signal between the reference command and the and 1 unit of time delay (i.e., dt = 2 and d2 = 1 in Eq. (3)),
measured output. Since there are time delays of two samples respectively; the time delays are identified from the open-
(delay of 1 s) and a computational delay of one sample in loop step tests. Therefore, the predictor forms can be ob-
the width control, the control action is always delayed by tained by successive substitutions as follows:
at least three samples. In Fig. 1(b), a time delay of five or
six samples (2.5 to 3.0 s) is observed in the step change. yi(k + 2) = anyi(k - 1) + oiUyi(k - 2)
This delay can be slightly reduced by adjusting the gains
+ 0 n m ( * ) + Pi2»i(* - 1) + Pi3«i(* - 2)
(e.g., increasing the proportional gain Kp), but at the expense
of a possible overshoot. When compared to the PI controller, + Pl4«l(* - 3) + Pl5"2(£) + Pl6"2(£ - 1)
the OSA adaptive controller shows much faster time re-
sponse (Fig. 7(a)). This is because the OSA adaptive con- + Pl7«2(*-2) + Pl8«2(*-3) (17)
troller is based not on the error signal but on the predictor and
form which accounts for the time delays including a compu-
tational delay. In the width control case involving three y2(k + 1) = a2iy2(k - 1) + 0L22y2(k - 2)
samples of time delay, for example, the control signal at + P21«lW + 022"l(£ - 1) + P23"l(£ - 2)
time kTs is commanded to track the reference input at time
(k + 3)7;, thus leading to good tracking for the step change. + P24"2(fc) + ?>25»2(k " 1) + $26U2(k - 2) (18)
Figure 7(c) shows the performance of the nonadaptive where y\(z) and y2(z) are the width and depth, u\(z) and u2(z)
generalized OSA controller. The parameters used for the are the wire feedrate (f) and inverse travel speed or inverse
controller were obtained from the operating condition corres- velocity (v_1). The coefficients a's and P's are obtained by
Journal of Dynamic Systems, Measurement, and Control SEPTEMBER 1994, Vol. 116/411
& 25
5 10 15 20 25 30 35
I 20 Time (sec)
CD
^
0)
"- 15 Fig. 10 Output responses of Indirect OSA adaptive controller
-0.1
0.4 ; „/A/\
^^Vv
0.0
w
1-0-3
^r jsT \
„ , 0 <j*r
J\
0.3 \
5.0
rect approaches; in indirect approach, the process parameters
are obtained first (i.e., the parameters of Eq. (1)), and then
'•
!\S/
12 16 20 24 28 32 12 16 20 24 28 32
the coefficients are computed. From the above predictor Tims Time
forms, the control law can be defined by Eq. (12) with Fig. 11 Parameter estimates for OSA adaptive controller (dotted lines
indicate changes in reference commands)
"3n3i5~ riO" y?(k + :
3o = ,R = , J*(k) =
321 324- .0 r2- •y$(k +
Figure 10 shows the output responses of an indirect adaptive
anvi(fc- 1) + otnyiik - 2) controller. The indirect controller also produces a reasonably
«(«"')y(i) = »
u-i\yi(k - 1) + a22y2(k - 2). good performance, but die steady-state responses are some-
what oscillatory when compared to those of the direct con-
and p ' ^ M A : - 1) troller.
3l2«l(& - 1) + - + Pi 8 w 2 (fc- Figure 11 plots some of the parameter estimates that are
identified by the RLS method. The parameters around time
132 iu\{k- - 1 ) + - + $26U2(k - 2 ) J ' t = 22 s are plotted in order to emphasize the parameter
where r\ and r2 are the control weighting factors for the changes in a step change of a reference command. It is
feedrate and the inverse travel speed, respectively. observed that all parameters change significantly after the
The control weighting factors are also important as in the set point change. The control signals after t = 22 s are
SISO case. With no control weights, either unstable or highly computed based on these changed control parameters and
oscillatory behavior was observed. In addition to the proper thus improves performance of the control system.
magnitudes of the control weights, the proper balance be- In the MIMO case, the control range is also an important
tween two control weights are important as well in the issue (Hale and Hardt, 1990a, Doumanidis and Hardt, 1990,
MIMO case. Since the control weight tends to decrease the and Hardt, 1992). In gas metal arc welding process, two
control signal, excessively large control weight reduces the outputs (i.e., width and depth) are highly coupled partly
corresponding control signal more than the other, as a result, because two inputs (i.e., feedrate and travel speed) have
the whole process is dominated by the other control signal, similar effects on each output. Therefore, there is a limit to
which is undesirable. In most cases, it is desirable to achieve the range which combinations of the two control inputs
equal participation from both control inputs, so that a good can achieve. For example, an increase in die width while
control performance may be accomplished. Therefore, an decreasing or maintaining a constant depth cannot be at-
appropriate choice of the control weighting factors is an tained with any combination of the control inputs. No control
important step in the design of the controller. The proper algorithm can overcome this inherent problem since the
level and balance can be found either by simulation or by limitations are due to the welding process and machines
experiment as indicated in Section 3.4. used. Therefore, the desired bead geometries must be in die
Figure 9 shows the output responses and control inputs achievable control range. The adaptive controller cannot
of direct adaptive controller. With proper control weights, provide good command following performance when de-
both control signals show reasonable variations. Both output sired bead geometries derive greatly from die achievable
responses are shown to follow the reference commands rea- control range.
sonably well even with the step changes in the commands. This output coupling may make the disturbance rejection
10
4r 1992).
10 15 Acknowledgment
> Disturbed
This project was supported by the U.S. Department of En-
ergy under contract number DE-FG02-85ER13331.
%r
References
0 5 10 15 20 25 30 35
Andersen, K., Cook, G. E., Karsai, G., and Ramaswamy, K., 1990, "Artificial
Fig. 12 Simulation of disturbance rejection by OSA adaptive controller Neural Networks Applied to Arc Welding Process Modeling and Control," IEEE
(disturbances; changes in steady-state gains) Transactions on Industry Applications, Vol. 26, No. 5, pp. 824-830.
Astrom, K. J., and Wittenmark, B., 1984, Computer Controlled Systems,
Prentice-Hall.
Astrom, K. J., and Wittenmark, B., 1989, Adaptive Control, Addison Wesley.
difficult. If the disturbance affects one output much more Boo, K. S., and Cho, H. S., 'A Fuzzy Linguistic Control Approach to the
Control of Weld Pool Size in Gas Metal Arc Welding Processes," Welding and
than the other, then it is difficult to reject this disturbance. Joining Processes, ASME WAM, PED-Vol. 51, pp. 73-84.
For example, a certain disturbance tends to increase the Cook, G. E., Andersen, K., and Barrett, R. J., 1989, "Feedback and Adaptive
width with no change in the depth. Then, the control signals Control in Welding," Proceedings of the 2nd International Conference on Trends
in Welding Research, Gatlinburg, TN, ASM International, pp. 891-903.
must decrease the width while maintaining the depth to Dornfeld, D. A., Tomizuka, ML, and Langari, G„ 1982, "Modeling and Adap-
achieve the desired bead geometries. As mentioned before, tive Control of Arc Welding Processes," Measurement and Control for Batch
Manufacturing, ASME WAM, pp. 53-64.
such a combination of the control inputs are hard to find Doumanidis, C. C , and Hardt, D. E., 1989, "A Model for In-Process Control
because of the strong I/O coupling. However, if the distur- of Thermal Properties during Welding," ASME JOURNAL OF DYNAMIC SYSTEMS,
MEASUREMENT, AND CONTROL, Vol. I l l , No. 1, pp. 40-50.
bance affects both outputs and they are in the achievable Doumanidis, C. C , and Hardt, D. E., 1991, "Multivariable Adaptive Control
control range, then the adaptive controller can reject it by of Thermal Properties During Welding," ASME JOURNAL OF DYNAMIC SYSTEMS,
adapting to new process parameters as in the SISO case. MEASUREMENT, AND CONTROL, Vol. 113, No. 1, pp. 82-92.
Fortescue, T. R., Kershenbaum, L. S., and Ydstie, B. E., 1981, "Implementa-
This can be shown in Fig. 12. In this simulation, the gain tion of Self-Tuning Regulators with Variable Forgetting Factors," Automatica,
disturbances (b's in Eq. (1)) were injected as step functions Vol. 17, pp. 831-835.
Goodwin, G. C , and Sin, K. S., 1984, Adaptive Filtering Prediction and
from the nominal values at t = 14 s. These disturbances Control, Prentice-Hall.
make both responses increase from the nominal conditions Goodwin, G. C , Ramadge, P. J., and Caines, P. E., 1980, "Discrete-Time
significantly at the introduction of the disturbances. It is Multivariable Adaptive Control," IEEE Transactions on Automatic Control,
Vol. AC-25, No. 3, pp. 449-456.
noted that the disturbed outputs are in the achievable control Hale, M. B., and Hardt, D. E., 1990a, "Multivariable Geometry Control of
range. The adaptive controller quickly adapts to the new Welding—Part I: Process Modeling," Automation of Manufacturing Processes,
ASME Winter Annual Meeting, DSC-Vol. 22, pp. 1-9.
parameters and commands the proper control inputs to reject Hale, M. B., and Hardt, D. E., 1990b, "Multivariable Geometry Control of
the disturbances. Welding—Part U: Process Control," Automation of Manufacturing Processes,
ASME Winter Annual Meeting, DSC-Vol. 22, pp. 143-151.
Hardt, D. E., Garlow, D. A., and Weinert, J. B„ 1985, "A Model of Full
5 Conclusions Penetration Arc-Welding for Control System Design," ASME JOURNAL OF DY-
NAMIC SYSTEMS, MEASUREMENT, AND CONTROL, Vol. 107, No. 1, pp. 40-46.
Gas metal arc welding process is highly nonlinear and its Hardt, D. E., 1992, "Welding Process Modeling and Redesign for Control,"
Proceedings of Japan/USA Symposium on Flexible Automation Vol. 1, ASME,
dynamics depends on the operating and boundary conditions. pp. 275-281.
It is therefore very difficult to obtain the process model Langari, G., and Tomizuka, M., 1988, "Fuzzy Linguistic Control of Arc
Welding," Sensors and Controls for Manufacturing, ASME WAM, pp. 157-162.
accurately in advance and on-line parameter estimation and Masmoudi, R. A., 1992, "Modeling and Control of Thermal and Geometric
adaptive control algorithm are required. A one-step-ahead Properties in Arc Welding," Ph.D. thesis, Massachusetts Institute of Technology,
adaptive control strategy combined with a recursive least- Cambridge, MA.
Nomura, H., Satoh, Yu., Tohno, K., Satoh, Yo., and Kurotori, M., 1980, "Arc
squares methods has been investigated. Command following Light Intensity Controls Current in SA Welding System," Welding and Metal
and disturbance rejection properties of the adaptive control Fabrication, Vol. 48, pp. 457-463.
Richardson, R. W., Gutow, D. A., and Rao, S. H., 1982, "A Vision Based
system for both SISO and MIMO cases are investigated by System for Arc Weld Pool Size Control," Measurement and Control for Batch
simulations and experiments. The adaptive control system Manufacturing, ASME WAM, November, pp. 65-75.
shows reasonable performance on command following for Smartt, H. B., 1992, "Intelligent Control of Arc Welding," Proceedings of
the 3rd International Conference on Trends in Welding Research, Gatlinburg,
both SISO and MIMO cases. This system also shows good TN, ASM International, June.
disturbance rejection in the SISO case, but is shown to be Song, J.-B., and Hardt, D. E., 1991, "Multivariable Adaptive Control of Bead
Geometry in GMA Welding," Welding and Joining Processes, ASME WAM,
of limited success in the MIMO case due to inherent lack PED-Vol. 51, pp. 123-124.
of output decoupling in the GMAW process. Process modifi- Song, J.-B., 1992, "Multivariable Adaptive Control in GMA Welding Using
a Thermally Based Depth Estimator," Ph.D. thesis, Massachusetts Institute of
cation to decouple the process outputs to a great extent is Technology, Cambridge, MA.
underway (Masmoudi, 1992). Song, J.-B., and Hardt, D. E., 1992, "A Thermally Based Weld Pool Depth
Estimator for Real-Time Control," Proceedings of the 3rd International Confer-
Even though a truly independent control of the outputs ence on Trends in Welding Research, Gatlinburg, TN, ASM International.
is difficult to implement due to a strong output coupling Suzuki, A., Hardt, D. E., and Valavani, L., 1991, "Application of Adaptive
Control Theory to On-Line GTA Weld Geometry Regulation," ASME JOURNAL
inherently existing in the process, a control system for simul- OF DYNAMIC SYSTEMS, MEASUREMENT, AND CONTROL, Vol. 113, No. 1, pp.
taneous control of bead width and depth was successfully 93-103.
Vroman, A. R., and Brandt, H., 1976, "Feedback Control of GTA Welding
implemented in this paper. More important, a closed-loop Using Puddle Width Measurement," Welding Journal, Vol. 57, No. 9, pp.
control system to directly regulate the depth of penetration 742-749.
Journal of Dynamic Systems, Measurement, and Control SEPTEMBER 1994, Vol. 116/413