- Email: [email protected]
Decoupling control scheme for pulsed GMAW process of aluminum
Journal of Materials Processing Technology 212 (2012) 801–807
Contents lists available at SciVerse ScienceDirect
Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec
Decoupling control scheme for pulsed GMAW process of aluminum Lihui Lu a , Ding Fan b,∗ , Jiankang Huang a , Yu Shi b a b
Key Laboratory of Non-ferrous Metal Alloys and processing, The Ministry of Education, Lanzhou University of Technology, Lanzhou 730050, China State Key Laboratory of Gansu Advanced Non-ferrous Metal Materials, Lanzhou University of Technology, Lanzhou 730050, China
a r t i c l e
i n f o
Article history: Received 27 March 2011 Received in revised form 18 September 2011 Accepted 2 November 2011 Available online 6 November 2011 Keywords: GMAW-P of aluminum Welding parameters Control scheme
a b s t r a c t A double-variable decoupling control scheme was proposed for GMAW-P process of aluminum helping to efficiently develop welding procedure. Weld pool width and arc length were both measured through vision sensing in welding process. Weld bead shape was improved by changing the current waveforms to adjust the heat input while the arc length was controlled to stabilize the welding process. An experimental system was developed to sense, observe and control the welding process real-timely. Experiments were conducted to verify the effectiveness of the scheme. The results show that good weld bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without complex metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. This control scheme provides an alternative to obtain proper weld quality for GMAW-P. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Recent trend in the selection of material for the automotive industry shows there has been a transition from conventional materials to light-weight material like aluminum. Praveen and Yarlagadda (2005) have recently shown achieving good quality with aluminum is a challenging job due to its significant differences from conventional materials like steel. Among the aluminum welding process, automated pulsed gas metal arc welding (GMAW-P) has been recognized as an efficient alternative for minimizing defects. In GMAW-P, the welding current is alternatively and periodically varied between base and peak values. The main setting parameters which influence weld quality or wire melting are base current Ib , peak current Ip , base time Tb , peak time Tp and wire feed rate Vwire . As illustrated in Fig. 1, the welding parameters are more numerous than they are in conventional GMAW, and the process is typically more sensitive. Palani and Murugan (2006) have claimed that improper selection of these pulse parameters may cause weld defects including irregular bead surface, lack of fusion, undercuts, burn-back and stubbing-in. Therefore, it is important to select a proper combination of pulse parameters to obtain stable welding processes and better quality welds. However, identifying suitable combinations of welding parameters for use with GMAW-P can be a time-consuming process, involving considerable trial and error, especially for aluminum. Subramaniam et al. (1999) proposed
∗ Corresponding author. Tel.: +86 13321224851. E-mail addresses: [email protected] (L. Lu), [email protected] (D. Fan). 0924-0136/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2011.11.001
a method of identifying power supply pulsing parameters for GMAW-P based on statistical experimental design. A linear wire feed rate model using data on the values of wire feed rate obtained from the experiments conducted using a two level factorial experimental design was developed. However, in the GMAW-P process of aluminum, even if the welding process is stable, the weld bead geometry may be poor. The weld pool width often become wider and wider, even collapse at constant welding parameters in GMAW-P process due to the strong heat accumulation and small surface tension. Joseph et al. (2002) have characterized the differences in welding heat input and weld bead shape that could be produced by the pulsed current waveforms. Hirai et al. (2001) proposed a penetration depth model based on vision sensing. During the welding, a fuzzy controller adjusted the welding current waveform so as to get the desired penetration depth. Therefore, adjusting welding current waveform suitably may be an alternative to obtain suitable heat input and improve the weld bead shape. How to sense the weld pool is a problem. Previous studies show that the use of vision sensing to achieve real-time control for weld pool is an effective means. Chen et al. (2002) designed a weld pool width control test through vision sensing and the results showed that the realtime and precision requirements for detecting and control the weld pool changes of GMAW-P could be met. In GMAW-P, droplets are regularly detached at a fixed frequency and directionally transferred to the workpiece under the influence of the current waveform. The influence of the metal transfer mode on the weld quality and stability is well known for the GMAWP process. Changing of the current waveform obviously increases the difficulty of identifying suitable combinations of welding
Ip : Peak current Ib : Background current Tp : Peak time Tb : Background time Frequency = 1 / (Tp + Tb) Dutycycle = Tp / (Tp + Tb)
Tp
Current (A)
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Current
802
Ip Tb Ib
Time
High energy pulse Low energy pulse Time
Ta
Tb
Fig. 1. Current waveform of GMAW-P.
T parameters. In GMAW-P process of aluminum, the coupling effect of welding parameters is obvious. Changing one welding parameter may cause variation of other parameters and every parameter is influenced by all other parameters. Ideal control model for aluminum welding should be ensuring the stability of welding process while adjusting the current waveform to improve weld bead shape. Therefore, to obtain a good weld formation and high welding quality on the basis of ensuring the welding process stable, multiinformation sensing and multi-variable decoupling control should be adopted for dynamic process of GMAW-P of aluminum. In this paper, a double-variable decoupling control scheme was proposed. The weld bead shape was improved by changing the current waveforms to adjust the heat input while the arc length was controlled to stabilize the welding process; hence, both the weld bead shape and stability of the welding process were improved. The suitable combinations of welding parameters were identified in the control process and automatically avoiding trial and error.
Fig. 2. Current waveform of double-pulsed GMAW welding.
Niederlinski index (NI) is obtained according to the steady-state gain matrix. NI =
˚
2
=
0.1845 × 0.9594 − 0.96 × (−0.24) = 2.3017 > 0 0.1845 × 0.9594
˚ii
i=1
(3)
The second amplification factor can be calculated by the first amplification factor. There is:
P=
0.4247 1.6975 −0.4244 2.2080
RGA can be obtained as follows:
0.4345 0.5655 0.5655 0.4345
2. Coupling analysis for control model
=
According to the identification test, the transfer function of TITO variable control system was obtained as follows:
The coupling index D is obtained after calculation:
⎡
yw yL
=
G11
G12
G21
G22
0.1845
⎢ 5.0459s + 1 ⎣ −0.24 0.0242s + 1
ı Vwire
e−1.1364s e−0.0253s
D=
=
⎤
0.96 e−0.4174s ı ⎥ 1.1691s + 1 ⎦ 0.9594 Vwire −0.4928s e 0.2663s + 1
(1)
where s is differential operator; G11 is the transfer function of duty cycle ı and weld width yw ; G12 is the transfer function of wire feed speed Vwire and weld pool width yw ; G21 is the transfer function of duty cycle ı and wire extension yL ; G22 is the transfer function of wire feed speed Vwire and wire extension yL . In the welding process, keep contact-tube-to-work distance (CTWD) constant, namely the arc length can be expressed by wire extension and a constant. Therefore, coupling degree of the control model can be analyzed through Eq. (1). For MIMO control system, relative gain array (RGA) theory can be used to determine the degree of coupling. First, the first amplification factor ˚ and the second amplification factor P must be determined. The first amplification factor is the steady-state gain matrix:
˚=
0.1845 −0.24
0.96 0.9594
(2)
(4)
0.5655 × 0.5655 = 1.6939 > 1 0.4345 × 0.4345
(5)
(6)
As for the above control system, it can be judged that system stability can be achieved using a controller with integrator, but the coupling index does not meet the requirements. Change of single input signal may cause changes of multiple outputs, and each output is affected by more than one input. Therefore, decoupling design can only be taken to eliminate or reduce the coupling effect among the parameters. 3. Control scheme Double-pulsed GMAW is that the welding current waveform changes from a group of high-energy pulses (large average current) to a group of low-energy pulses (small average current) alternately, as shown in Fig. 2. T is a double-pulse period; Ta is high-energy pulse time in a double-pulse period; Tb is low-energy pulse time in a double-pulse period. Ta /T (the ratio of high-energy pulse time in a period) is defined as double-pulse duty cycle. The double-pulsed GMAW technique is a variation of the PGMAW. Tong and Tomoyuki (2001) discussed the double-pulsed GMAW technique and claimed that this welding process provided beautiful scaly bead, improved gap-bridging ability for lap joint, restrained blowholes for formation, refined grain size, and decreased crack sensibility. Further, Silva and Scotti (2006) proved
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 3. Route of control scheme.
803
Fig. 5. Typical image captured by vision sensing.
that the double-pulsed GMAW, in spite of having theoretically higher potential for porosity generation, did not increase the porosity susceptibility in aluminum welding when compared with the P-GMAW. Therefore, the control scheme was designed based on double-pulsed GMAW of aluminum. In control process, change of weld pool width can be reflected through vision sensing system, and then the weld pool width can be controlled by changing the double-pulse duty cycle to adjust the heat input. In the control process, the current waveform is constantly changeable. However, as is known from the analysis of control model, GMAW-P process of aluminum is a nonlinear and strong coupling process, by improving one parameter to obtain a satisfactory weld quality is very difficult. Therefore, on the basis of single-variable control of weld pool width, a double-variable decoupling control method is proposed adopting compensation decoupling. In order to ensure the accuracy and avoid the signal interference from droplet transfer, in welding process, weld pool width and arc length were both obtained through vision sensing. On the one hand, control for weld pool width was realized by changing double-pulse duty cycle ı to adjust heat input to improve weld formation. However, change of single parameter can easily lead to instability of the welding process. So on the other hand, arc length was controlled simultaneously by changing wire feed speed Vwire to stabilize the welding process. Therefore, good weld formation can be realized on the basis of the stability of welding process. Route of control scheme is shown in Fig. 3.
4. Experiments 4.1. Experimental setup The experimental setup used in this study is shown in Fig. 4. Through-the-arc sensing of the welding current, arc voltage and CCD camera of welding zone image acquisition are assembled in conjunction with each other to study the control scheme during GMAW-P process of aluminum. The experimental setup consisted of four parts: (1) the welding system including the DALEX VIRIO MIG-400L-digital power source, welding gun, welding table and its moving-control unit, (2) rapid prototyping control system consisting two industrial PC connected through TCP/IP network, one as a central control computer and the other as xPC target machine, (3) the data acquisition and output system including sensors for welding current and arc voltage, PCL-812PG, NI PCI-6221 data acquisition card and PCL-728D/A data output card, (4) vision sensing system including CCD camera and NI PCI-1405 image acquisition card. The arc voltage was measured between the contact tip of the welding gun and the fixture. This allowed voltage measurements as close to the arc as possible to maximize the measurement accuracy of the arc voltage. The welding current was measured with a Hall sensor, which was attached to the ground. A data acquisition card in combination with the xPC target machine was used to acquire
Fig. 4. Schematic diagram of experimental setup.
804
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 6. Image processing for weld pool. (a) Original weld pool image, (b) binaryzation, (c) noise elimination, (d) erosion, (e) dilation and (f) edge detection.
voltage and current signals and display the results in a graphical format. The sampling rate was 10 kHz. Based on the aforementioned hardware, special control software was developed to synchronize the running of the welding system, the vision sensing system and the rapid prototyping control system. Both image-processing and data-analyzing algorithms were designed to display the image and voltage and current waveforms on the computer screen simultaneously.
In control process, welding zone image was acquired through vision sensing system as shown in Fig. 5 and with corresponding image processing algorithm weld pool width and arc length signals were obtained, as illustrated in Figs. 6 and 7. Then the obtained weld pool width and arc length signals were exported to the rapid prototyping control system for control. The flow chart of control program is shown as Fig. 8.
Fig. 7. Image processing for arc length. (a) Gray image of welding zone. (b) Wire location image in x direction. (c) Image cut and enhancement. (d) Pixels in y direction.
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
805
Fig. 8. The flow chart of control program.
Table 1 Welding parameters used in group 1.
4.2. Welding conditions Bead on plate welds were made in the flat position. The filler material used was a 5356 aluminum alloy welding wire with a 1.2 mm diameter. The base material was 6061 aluminum alloy with a thickness of 4 mm. The shielding gas used throughout the experiments was pure argon with a gas flow rate of 25 L/min. CTWD and welding speed was kept constant through out the experiments. The values of CTWD and welding speed used for all the experiments were 20 mm and 15 cm/min. For comparison, three groups of tests were carried out respectively. In the first group, GMAW-P of aluminum was conducted at constant welding parameters as followed in Table 1. Then the
Parameters
Value
Duty cycle Peak current Base current Pulse frequency Average current
50% 200 A 20 A 40 Hz 110 A
single-variable weld width control was done with the method of changing double-pulse duty cycle to adjust the heat input. Table 2 shows the setting conditions of pulsing parameters used for the
Fig. 9. Weld bead at constant welding parameters.
Fig. 10. Weld bead obtained in single-variable weld width control process.
Fig. 11. Weld bead obtained in double-variable decoupling control process.
806
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Table 2 Welding parameters used in group 2 and 3.
Duty cycle Peak current Base current Pulse frequency Average current
Value High-energy pulse
Low-energy pulse
50% 200 A 20 A 40 Hz 110 A
10% 250 A 40 A 40 Hz 61 A
experimentation. Finally, the double-variable decoupling control test was carried out. The double-pulse parameters are also shown in Table 2. In the control process, the change range of double-pulse duty cycle was set to 20–80%. The initial double-pulse duty cycle was set to 80%. Obviously, the mean current changed between 61A and 110A.
Width sizes of weld beads mm
Parameters
20
weld bead in first group weld bead in senond group weld bead in third group
18
16
14
12
10
8
0
20
40
60
80
100
120
140
160
Distance away from starting point of weld bead mm Fig. 12. Weld width sizes of the obtained three weld beads.
Fig. 13. Change of related signals in decoupling control process.
180
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 14. Typical image captured in first 5 s of welding process.
807
twelfth second under the action of fuzzy PID controller in order to offset the heat accumulation of welding with high-energy pulse at the beginning of the welding process. With the change in thermal conditions and decrease of heat input, in order to achieve the set value of weld pool width, double-pulse duty cycle increased. The fluctuation of arc length is generally observed in Fig. 13(c), implying the stability of welding process. The typical welding current and arc voltage in 2 s are shown in Fig. 13(d) and Fig. 13(e). As is illustrated in Fig. 13(d) that current waveform was changing obviously with the change of double-pulse duty cycle. As seen from above tests, the good welding bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. It provides an alternative to obtain proper weld quality for GMAW-P.
4.3. Results and discussion
5. Conclusions
The welds obtained in the three groups of tests are shown in the Figs. 9–11. Fig. 12 shows the weld width sizes of the three welds measured every 0.5 mm. As is illustrated in the Fig. 9 and Fig. 12, the first weld bead became wider and wider obviously. The weld pool width sizes changed from the initial value 9.94 mm to the end value 18.53 mm. This maybe resulted from the strong heat accumulation at constant welding parameters and small surface tension of aluminum. The weld pool width of the second weld bead was improved through the single-variable weld width control; however the weld bead formation was not well and the fish-scale pattern was not obvious. In the control process, though the change of welding current waveform improved the weld bead geometry, it also influenced metal transfer and maybe caused unsuitable combinations of welding parameters, further resulted in the instability of control process. The weld bead obtained through double-variable decoupling control scheme was shown in Fig. 11. The weld bead shape was improved obviously compared with the first and second welds and the spatter was low. Compared with the first weld bead, the heat accumulation effect had been compensated through doublevariable decoupling control scheme. As is illustrated in Fig. 12, the weld width sizes had small fluctuations in welding process. This control scheme also provided beautiful scaly bead. The good weld bead shape and low spatter also imply the stability of welding process adopting double-variable decoupling control scheme. Fig. 13 shows the corresponding signals in double-variable decoupling control process. The change of weld pool width pixels obtained by vision sensing is shown in Fig. 13(a). The pixels of weld pool width kept stable after a stable welding and were consistent with weld width sizes of the third weld bead in Fig. 12. In the first 5 s of welding process, it was difficult to obtain the weld pool width through image processing algorithm due to the strong arc light as shown in Fig. 14. Gradually, wire extension became long and arc light was suppressed, the weld pool width could be sensed successful. In the Fig. 13(b), the double-pulse duty cycle decreased from the
(1) Coupling analysis for TITO control model in GMAW-P of aluminum was done. The strong coupling effect among welding parameters influences the stabilization of control system in welding process. A double-variable decoupling control scheme was proposed for this problem. (2) The developed experimental system makes it possible to sense, observe and control the welding process in real time. (3) Good weld bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without complex metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. This control scheme provides an alternative to obtain proper weld quality for GMAW-P. Acknowledgment This research work is supported by National Nature Science Foundation of China (50805073). References Chen, S.B., Cao, J.M., Xu, C.M., 2002. Visual sensing and real-time control of weld pool dynamics in pulsed GMAW. Transactions of the China Welding Institution 23, 10–14. Hirai, A., Kaneko, Y., Hosoda, T., 2001. Sensing and control of weld pool by fuzzyneural network in robotic welding system. In: Industrial Electronics Conference, Denver, Colorado, USA, pp. 238–242. Joseph, A., Farson, D., Harwig, D., 2002. Influence of GMAW-P current waveforms on heat input and weld bead shape. Science and Technology of Welding and Joining 10, 311–318. Palani, P.K., Murugan, N., 2006. Selection of parameters of pulsed current gas metal arc welding. Journal of Materials Processing Technology 172, 1–10. Praveen, P., Yarlagadda, P.K.D.V., 2005. Meeting challenges in welding of aluminum alloys through pulse gas metal arc welding. Journal of Materials Processing Technology 164–165, 1106–1112. Silva, C.L.M., Scotti, A., 2006. The influence of double pulse on porosity formation in aluminum GMAW. Journal of Materials Processing Technology 171, 366–372. Subramaniam, S., White, D.R., Jones, J.E., 1999. Experimental Approach to Selection of Pulsing Parameters in Pulsed GMAW. Welding Journal, 166–172. Tong, H.J., Tomoyuki, U., 2001. Features of low frequency modulated type pulsed MIG welding process. Welding & Joining 11, 33–35, 40.
Contents lists available at SciVerse ScienceDirect
Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec
Decoupling control scheme for pulsed GMAW process of aluminum Lihui Lu a , Ding Fan b,∗ , Jiankang Huang a , Yu Shi b a b
Key Laboratory of Non-ferrous Metal Alloys and processing, The Ministry of Education, Lanzhou University of Technology, Lanzhou 730050, China State Key Laboratory of Gansu Advanced Non-ferrous Metal Materials, Lanzhou University of Technology, Lanzhou 730050, China
a r t i c l e
i n f o
Article history: Received 27 March 2011 Received in revised form 18 September 2011 Accepted 2 November 2011 Available online 6 November 2011 Keywords: GMAW-P of aluminum Welding parameters Control scheme
a b s t r a c t A double-variable decoupling control scheme was proposed for GMAW-P process of aluminum helping to efficiently develop welding procedure. Weld pool width and arc length were both measured through vision sensing in welding process. Weld bead shape was improved by changing the current waveforms to adjust the heat input while the arc length was controlled to stabilize the welding process. An experimental system was developed to sense, observe and control the welding process real-timely. Experiments were conducted to verify the effectiveness of the scheme. The results show that good weld bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without complex metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. This control scheme provides an alternative to obtain proper weld quality for GMAW-P. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Recent trend in the selection of material for the automotive industry shows there has been a transition from conventional materials to light-weight material like aluminum. Praveen and Yarlagadda (2005) have recently shown achieving good quality with aluminum is a challenging job due to its significant differences from conventional materials like steel. Among the aluminum welding process, automated pulsed gas metal arc welding (GMAW-P) has been recognized as an efficient alternative for minimizing defects. In GMAW-P, the welding current is alternatively and periodically varied between base and peak values. The main setting parameters which influence weld quality or wire melting are base current Ib , peak current Ip , base time Tb , peak time Tp and wire feed rate Vwire . As illustrated in Fig. 1, the welding parameters are more numerous than they are in conventional GMAW, and the process is typically more sensitive. Palani and Murugan (2006) have claimed that improper selection of these pulse parameters may cause weld defects including irregular bead surface, lack of fusion, undercuts, burn-back and stubbing-in. Therefore, it is important to select a proper combination of pulse parameters to obtain stable welding processes and better quality welds. However, identifying suitable combinations of welding parameters for use with GMAW-P can be a time-consuming process, involving considerable trial and error, especially for aluminum. Subramaniam et al. (1999) proposed
∗ Corresponding author. Tel.: +86 13321224851. E-mail addresses: [email protected] (L. Lu), [email protected] (D. Fan). 0924-0136/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2011.11.001
a method of identifying power supply pulsing parameters for GMAW-P based on statistical experimental design. A linear wire feed rate model using data on the values of wire feed rate obtained from the experiments conducted using a two level factorial experimental design was developed. However, in the GMAW-P process of aluminum, even if the welding process is stable, the weld bead geometry may be poor. The weld pool width often become wider and wider, even collapse at constant welding parameters in GMAW-P process due to the strong heat accumulation and small surface tension. Joseph et al. (2002) have characterized the differences in welding heat input and weld bead shape that could be produced by the pulsed current waveforms. Hirai et al. (2001) proposed a penetration depth model based on vision sensing. During the welding, a fuzzy controller adjusted the welding current waveform so as to get the desired penetration depth. Therefore, adjusting welding current waveform suitably may be an alternative to obtain suitable heat input and improve the weld bead shape. How to sense the weld pool is a problem. Previous studies show that the use of vision sensing to achieve real-time control for weld pool is an effective means. Chen et al. (2002) designed a weld pool width control test through vision sensing and the results showed that the realtime and precision requirements for detecting and control the weld pool changes of GMAW-P could be met. In GMAW-P, droplets are regularly detached at a fixed frequency and directionally transferred to the workpiece under the influence of the current waveform. The influence of the metal transfer mode on the weld quality and stability is well known for the GMAWP process. Changing of the current waveform obviously increases the difficulty of identifying suitable combinations of welding
Ip : Peak current Ib : Background current Tp : Peak time Tb : Background time Frequency = 1 / (Tp + Tb) Dutycycle = Tp / (Tp + Tb)
Tp
Current (A)
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Current
802
Ip Tb Ib
Time
High energy pulse Low energy pulse Time
Ta
Tb
Fig. 1. Current waveform of GMAW-P.
T parameters. In GMAW-P process of aluminum, the coupling effect of welding parameters is obvious. Changing one welding parameter may cause variation of other parameters and every parameter is influenced by all other parameters. Ideal control model for aluminum welding should be ensuring the stability of welding process while adjusting the current waveform to improve weld bead shape. Therefore, to obtain a good weld formation and high welding quality on the basis of ensuring the welding process stable, multiinformation sensing and multi-variable decoupling control should be adopted for dynamic process of GMAW-P of aluminum. In this paper, a double-variable decoupling control scheme was proposed. The weld bead shape was improved by changing the current waveforms to adjust the heat input while the arc length was controlled to stabilize the welding process; hence, both the weld bead shape and stability of the welding process were improved. The suitable combinations of welding parameters were identified in the control process and automatically avoiding trial and error.
Fig. 2. Current waveform of double-pulsed GMAW welding.
Niederlinski index (NI) is obtained according to the steady-state gain matrix. NI =
˚
2
=
0.1845 × 0.9594 − 0.96 × (−0.24) = 2.3017 > 0 0.1845 × 0.9594
˚ii
i=1
(3)
The second amplification factor can be calculated by the first amplification factor. There is:
P=
0.4247 1.6975 −0.4244 2.2080
RGA can be obtained as follows:
0.4345 0.5655 0.5655 0.4345
2. Coupling analysis for control model
=
According to the identification test, the transfer function of TITO variable control system was obtained as follows:
The coupling index D is obtained after calculation:
⎡
yw yL
=
G11
G12
G21
G22
0.1845
⎢ 5.0459s + 1 ⎣ −0.24 0.0242s + 1
ı Vwire
e−1.1364s e−0.0253s
D=
=
⎤
0.96 e−0.4174s ı ⎥ 1.1691s + 1 ⎦ 0.9594 Vwire −0.4928s e 0.2663s + 1
(1)
where s is differential operator; G11 is the transfer function of duty cycle ı and weld width yw ; G12 is the transfer function of wire feed speed Vwire and weld pool width yw ; G21 is the transfer function of duty cycle ı and wire extension yL ; G22 is the transfer function of wire feed speed Vwire and wire extension yL . In the welding process, keep contact-tube-to-work distance (CTWD) constant, namely the arc length can be expressed by wire extension and a constant. Therefore, coupling degree of the control model can be analyzed through Eq. (1). For MIMO control system, relative gain array (RGA) theory can be used to determine the degree of coupling. First, the first amplification factor ˚ and the second amplification factor P must be determined. The first amplification factor is the steady-state gain matrix:
˚=
0.1845 −0.24
0.96 0.9594
(2)
(4)
0.5655 × 0.5655 = 1.6939 > 1 0.4345 × 0.4345
(5)
(6)
As for the above control system, it can be judged that system stability can be achieved using a controller with integrator, but the coupling index does not meet the requirements. Change of single input signal may cause changes of multiple outputs, and each output is affected by more than one input. Therefore, decoupling design can only be taken to eliminate or reduce the coupling effect among the parameters. 3. Control scheme Double-pulsed GMAW is that the welding current waveform changes from a group of high-energy pulses (large average current) to a group of low-energy pulses (small average current) alternately, as shown in Fig. 2. T is a double-pulse period; Ta is high-energy pulse time in a double-pulse period; Tb is low-energy pulse time in a double-pulse period. Ta /T (the ratio of high-energy pulse time in a period) is defined as double-pulse duty cycle. The double-pulsed GMAW technique is a variation of the PGMAW. Tong and Tomoyuki (2001) discussed the double-pulsed GMAW technique and claimed that this welding process provided beautiful scaly bead, improved gap-bridging ability for lap joint, restrained blowholes for formation, refined grain size, and decreased crack sensibility. Further, Silva and Scotti (2006) proved
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 3. Route of control scheme.
803
Fig. 5. Typical image captured by vision sensing.
that the double-pulsed GMAW, in spite of having theoretically higher potential for porosity generation, did not increase the porosity susceptibility in aluminum welding when compared with the P-GMAW. Therefore, the control scheme was designed based on double-pulsed GMAW of aluminum. In control process, change of weld pool width can be reflected through vision sensing system, and then the weld pool width can be controlled by changing the double-pulse duty cycle to adjust the heat input. In the control process, the current waveform is constantly changeable. However, as is known from the analysis of control model, GMAW-P process of aluminum is a nonlinear and strong coupling process, by improving one parameter to obtain a satisfactory weld quality is very difficult. Therefore, on the basis of single-variable control of weld pool width, a double-variable decoupling control method is proposed adopting compensation decoupling. In order to ensure the accuracy and avoid the signal interference from droplet transfer, in welding process, weld pool width and arc length were both obtained through vision sensing. On the one hand, control for weld pool width was realized by changing double-pulse duty cycle ı to adjust heat input to improve weld formation. However, change of single parameter can easily lead to instability of the welding process. So on the other hand, arc length was controlled simultaneously by changing wire feed speed Vwire to stabilize the welding process. Therefore, good weld formation can be realized on the basis of the stability of welding process. Route of control scheme is shown in Fig. 3.
4. Experiments 4.1. Experimental setup The experimental setup used in this study is shown in Fig. 4. Through-the-arc sensing of the welding current, arc voltage and CCD camera of welding zone image acquisition are assembled in conjunction with each other to study the control scheme during GMAW-P process of aluminum. The experimental setup consisted of four parts: (1) the welding system including the DALEX VIRIO MIG-400L-digital power source, welding gun, welding table and its moving-control unit, (2) rapid prototyping control system consisting two industrial PC connected through TCP/IP network, one as a central control computer and the other as xPC target machine, (3) the data acquisition and output system including sensors for welding current and arc voltage, PCL-812PG, NI PCI-6221 data acquisition card and PCL-728D/A data output card, (4) vision sensing system including CCD camera and NI PCI-1405 image acquisition card. The arc voltage was measured between the contact tip of the welding gun and the fixture. This allowed voltage measurements as close to the arc as possible to maximize the measurement accuracy of the arc voltage. The welding current was measured with a Hall sensor, which was attached to the ground. A data acquisition card in combination with the xPC target machine was used to acquire
Fig. 4. Schematic diagram of experimental setup.
804
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 6. Image processing for weld pool. (a) Original weld pool image, (b) binaryzation, (c) noise elimination, (d) erosion, (e) dilation and (f) edge detection.
voltage and current signals and display the results in a graphical format. The sampling rate was 10 kHz. Based on the aforementioned hardware, special control software was developed to synchronize the running of the welding system, the vision sensing system and the rapid prototyping control system. Both image-processing and data-analyzing algorithms were designed to display the image and voltage and current waveforms on the computer screen simultaneously.
In control process, welding zone image was acquired through vision sensing system as shown in Fig. 5 and with corresponding image processing algorithm weld pool width and arc length signals were obtained, as illustrated in Figs. 6 and 7. Then the obtained weld pool width and arc length signals were exported to the rapid prototyping control system for control. The flow chart of control program is shown as Fig. 8.
Fig. 7. Image processing for arc length. (a) Gray image of welding zone. (b) Wire location image in x direction. (c) Image cut and enhancement. (d) Pixels in y direction.
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
805
Fig. 8. The flow chart of control program.
Table 1 Welding parameters used in group 1.
4.2. Welding conditions Bead on plate welds were made in the flat position. The filler material used was a 5356 aluminum alloy welding wire with a 1.2 mm diameter. The base material was 6061 aluminum alloy with a thickness of 4 mm. The shielding gas used throughout the experiments was pure argon with a gas flow rate of 25 L/min. CTWD and welding speed was kept constant through out the experiments. The values of CTWD and welding speed used for all the experiments were 20 mm and 15 cm/min. For comparison, three groups of tests were carried out respectively. In the first group, GMAW-P of aluminum was conducted at constant welding parameters as followed in Table 1. Then the
Parameters
Value
Duty cycle Peak current Base current Pulse frequency Average current
50% 200 A 20 A 40 Hz 110 A
single-variable weld width control was done with the method of changing double-pulse duty cycle to adjust the heat input. Table 2 shows the setting conditions of pulsing parameters used for the
Fig. 9. Weld bead at constant welding parameters.
Fig. 10. Weld bead obtained in single-variable weld width control process.
Fig. 11. Weld bead obtained in double-variable decoupling control process.
806
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Table 2 Welding parameters used in group 2 and 3.
Duty cycle Peak current Base current Pulse frequency Average current
Value High-energy pulse
Low-energy pulse
50% 200 A 20 A 40 Hz 110 A
10% 250 A 40 A 40 Hz 61 A
experimentation. Finally, the double-variable decoupling control test was carried out. The double-pulse parameters are also shown in Table 2. In the control process, the change range of double-pulse duty cycle was set to 20–80%. The initial double-pulse duty cycle was set to 80%. Obviously, the mean current changed between 61A and 110A.
Width sizes of weld beads mm
Parameters
20
weld bead in first group weld bead in senond group weld bead in third group
18
16
14
12
10
8
0
20
40
60
80
100
120
140
160
Distance away from starting point of weld bead mm Fig. 12. Weld width sizes of the obtained three weld beads.
Fig. 13. Change of related signals in decoupling control process.
180
L. Lu et al. / Journal of Materials Processing Technology 212 (2012) 801–807
Fig. 14. Typical image captured in first 5 s of welding process.
807
twelfth second under the action of fuzzy PID controller in order to offset the heat accumulation of welding with high-energy pulse at the beginning of the welding process. With the change in thermal conditions and decrease of heat input, in order to achieve the set value of weld pool width, double-pulse duty cycle increased. The fluctuation of arc length is generally observed in Fig. 13(c), implying the stability of welding process. The typical welding current and arc voltage in 2 s are shown in Fig. 13(d) and Fig. 13(e). As is illustrated in Fig. 13(d) that current waveform was changing obviously with the change of double-pulse duty cycle. As seen from above tests, the good welding bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. It provides an alternative to obtain proper weld quality for GMAW-P.
4.3. Results and discussion
5. Conclusions
The welds obtained in the three groups of tests are shown in the Figs. 9–11. Fig. 12 shows the weld width sizes of the three welds measured every 0.5 mm. As is illustrated in the Fig. 9 and Fig. 12, the first weld bead became wider and wider obviously. The weld pool width sizes changed from the initial value 9.94 mm to the end value 18.53 mm. This maybe resulted from the strong heat accumulation at constant welding parameters and small surface tension of aluminum. The weld pool width of the second weld bead was improved through the single-variable weld width control; however the weld bead formation was not well and the fish-scale pattern was not obvious. In the control process, though the change of welding current waveform improved the weld bead geometry, it also influenced metal transfer and maybe caused unsuitable combinations of welding parameters, further resulted in the instability of control process. The weld bead obtained through double-variable decoupling control scheme was shown in Fig. 11. The weld bead shape was improved obviously compared with the first and second welds and the spatter was low. Compared with the first weld bead, the heat accumulation effect had been compensated through doublevariable decoupling control scheme. As is illustrated in Fig. 12, the weld width sizes had small fluctuations in welding process. This control scheme also provided beautiful scaly bead. The good weld bead shape and low spatter also imply the stability of welding process adopting double-variable decoupling control scheme. Fig. 13 shows the corresponding signals in double-variable decoupling control process. The change of weld pool width pixels obtained by vision sensing is shown in Fig. 13(a). The pixels of weld pool width kept stable after a stable welding and were consistent with weld width sizes of the third weld bead in Fig. 12. In the first 5 s of welding process, it was difficult to obtain the weld pool width through image processing algorithm due to the strong arc light as shown in Fig. 14. Gradually, wire extension became long and arc light was suppressed, the weld pool width could be sensed successful. In the Fig. 13(b), the double-pulse duty cycle decreased from the
(1) Coupling analysis for TITO control model in GMAW-P of aluminum was done. The strong coupling effect among welding parameters influences the stabilization of control system in welding process. A double-variable decoupling control scheme was proposed for this problem. (2) The developed experimental system makes it possible to sense, observe and control the welding process in real time. (3) Good weld bead shape and stable welding process can be obtained through the double-variable decoupling control scheme without complex metal transfer control and considerable trial and error to identify suitable combinations of welding parameters in GMAW-P. This control scheme provides an alternative to obtain proper weld quality for GMAW-P. Acknowledgment This research work is supported by National Nature Science Foundation of China (50805073). References Chen, S.B., Cao, J.M., Xu, C.M., 2002. Visual sensing and real-time control of weld pool dynamics in pulsed GMAW. Transactions of the China Welding Institution 23, 10–14. Hirai, A., Kaneko, Y., Hosoda, T., 2001. Sensing and control of weld pool by fuzzyneural network in robotic welding system. In: Industrial Electronics Conference, Denver, Colorado, USA, pp. 238–242. Joseph, A., Farson, D., Harwig, D., 2002. Influence of GMAW-P current waveforms on heat input and weld bead shape. Science and Technology of Welding and Joining 10, 311–318. Palani, P.K., Murugan, N., 2006. Selection of parameters of pulsed current gas metal arc welding. Journal of Materials Processing Technology 172, 1–10. Praveen, P., Yarlagadda, P.K.D.V., 2005. Meeting challenges in welding of aluminum alloys through pulse gas metal arc welding. Journal of Materials Processing Technology 164–165, 1106–1112. Silva, C.L.M., Scotti, A., 2006. The influence of double pulse on porosity formation in aluminum GMAW. Journal of Materials Processing Technology 171, 366–372. Subramaniam, S., White, D.R., Jones, J.E., 1999. Experimental Approach to Selection of Pulsing Parameters in Pulsed GMAW. Welding Journal, 166–172. Tong, H.J., Tomoyuki, U., 2001. Features of low frequency modulated type pulsed MIG welding process. Welding & Joining 11, 33–35, 40.