Jinhong Zhu1,2, Shuzhong Song1, Hongxin Shi2, Kang Yong Lee3
1 School of Electronic and information engineering, Henan University of Science and Technology, Luoyang, China, 471003
2 Henan Province Key Laboratory of nonferrous metals processing technology, Luoyang, China, 471003
3 School of Mechanical Engineering, Yonsei University, Seoul, South Korea, 120749
Abstract-Pulse gas metal arc welding (Pulse GMAW) plays an important role in advanced manufacturing technology, and it is transforming the conventional welding process into fine controllable one. However, the development of sophisticated Pulse GMAW machines becomes very complex. The synergic parameters preset are needed in order to achieve optimum weld process. In addition, there are many facets of disturbances and adaptive process control is necessary to maintain optimum state. This paper reviews progress in control of Pulse GMAW technology and summarize control strategies for embedded digital system detailedly. Key words- arc welding, Pulse GMAW, control strategy, power supply, digital control I. INTRODUCTION
Pulse gas metal arc welding (Pulse GMAW) is an advanced welding process that plays a significant role in manufacturing industry. However, its quality is highly dependent on the property of power supply. In order to develop the sophisticated and high performance welding machine, understanding the complexity of this process and implementation of proper control strategy are necessary.
The Pulse GMAW system is composed of wire feeder and power supply. The wire feeder usually runs at a feeding rate which can be varied from 2 to 20 meters per minute, but keeps a constant rate during welding. The power supply must provide suitable current or voltage parameter (about 30 to 600 amps and 15 to 45 volts) which matches the feeding rate and leads to optimum welding process. The welding process is characterized by automatic wire feeding, multiple parameters, various and frequently occurred external disturbances and involving complicated physical phenomenon of molten droplet transfer. The final aim of welding is to generate good weld profile, smooth droplet transfer and no spatters. Due to differential wire materials, diameters, shielding gases and welding modes, no available approach can be provided as universal solution. Various control strategies for typical applications have to be adopted, for example, CO2/MAG/MIG welding imply to pure CO2, mixed argon and pure argon gas shielded welding respectively and the corresponding control approaches may differ from each other.
In this paper, Pulse GMAW technologies are summarized on the basis of present researches and advancements. While emphases are paid on synergic parameters integration and adaptive process control, the control strategies of typical welding modes are introduced
detailedly, focusing on fuzzy process control. The new digital control algorithm and system scheme are suggested as well, which may be promising for future development. II. WELDING MODES AND SELECTIONS
According to types of welding current, parameter range or arc length, and forms of droplet
transfer, Pulse GMAW is divided into various categories. There are several situations: In terms of current type, there are DC, pulse and waveform control welding. In terms of parameter range or arc length, there are short-circuit arc (or dip arc) and long arc (or free space arc) welding. In terms of droplet transfer, GMAW arc can be further classified into so typical transfer modes as stated below. These points lay the basis for Pulse GMAW control.
An automatic wire-feeder with even speed and DC power supply are usually adopted for GMAW process. While the wire feed speed and current are getting increased gradually, molten metal at the tip of wire will transfer to welding pool on the work-piece in the form of globular droplet, spray droplet and rotary spray droplet separately. Short circuit mode is suitable for thin sheet welding, but weld profile with excess metal height is undesirable. Globular mode suffers from lack of control over molten droplets and arc instability, not preferable for fine control. Spray mode offers high deposition rate but minimum current for spray mode is necessary and too high. Rotary spray can only occur at very high current and suitable for high efficiency thick plate welding, but seldom adopted for ordinary sheet welding.
Pulse GMAW overcomes the drawbacks of globular mode while achieving the benefits of spray transfer. This mode is characterized by pulsing of current between low-level background current and high-level peak current. The purpose of background current is to maintain arc, and peak currents are long enough to make sure detachment of the molten droplet, and the mean current is always below the threshold level of DC spray transfer. At the same time, although differential pulse parameter may be utilized for welding, the so called one pulse one droplet mode realizes the optimal process controllability. Pulse modulation, such as low and high frequency hybrid double pulse, can add more flexibility to control of welding quality and formation.
In its broadest sense, DC welding does not mean pure conventional DC current. Fine waveform control instead can be used to eliminate spatters and improve weld formation. As shown in Figure 3, it plays an important role to improve dip welding process and quality. For example, cold metal transfer (CMT) and cold arc process (CP) are emerging recently. Additionally, pulse sub-spray welding, combining the characteristics of dip and pulse spray welding, is preferable for medium thickness aluminum welding. These processes can be regarded as special types of Pulse GMAW.
After all, it is welding material and shielding gas that determines welding mode which is suitable for application. In order to develop a Pulse GMAW machine, proper welding modes should be selected and integrated into the final product. This difference of selections determines performance of a particular series of machine. Commonly, the following cases will be included in main design preferences:
For steel welding Pure CO2 gas
DC current: dip to globular transfer
Usually pulse inapplicable, waveform may be desirable For steel and stainless steel welding
MAG gas (82% Ar+18% CO2, 98% Ar+2% CO2) DC current: short-circuit, globular to spray transfer Pulse current: one pulse one drop spray transfer For aluminum alloys welding
MIG gas (100% Ar)
DC current: globular, sub-spray to spray transfer
Pulse current: sub-spray to one pulse one drop transfer
Briefly, the typical welding modes are short-circuit (conventional dip or waveform control), pulse spray and pulse sub-spray welding. The optimal welding conditions should be integrated. Besides, as new materials and control technology emerge, new welding modes will be added as well.
III. SYNERGIC PARAMETERS PRESET
In order to achieve a stable welding process, wire feeding rate and output parameters of power supply must be matched properly. However, it may be rather difficult for an operator to decide what values should be selected, especially the multiple parameters during Pulse GMAW. In fact, it is not a trial and do job on field, but must be left to the control technology and integrated into welding power supply. It is called one-knob or parameters uniform adjustment.
Based on common experiences, welding current is preset in accordance with work-piece thickness. Other parameters will be linked with this value and set accordingly, aiming to obtain the best welding effect. On the other hand, the GMAW system is composed of wire feeder and power supply. Wire feeding rate is equal to melting rate, and both rates increase as welding current rises. It indicates that adjusting welding current is changing wire feeding rate in essential. For the convenience of power supply design, wire feeding rate be changed as reference instead. The relation analysis is given below.
In the simple case of DC free space welding, there is a relation between welding current and wire melting rate as follows:
Here Vm is the melting rate, α and β are constants dependent on wire material and diameter, L is the wire extension length (or stickout) from torch. Generally, the value of later item βLI2 is small and has a weak influence. For aluminum alloys, βLI2 may even be ignored. So, Vm is nearly proportional to current.
As for DC dip welding process, the relation can be modified as follows.
Here i(t) is dynamic welding current, um is voltage drop near electrode, ts is short-circuiting time, tb is arc burning time, α, β and γ are material constant corresponding to short-circuiting and arc burning duration respectively.
The GMAW system of constant voltage (CV) power supply and constant rate wire feeder is most commonly used not only in conventional machines, but also in new inverter based control system. The benefit of CV power supply is easy to maintain arc burning, and obtain a balance between feeding rate and current automatically. However, the drawbacks are droplet repelling effect during arc burning period, and proper dynamic response is required for good welding process.
Although the relation of welding current, wire feeding rate and output voltage has been
investigated, it is found that, due to nonlinear factors which are difficult to be described in mathematic approach, the best matched parameters can be only obtained by experimental approach. Optimal dynamic current slope during short-circuit period is fixed to simplify the adjustment.
In regard of one pulse one drop GMAW welding, there are more parameters and choices. The power supply may be CV or CC characteristics, or changeable during peak and base period, namely the combination of U-U, U-I, I-U, I-I respectively, and each has its merits and shortcomings. In order to achieve one pulse one droplet effect, I-I mode is more preferable.Parameters such as peak current, peak time, base current, base time and wire feeding rate must be properly matched.
Assuming the wire diameter is D, ideal droplet diameter is kD, k is a constant around 1, and pulse frequency is f, wire length melted per pulse is expressed as
So it implies that pulse frequency is proportional to wire feed rate. In other words, once the reference feeding rate is set, pulse period will be fixed.
The relation between wire melting rate and pulse parameters is described as
During the pulse peak stage, peak current must exceed the threshold value for spray transfer generation. If one pulse one drop is desired, the peak time is also decided to generate enough heat for a proper droplet. Due to practical nonlinear factors, the relation is
The constant C is related to wire material and diameter.
As for base current, it is generally not more than 50 ampere, i.e., enough to maintain arc burning but not accumulate heat to affect melting rate. In fact, the low base current is also used to obtain a wide current range and maintain pulse effect.
Then, average current and base time can be determined by
Although the above relations are helpful to understand welding control, but nonlinear characteristics is obvious regarding to P-GMAW welding process. Suitable pulse parameters are very dependent on experimental results.
Waveform control and pulse sub-spray can be regarded as special pulse welding. In waveform control, CC power supply is adopted, current values during various durations are also determined through experiment. In sub-spray mode, CV and even CC power supply can work well
due to its particular characteristics, but arc voltage should be strictly controlled. In pulse sub-spray welding, peak time should be long enough to produce more than one droplet per pulse.
The experimental database is integrated in memory of a digital system. An operator just need to select one reference of thickness, average current, or wire feeding rate, other parameters such as voltage, pulse parameters, or waveform parameters will be recalled conveniently. IV. PULSE-GMAW PROCESS CONTROL
Although synergic parameters preset provides optimum welding outputs in accordance with wire feeding rate, it should be noticed that the above mentioned welding database is obtained under certain conditions. Real conditions such as gas flow and torch height may change more or less and deflect from standard state. In order to maintain optimal welding process and droplet transfer, some remedies or adaptive control methods must be applied. The following situations must be taken into consideration.
A. Open-loop Fine Tuning for CV Power Supply
CV characteristic is commonly adopted in CO2 welding supply. In regard to this mode, its output voltage should be matched with wire feeding rate. Usually, its output voltage is allowed to be adjusted slightly by operator. This is an open-loop fine tuning method.
In addition, the CV characteristic of digital inverter power supply is realized through duty cycle or current control.
B. PID Arc Length Control of Pulse Welding
As for pulse welding of free space transfer, it is ideal to achieve result of one pulse one drop. Nowadays I-I mode P-GMAW is the most common pulse welding process. It doesn’t have the ability of arc length self-regulation. In order to maintain the welding process stable, close-loop arc voltage control should be added. Investigations proved that PI control is feasible for arc length control. Additonaly, digital PI is more flexible and adaptive. Researches also claimed Fuzzy-PID controller with self-adjusting coefficients is possible.
Because average arc voltage control has slow response, peak time voltage detection and arc control can be used for digital control.
Digitalized PI control can be described as follows.
Here, u(k) is the control output after this period, e(k) is the error, Kp is proportional coefficient, Ti is integral coefficient, T is period. It can then be simplified as
In fact, the control output required is base time, and can be written as follows.
This value can be used to readjust the constant of timer/counter in software programming. C. Fuzzy/PID Constant Frequency Waveform Control of Dip Welding
If open-loop tuning is used, even slight change over a range of several volts may deflect from optimum welding state. An adaptive control strategy is needed. Some researchers have suggested a kind of fuzzy self-optimizing algorithm on basis of characteristic value control. With the rapid development of waveform controlled inverter supplies, constant frequency control has been studied in recent years.
A simplified one input one output fuzzy controller was tested as shown in Table I. The strategy is a dual mode Fuzzy/PID algorithm. Its principle is that the period of arc burning pulse will be adjusted according to short-circuit transfer frequency in waveform control. Since fuzzy control could not avoid static error, nonlinear PI control was combined to improve the quality, as shown in Equation (11). Experiments proved that it is a feasible and effective approach.
D. Fuzzy Short-circuit Frequency Control of Pulse Sub-spray Welding
Pulse sub-spray transfer is a choice suitable for sheet aluminum alloy welding, so this mode should also be considered for welding machine design within full range of output parameters. Selection of arc voltage value is very critical to obtain ideal sub-spray droplet transfer, and can only be judged by the frequency of temporary short-circuit transfer.
Therefore, with arc voltage being chosen as control variable, short-circuit number during a certain period as control aim, a fuzzy controller of double inputs and single output can be designed for P-GMAW.
Defining E, EC and ∆U are three fuzzy linguistic sets. E is error set of short circuit number, EC is varing rate set of error, ∆U is output set. According to fuzzy logic principles, we set the universal space for our inputs and output, defining membership functions as Gaussian function. The control rules are summarized in Table II. According to fuzzy inference rules, defuzzification is implemented by using max-product method. Finally, a data table for fuzzy control can be obtained. This data table is calculated offline, stored in system memories and easy for fast accessing during real time control.
V. EMBEDDED SCHEME AND FNC APPROACH
Digitalized welding power supply with intelligent control is the trend of welding equipment development. In regard to Pulse GMAW, either full digital or mixed digital power supply can be adopted. The difference is that the former can generate pulse width modulation (PWM) signals directly, but the later uses analog circuit for PWM generation and thus simplifies basic software design. There are no remarkable differences hereby, but control strategies may lead to essential progress.
Digital control algorithm will meet the following stated needs: in accordance with the given mean current, decide wire feed rate and pulse welding parameters, adjust arc length to make a stable welding process, and realize optimal droplet transfer. It mainly includes synergic (or pre-feedback) adjustment of wire feed rate, arc length close loop control, and short-circuit frequency control as well.
An embedded MCU system mixed with linear PWM IC has been designed in combination to constitute an analog-digital hybrid control system. It is designed according the principles stated before. The function of this system is aimed at CO2/MAG/MIG welding. It contains inverter power supply and wire feeder. The reference current (or wire feeding rate) and voltage can be set independently on feeder panel or in synergic way via front panel encoder. The feeding rate is set through D/A converters, welding current and short-circuit output current are set through D/A outputs as well. Feedback current, feedback voltage, reference current and voltage given in wire feeder are inputted through A/D channels. Software programming implements welding sequence and control algorithm. Communication interfaces can fulfill system upgrading and welding data transmission as well.
In order to overcome the limitation of various algorithms above stated, a fuzzy neural controller (FNC) is proposed to realize arc length and short-circuit frequency control. As shown in Figure 4, neural network controller is trained before practical use, which consists of 2-H-1 (2 inputs, 1 hidden layer, 1 output) BP structure and is used to store rules for fuzzy logic control. The introduction of neural network overcomes the empirical shortage, and the controller may be more adaptive and universal. In addition, since control rules are reflected by weight values of FNC controller, it can realize fine control and improve the precision. So, it overcomes the disadvantage of high static error, and is promising for final quality control and further applications.
VI. CONCLUSION
Advancements in the control of Pulse GMAW have achieved better controlled droplet transfer and welding quality. In this paper, the control requirements of Pulse GMAW were summarized, and typical control strategies were described. Attentions were paid on the approach of digitalized welding power supply. Synergic parameters preset, open-loop fine tuning, arc length control of Pulse GMAW, fuzzy control of droplet transfer frequency for dip welding waveform control and pulse sub-spray welding. The control scheme based on embedded system has been illustrated with discussion on a fuzzy neural controller which has remarkable advantages and is promising for future applications. REFERENCES
[1] Ghosh, PK, et al.Arc characteristics and behaviour of metal transfer in pulsed current GMA welding of aluminium alloy. J MATER PROCESS TECH, 194 (1-3), pp.163-175, Nov 2007. [2] Kim Y S, Eagar T. Metal transfer in pulsed current gas metal arc welding. Welding Journal, 1993 (7), pp.279 s -287 s.
[3] Jiang Weiyan etal. Behaviors of metal transfer in pulse MIG/MAG welding. Transactions
of Chinese Welding Society. 1994 (3) pp.50-56
[4] Modenesi, PJ, et al.A model for melting rate phenomena in GMA welding. J MATER PROCESS TECH 189 (1-3): 199-205 July 2007
[5] Ueyama, T, et al. Occurrence of arc interaction in tandem pulsed gas metal arc welding. SCI TECHNOL WELD. 12 (6): pp.523-529, Nov 2007
[6] Amin M. Synergic pulse MIG welding. Metal Construction, 1981, 13 (6), pp.349 - 353. [7] Amin M. Pulsed current parameters for arc stability and controlled metal transfer in arc welding. Metal Construction,1983 (5), pp.272-278.
[8] Liu Huijie. Pulse MIG welding microcomputer control. Welding Machine. 1993 (4), pp24-27.
[9] Zhao Judong. Twin Fuzzy Controller for Pulse MIG/MAG welding. Electric Automation. 1997 (6), pp.22–25.
[10] Danut Iordachescu and Luisa Quintino. Steps toward a new classification of metal transfer in gas metal arc welding. Journal of Materials Processing Technology, In Press.
[11] P.K. Palani and N. Murugan. Selection of parameters of pulsed current gas metal arc welding. Journal of Materials Processing Technology, Vol 172, Issue 1, February 2006, pp.1-10. [12] P. Praveen, P.K.D.V. Yarlagadda and M.J. Kang. Advancements in pulse gas metal arc welding. Journal of Materials Processing Technology, Vol 164-165, May 2005, pp.1113-1119
因篇幅问题不能全部显示，请点此查看更多更全内容