Method for calculating arc voltage estimate

By modeling the welding cable as a higher-order transmission system, the method addresses the inaccuracy of conventional models, enhancing arc voltage estimation and weld quality through improved process control and characteristic calculation.

JP7590570B2Active Publication Date: 2024-11-26FRONIUS INT GMBH
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2023530863
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-11-25
Filing Date
2021-11-23
Publication Date
2024-11-26
Estimated Expiration
2041-11-23

AI Technical Summary

Technical Problem

Conventional R/L cable models fail to accurately estimate arc voltage due to neglecting electromagnetic interactions between welding cables and nearby conductive materials, leading to inaccuracies in welding process control and weld quality.

Method used

Model the welding cable as a transmission system with orders higher than second order to account for magnetic coupling with conductive members, using a welding cable model that includes resistance, inductance, and capacitance to accurately estimate arc voltage.

Benefits of technology

This approach provides a more precise estimation of arc voltage, improving weld seam quality and enabling better control of welding processes by accurately calculating characteristic values such as section energy and heat input.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007590570000019
    Figure 0007590570000019
  • Figure 0007590570000020
    Figure 0007590570000020
  • Figure 0007590570000021
    Figure 0007590570000021
Patent Text Reader

Abstract

To improve the estimation of the arc voltage for use during the welding process, the arc voltage u^ was calculated by the welding cable model 13. LBw To estimate the arc voltage, the welding cable 4 is modeled as a transmission system with a second or higher order. LBw , the measured voltage u for measurement point 19 M and the calculated cable voltage drop u S It is proposed to calculate it as the difference between
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The invention relates to a method for calculating an estimate of an arc voltage of an arc occurring between a welding electrode and a workpiece, in which a cable voltage drop of a welding cable between a measuring point on the welding cable and the arc is calculated by a welding cable model having model parameters, the welding cable model models the welding cable as a transfer system with the welding current passing through the welding cable or the measured voltage applied to the measuring point as input variable and the cable voltage drop as output variable in order to calculate an actual estimate for the cable voltage drop for an actual welding current or an actual measured voltage, and an estimate of the arc voltage is calculated by a known relationship between the cable voltage drop and the arc voltage.Furthermore, the invention relates to a welding power source in which such a welding cable model is implemented. [Background technology]

[0002] The main task of a welding power source for carrying out a welding process is to generate, control and adjust the process variables welding current and / or welding voltage and, if necessary, the welding wire feed depending on the welding objective, so that a good weld result with the desired weld seam quality and a good weld joint are finally obtained. In this case, the control of the welding current plays a special role, since the power for the welding process is provided by this control. The arc voltage is also required for the control of the welding process, since it is an important parameter for the welding process. The arc voltage, which falls at the arc, occurs over the length of the arc and therefore over the distance between the welding electrode and the workpiece (or the weld pool). However, the arc voltage cannot be measured directly in practice. It is therefore reconstructed from indirect measurements.

[0003] A welding cable consisting of a welding current cable (running from the welding power source to the welding torch or welding electrode) and a ground cable (running from the work to the welding power source) connects the welding current source to the welding torch, i.e. the welding electrode and the work. Typically, the welding cable is connected to terminal sockets of the welding power source. The socket voltage between these terminal sockets, i.e. between the welding current cable and the ground cable, and the socket current passing through the welding cable can be easily measured at the welding power source. The socket current can be equal to the welding current, since in many cases, when welding, the current flow parallel to the arc can be neglected. However, due to the voltage drop across the welding cable, the socket voltage does not correspond to the arc voltage. In the known prior art, in order to reconstruct the arc voltage from the socket voltage, a cable model (R / L model) that models the welding cable as a series circuit with a cable resistance and a cable inductance is used to calculate the voltage drop over the welding cable. An estimate for the arc voltage can be calculated by the measured socket voltage, the measured socket current and the cable model. Therefore, to obtain an estimate of the arc voltage, the voltage drop on the welding cable is corrected by means of the cable model, which is known for example from WO 2000 / 74888 or DE 102005005771. This model has a first order, since it consists of a hose packet which only has an energy storage component or a dynamic component (cable inductance).

[0004] The parameters of the R / L cable model, i.e. the cable resistance and the cable inductance, are known or identified. The identification of the parameters can be performed during welding or during the welding phase. For example, the welding electrode (or the contact tube of the welding torch) can be short-circuited to the workpiece. The cable resistance is then calculated at a constant current. Afterwards, a current impulse can be applied and the cable inductance can be calculated from the application. During welding, for example, a current impulse that does not interfere with the welding process can be superimposed and modulated on the welding current and the model parameters can be calculated from the response. The model parameters can be calculated continuously from a relationship consisting of existing measured values, for example as in DE 102005005771.

[0005] Therefore, the arc voltage can be estimated while the welding power source is in operation. However, in practice, it has been found that with the conventional R / L cable model, significant estimation errors can occur when reproducing the arc voltage, which can have a detrimental effect on the welding process.

[0006] The applicant's investigations have confirmed that the laying of the welding cable has a significant influence on the reproduction of the cable voltage. It has been confirmed that in many cases the welding cable is laid close to electrically conductive members / components / structures. These may be the robot arm of a welding robot, or metal members (for example in a ship), or reinforcing steel in reinforced concrete, or metal cable ducts, etc. Due to the highly dynamic welding current (time-varying current with sharp edges and high frequency) passing through the welding cable, electromagnetic interactions with the members may occur. The time-varying welding current causes a time-varying magnetic field around the welding cable. As is known, an electric field intensity is generated in the conductor, which is directly related to the current density in the conductor due to its electrical conductivity, being present in a changing magnetic field. On the other hand, the current density occurs such that the magnetic field of the member opposes the initial magnetic field. This causes further interactions of the current-carrying components. The time-varying welding current in the welding cable causes such phenomena in the adjacent electrically conductive materials. Conventional R / L cable models are unable to represent such coupling and therefore cannot provide a sufficiently accurate reproduction of the arc voltage in such cases.

[0007] This effect occurs very strongly in the case of ferromagnetic materials, such as iron or steel, because the magnetic flux density in such materials can be strong, but in the case of paramagnetic materials, such as aluminum, the effect can also be observed, although it is weaker than in the case of ferromagnetic materials.

[0008] Since the welding cable must always be laid close to such conductive materials during practical use - which must also be welded - this effect always affects the execution of the welding process in practice. This effect cannot be detected by the currently used R / L cable model consisting of a series-connected cable resistance and cable inductance. This can lead to an inaccurate reproduction of the arc voltage. On the one hand, this can have a negative effect on the obtained weld quality.

[0009] The arc voltage is not only an important variable for controlling the welding process, but is also required to calculate important known characteristics of the welding process, such as the heat transferred into the workpiece to be welded or the section energy (also called arc energy), for which an estimate of the arc voltage as accurate as possible is also required. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] International Publication No. 2000 / 74888 [Patent Document 2] German Patent No. 102005005771 Summary of the Invention [Problem to be solved by the invention]

[0011] It is an object of the present invention to provide an improved estimation of arc voltage for use during the welding process. [Means for solving the problem]

[0012] This problem is solved in that the welding cable is modeled as a transmission system with a second or higher order in order to estimate the arc voltage by means of the welding cable model, and the estimated value of the arc voltage is calculated as the difference between the measured voltage for the measuring point and the calculated cable voltage drop. It was found that by using a welding model with a second or higher order when estimating the cable voltage drop, the magnetic coupling of the welding cable with the conductive members around the welding cable can also be taken into account. The estimation of the arc voltage can therefore also take into account the laying of the welding cable, and the estimation of the arc voltage can be performed more accurately than before. The estimated arc voltage can be used to control the welding process, for example to control the welding current or the wire feed speed of the welding wire. Due to a more accurate estimation of the arc voltage, the achievable quality of the weld seam can also be improved, since the control can provide a more accurate estimation of the arc voltage. However, the estimation of the arc voltage can also be used to calculate characteristic values ​​of the welding process, such as section energy, resistance value, power value or heat input. Such characteristic values ​​can also be calculated more accurately by more accurately estimating the arc voltage.

[0013] If the model parameters of the welding current model are not known, they can be identified by parameter estimation methods. Several parameter estimation methods are well known and readily available, so that the model parameters can be easily calculated.

[0014] The identification of the model parameters can also be carried out during the short-circuit phase during welding. This allows a continuous new calculation of the model parameters for a given welding method, thus allowing for changing conditions during the laying of the welding cable to be addressed. However, the model parameters can also be identified for pre-defined time progressions of the input variables and the associated output variables before welding. This allows for the selection of the desired input variable progressions for identifying the parameters. As a result, the identification of the parameters can be improved.

[0015] In this case, it has been found to be very useful if a time course of an input variable with several rising edges is used, where the slope of at least two edges is variable, and where the output variable is measured, so that the transmission system as a welding cable can be well improved in order to obtain measurements and estimates of the output variable for estimating the parameters.

[0016] The present invention will now be described in detail, by way of example, schematic and non-limiting example, with reference to Figures 1 to 5 which show preferred configurations of the present invention. [Brief description of the drawings]

[0017] [Figure 1] 1 shows a welding power source for controlling a welding process. [Diagram 2] 1 shows the installation of a welding cable near a conductive member. [Diagram 3] 1 shows a welding cable model having second or higher order. [Figure 4] 13 shows the time course of the welding current as an impulse for identifying model parameters of a welding cable model. [Diagram 5] 1 shows the time transition of the arc voltage at a predetermined welding current using a cable model having a second or higher order of the present invention, and the time transition of the arc voltage at a predetermined welding current using a conventional R / L cable model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] FIG. 1 shows a welding power source 1 for performing a welding process. The welding power source 1 comprises a power unit 2 for generating a welding current. The power unit 2 provides a socket voltage u to perform the desired welding process. B and socket current i BThe power unit 2 is shown as a current source 3 which supplies a current to the connector sockets A+, A- of the welding power source 1. The power unit 2 is an electric circuit, for example a voltage converter such as an AC / DC converter or a DC / DC converter, which converts an input voltage, for example consisting of a single-phase (or multi-phase) AC mains supply or consisting of a direct current voltage (for example from a battery), into a current-time course (of the welding current). The current course depends on the welding method and is controlled by a welding control device 11. The welding control device 11 can be integrated in the welding power source 1, but can also be separate from the power unit 2 or the welding power source 1. The planned current course is provided to the welding control device 11 as a target variable SG. Naturally, like the current course, the voltage course can also be preset as a target variable SG. A welding cable 4 with a welding current cable 5 and a ground cable 6 is connected to the connector sockets A+, A- of the welding power source 1. Although not necessary, the ground cable 6 is generally laid separately to the workpiece 7 and connected to the workpiece 7 to be welded.

[0019] Typically, a welding current cable 5 is laid in a hose packet up to a welding torch 8 with a (consumable or non-consumable) welding electrode 9. Further cables, such as, for example, a coolant cable, a shielding gas cable, a cable for a consumable welding electrode, a control cable, etc., may be integrated into the hose packet in a known manner. A ground cable 6 may be laid in the hose packet.

[0020] The welding torch 8 can be arranged on a welding robot (not shown) or can be operated manually. For example, as in the case of metal inert gas welding (MIG welding) or metal active gas welding (MAG welding), the welding torch 8 can have a contact tube (shown in FIG. 1 ). This contact tube is connected to the welding current cable 5 and is contacted by a welding wire 18 which at the same time serves as the welding electrode 9. However, a non-consumable welding electrode 9 connected to the welding current cable 5 can also be arranged on the welding torch 8, as for example in the case of tungsten inert gas (WIG) welding. The filler material is supplied by the welding wire 9. The welding wire 18 can be fed to the welding point by a welding wire feeder 17.

[0021] During the welding, an arc 10 is generated between the welding electrode 9 and the workpiece 7. The arc voltage u LB falls in the arc 10. However, the configuration of the welding torch 8 and the welding method are not important to the present invention. In order to control the welding process, the socket current i B (which can be equated with sufficient accuracy to the welding current, i.e., the current passing through the arc 10) and the arc voltage u LB It is important to know the voltage drop across the arc 10. The socket current i B can be easily measured in the welding power source 1 by corresponding measuring techniques, while the arc voltage u LB It is difficult to measure directly.

[0022] In order to generate a desired welding current and / or a desired welding voltage, the power unit 2 of the welding power source 1 is controlled by a welding control device 11, which generates a control variable S for said power unit 2, e.g. a duty ratio for a PWM (Pulse Width Modulation) control of a power converter, in order to control the welding process. To this end, a welding process control 12 may be provided in the welding control device 11, which executes, monitors and controls a given welding process. Furthermore, a target variable SG of said control, e.g. a target time course of the welding current or welding voltage, is pre-set in the welding process control device 12. The control variable SG may for example be pre-set by adjusting the welding process, e.g. by selecting a welding program or by adjusting welding parameters at an I / O interface. To this end, the welding control device 11 and / or the welding process control device 12 may generate a control variable S for said power unit 2, e.g. a duty ratio for a PWM (Pulse Width Modulation) control of a power converter, in order to control the welding process. LB , the actual socket current (as welding current) i B,ist or welding wire feed speed v D may be obtained from the welding process as an actual variable of the control.

[0023] The control variable S is the power supply voltage Vcc for a given socket current i B or socket voltage u B For this reason, the welding control device 11 converts the actual socket current i B,ist and the preset target socket current i B,soll However, the control variable S may have a current regulator (e.g., a PI or PID controller) that calculates the control variable S from the difference between the target socket current i B,soll or the target socket voltage u B,soll In this case, such a current regulator is implemented in the power device 2. Naturally, the type of the control variable S depends on the respective implementation of the power device 2.

[0024] As shown in FIG. 1, to control the welding process, a welding controller 11 controls a desired or required welding wire feed rate v (shown in dashed line in FIG. 1). DAdditional or different components of the welding power source 1 may be controlled and monitored, such as the welding wire feeder 17, for supplying the welding wire 18 and the welding electrode 9 to the weld point (and over time). To this end, the welding controller 11 may calculate additional or different control variables S for controlling the welding process.

[0025] For example, the welding current or socket current i B , socket voltage u B or welding wire feed speed v D The actual variables required for control, such as the arc voltage u, can be measured by suitable measuring sensors or can be preset, for example, based on the welding program to be executed. LB can not be measured directly by the conventional welding power source 1 and the conventional welding torch 8, but is evaluated for reconstruction based on the welding cable model 13 implemented in the estimator 16. LB The estimate of is "^", i.e., u^ LB As shown in the figure.

[0026] All or certain "control units" or "units" may be configured as microprocessor-based hardware. In this case, the functions of these control units or units are implemented as software. In this case, one common microprocessor-based hardware may be used for multiple functions. However, one control unit or unit may also be configured on hardware as a field programmable gate array (FPGA), a programmable logic device (PLD) or an application specific integrated circuit (ASIC) or other integrated circuit. In this case, multiple control units and / or multiple units may be incorporated on such hardware. However, one control unit or unit may also be implemented as an analog circuit or an analog computer. Naturally, any combination of these configurations is also possible.

[0027] The welding cable model 13 mainly comprises a cable voltage drop u across the welding cable 4, e.g. the welding current cable 5 and the ground cable 6. S Calculate the cable voltage drop u S Estimated arc voltage u^ from LB In this case, the cable voltage drop u S is the sum of all voltage drops occurring in the welding cable 4 from the measuring point 19 in the welding cable 4 to the arc 10 and from this arc 10 to this measuring point 19 in the welding cable 4. The corresponding live part of the welding torch 8 up to the arc 10 and the workpiece 7 from the arc 10 to the ground cable 6 are also included in the welding cable 4. The measured voltage u between the welding current cable 5 and the ground cable 6 M is measured at measurement point 19.

[0028] Basically, the measuring point 19 can be provided at any position between the connector sockets A+, A- of the welding power source 1 and the welding torch 8. As in FIG. 1, for example the connector sockets A+, A- can be used as measuring point 19. However, as shown by the dashed line in FIG. 1, the measuring point 19 can also be provided at any position in the welding cable 4 or at the end position of the hose packet. It is also possible to provide several measuring points 19. The measured voltage u between the welding current cable 5 and the ground cable 6 M is measured at measuring point 19. In FIG. B But the measured voltage u M Then, the arc voltage u^ is measured. LB is the measured voltage u at each point in time. M and the cable voltage drop u calculated by the welding cable model 13 S For example, the measured voltage u M and cable voltage drop u S In this case, the known voltage drop across the length of the unfixed wire electrode may be taken into account in a known manner. In this case, the cable voltage drop u S does not need to be calculated as an intermediate step, but the estimated arc voltage u^ LBmay be calculated directly. The cable voltage drop u across the welding cable 4 between the measuring point 19 and the arc 10 S must be reproduced by the welding cable model 13.

[0029] For this purpose, the welding cable model 13 is connected to the actual socket current i B,ist The welding current i S , and / or the measured voltage u M , e.g. the actual socket voltage u B,ist may be obtained as input variables which may be measured by measurement sensors and provided as analog or digital measurements, for example in continuous or discrete time.

[0030] The welding cable model 13 may be implemented as software on microprocessor-based hardware, which may be shared by a single control unit, or all or part of the welding cable model 13 may be implemented as hardware in the form of an FPGA, PLD, ASIC or similar circuitry.

[0031] As shown in Fig. 2, the welding cable 4, the ground cable 6 and / or the welding current cable 5 may be arranged near a paramagnetic or ferromagnetic and conductive member (hereinafter, simply referred to as a conductive member 14 for short). In Fig. 2, the welding current cable 5 is, for example, laid through the conductive member 14a (e.g., a conductive profile or a robot arm), or laid in a loop on the conductive member 14b (which is assumed to be the workpiece 7 to be welded), or laid on the reinforced concrete member 14c.

[0032] In practice, of course, not all conductive members need to be laid as illustrated in Fig. 2, and further laying forms may be implemented within the conductive members 14. A loop 15 may for example be arranged around the conductive members 14, or a coil may be arranged. An electromagnetic interaction may occur between the conductive members 14 and the welding cable 4.

[0033] According to the present invention, the welding cable model 13 does not detect the cable voltage drop u caused by the welding cable 4 itself, as in the conventional R / L cable model. S Only the welding cable 4 should be modeled, but also the influence of the surroundings and the influence of the installation of the welding cable 4. In order to achieve this modeling, it has been found that a welding cable model 13 with second or higher order is required.

[0034] The welding cable model 13 is composed of a number of electrical elements, such as resistors, inductances and capacitances, connected together. In this case, a number of energy storage elements, such as a number of inductances and / or a number of capacitances, are included. A suitable welding cable model 13 is shown in FIG. 3. The cable voltage drop u caused by the welding cable 4 is S is applied to the input of the circuit of the welding cable model 13, and the welding current i S is applied to the welding cable model. Cable voltage drop u S and welding current i S The relationship between the output voltage drop u and the input voltage drop u can be expressed by a transfer function G as a division of the output and input variables. S denotes the output variable y, and the welding current i S denotes the input variable u. However, the measured voltage u M is used as the input voltage, and the cable voltage drop u S may be used as an output variable. The order of the transfer function G, i.e. the time derivative that is the maximum in the underlying differential equation, or equivalently the order of the underlying differential equation, determines the order of the welding cable model 13. According to the invention, the cable voltage drop u S and welding current i S or measurement voltage u MA welding cable model 13 is used, which represents the relationship between the load current I and the load current V by a transmission system with second or higher order, which can be considered equivalent to a transfer function G with second or higher order. The welding cable model 13 is therefore configured as a number of interconnected electric elements that behave exclusively passively. Due to the higher order than is the case with conventional R / L cable models (with first order), the surrounding influences can be modeled sufficiently.

[0035] A physical system such as the welding cable model 13 can be represented by differential equations in the continuous time domain or similarly by differential equations in the discrete time domain, and may also be represented as a system of simultaneous differential or differential equations. It is known from system theory that a system can be transformed from simultaneous first-order differential or differential equations to single-dimensional differential or differential equations of higher order, and from single-dimensional differential or differential equations of higher order to simultaneous first-order differential or differential equations. For example, a system can be transformed from a first-order differential equation with n dimensions to a single-order n-dimensional differential equation, and from a single-order n-dimensional differential equation to a first-order differential equation with n dimensions. The same holds for difference equations. Thus, the welding cable model 13 having an order of second order or higher can be mathematically represented in a similarly diverse manner.

[0036] The welding cable model 13 and the formulas are described with reference to Fig. 3. The welding cable 4 is modeled by an input series circuit consisting of one ohmic resistance R1 and one cable inductance L1 (i.e., by a conventional R / L cable model with a first order). Furthermore, the coupling with the conductive member 14 around the welding cable 4 is modeled by a coupling inductance L m In this model, the resistance R, which represents the iron loss and eddy current loss, F But the coupling inductance L m1. Another branch may be provided with a series circuit of an ohmic resistance R2 and an inductance L2 to represent a possible current flow i2 in the conductive member 14. It can be seen that this welding cable model 13 corresponds to a known equivalent circuit of a transformer with a short-circuited secondary.

[0037] It is also possible to model the conductive member 14, for example the cable inductance L1, nonlinearly to account for possible magnetic saturation in the conductive member 14. Thus, for example, the cable inductance L1 is proportional to the socket current i B function of L1(i B ) In this case, nonlinear system theory can be used to mathematically represent the welding cable model 13. It is pointed out again that, insofar as the welding cable model 13 is represented by a transmission system of order n (n>1), the configuration of the welding cable model 13 is not unique and can therefore differ from the configuration shown in FIG. 3. The welding cable model 13 according to FIG. 3 can be represented by applying Kirchhoff's laws to the transmission system by a system of first-order differential equations as follows:

[0038]

number

[0039]

number

[0040]

number

[0041]

number

[0042]

number

[0043] In order to be able to use the welding cable model 13 in the welding power source 1, the coefficients b0, b1, b2, a0, a1, a2, a3 of the transfer function G(s) and the underlying model parameters p M Here, R1, R2, R F ,L1,L2,L m must be known or calculated. The model parameters p M If these model parameters p M can be calculated by a well-known parameter estimation method. In this case, the coefficients b0, b1, b2, a0, a1, a2, and a3 of the transfer function G can be calculated. That is, the model parameters p M can be calculated indirectly or by using the model parameters p M can be calculated directly.

[0044] For known coefficients b0, b1, b2, a0, a1, a2, a3 due to known variables, e.g. welding current i s or measurement voltage u M ,Detected cable voltage drop u s can be calculated in the continuous-time domain, but this is practically laborious. For this, differential equations need to be solved, which is often not possible analytically and has to be performed numerically. Due to the computational effort involved in the numerical implementation, it is often not possible to perform the calculations online within a sufficiently short period of time. It is therefore advantageous to sample the time-continuous signal with a sampling time T and to install a sampling system, which calculates the actual value of the output variable from the preceding values ​​of the input and output variables by means of difference equations (instead of differential equations). Such a sampling system is much better suited for practical implementation in the welding power source 1.

[0045] According to system theory, a sampling system is a system in which sampling time t=k T a (Here, T ais the sampling time). In this case, k indicates each sampling point. Therefore, the welding current i S (t) and cable voltage drop u S (t) from the discrete time function i S (k) or alternatively written as i S,k and the cable voltage drop u S (k) or alternatively written as u S,k In this case, k is k·T a By using the well-known z-transform, we can obtain the time-discrete transfer function G Z (Z) can be calculated as a division of the output variables and the input variables. Similarly, the time-discrete transfer function G Z The degree n of (Z) is 2 or more, as in the case of a time-continuous transfer function. In this case, the z operator and the coefficient a are also polynomials. i ,b i is the input variable u k , e.g., welding current i S,k or measurement voltage u M,k and the output variable y k , e.g. cable voltage drop u S,k The transfer function G Z In (Z), the order n is determined by the maximum power of the z operator. Therefore, the welding current i for the actual sampling point k S,k (or the measured voltage u M,k ) and the welding current for the preceding time point (kj)·Ta iS,k-j (or the measured voltage u M,k-j ) and the cable voltage drop u from the preceding point (kj)·Ta S,k-j From the value of and the actual time point k (k is k·T a where T a is the sampling time) for the cable voltage drop u S,k A difference equation for calculating the transfer function

[0046]

number

[0047]

number

[0048] In this case, the time-discrete transfer function G(s) has a different value from the time-continuous transfer function G(s). Z The coefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) are known or must be calculated, for example, by a known parameter estimation method. However, these coefficients b0, b1, b2, a0, a1, a2, and a3 are also related to the model parameters p M As a result, the model parameter p M can be calculated from these coefficients.

[0049] The difference equation is given by: S,k Therefore, the output variable y k-j , e.g. cable voltage drop u S,k-j and the input variable u k-j , e.g., welding current i S,k-j (or the measured voltage u M,k ) are stored for the required time point (kj), and the input variables u k , e.g., welding current i S,k (or the measured voltage u M,k ) need only be calculated, for example by measuring and sampling the respective variable.

[0050] As mentioned above, the transfer function G(s) or G ZCoefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) (Then, from these coefficients, the model parameters p M can be calculated) or the model parameters p M can be calculated directly. Therefore, the model parameters p M are, for example, R1, R2, R F ,L1,L2,L m or indirectly, the transfer function G(s) or G Z The coefficients in (Z) are b0, b1, b2, a0, a1, a2, and a3.

[0051] Model parameters p M To identify the transfer function G(s) or G Z (Z) or exactly like the mathematical description of the welding cable model 13 according to FIG. 3, a finite number of model parameters p M In general, the model parameters p are estimated so that the quality criteria of the welding cable model 13 are optimally matched to the section to be identified, here the welding cable 4. M is determined indirectly or directly by a parameter estimation method. For this purpose, there are many known parameter estimation methods. In this case, the least squares method or the optimization of a cost function is frequently used. Both methods are basically known and will be described briefly below.

[0052] These parameter estimation methods estimate the output variable y, here the cable voltage drop u S The N actual measured values ​​of the welding current i S In particular, for the measurement in question, the welding electrode 9 is initially short-circuited, i.e. the arc voltage u LB is 0 V. The cable voltage drop u when the welding electrode 9 is short-circuited S is the measured voltage u at measurement point 19 M , i.e. for example the socket voltage u B Therefore, the welding current i Scan be generated as an input variable u, and the measurement voltage u M , e.g. socket voltage u B can be measured as the output variable y. Thus, any number of measurements i=1,...,N can be made of the input variables u i and the associated output variable y i It can be generated from:

[0053] Due to these measurements, the measured voltage u M The cable voltage drop u between the measuring point 19 and the arc 10, where S It is clear that the welding cable model 13 is estimated to provide the model parameters p M The deviation between the estimate of the output variable ŷ of interest, which depends on ŷ, must be minimized by the parameter estimation method.

[0054] In the case of least squares, the model parameters p M The linear system of equations that is solved to obtain the measurements u i ,y i Therefore, multiple model parameters p M is integrated directly in the parameter vector c or is integrated into the transfer function G(s) or G Z The coefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) are integrated in a parameter vector c.

[0055]

number

[0056]

number

[0057]

number

[0058]

number

[0059]

number

[0060]

number

[0061]

number

[0062]

number

[0063] In the case of optimizing the cost function J, the cost function is minimized by the model parameter p M is installed as a function of (a vector with individual parameters). This means that

[0064]

number

[0065]

number

[0066] For example, the model parameters p of the welding cable model 13 according to FIG. 3 for a given installation of the welding cable 4 M But the cost function

[0067]

number

[0068] However, in addition to the above methods, there are many other methods that can be used to estimate the parameters. There are also methods in which measurement noise can be taken into account. In general, autoregressive methods can be used that are based on known model structures, such as the ARX or ARMAX structures (mainly systems of multiple differential equations of a given structure), and that can also take measurement noise into account. Often, such model structures are solved by maximum likelihood estimation or least squares methods.

[0069] The values ​​of the components in the welding cable model 13 are directly estimated or calculated using a time-continuous transfer function G(s) or a time-discrete transfer function G Z The coefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) are estimated. Z If the coefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) are estimated, the values ​​of the component values ​​in the welding cable model 13 can be calculated directly from the estimates, as described above. Z If the coefficients b0, b1, b2, a0, a1, a2, and a3 of (Z) are estimated, the time-discrete transfer function can be transformed to the time-continuous domain by an inverse z-transform. Similarly, values ​​of the component values ​​in the welding cable model 13 can be calculated from the transformation.

[0070] Input variable u (welding current i S or measurement voltage u M ) and output variable y (cable voltage drop u S ) are required to estimate the parameters. In this case, the input variables u must be selected such that as much information as possible about the system to be identified can be derived from the output variables y. During welding, for example during pulsed arc welding or during short circuit arc welding, the welding current i S The welding current i SThe decay behavior of the current cannot be directly controlled, since the power caused by the current can only be consumed by the resistive power losses caused by the resistive losses of the welding cable model 4 and also by the power losses caused by the arc 10 when the arc is struck. Therefore, in order to estimate the parameters, it is necessary to calculate in particular several rising edges, e.g. the slope |Δi S The input variable u (welding current i S or measurement voltage u M ) is generated. In this case, the slopes of the individual rising edges are different. S Such a time course of an input variable u as the welding current i is illustrated in FIG. 4. It can be seen in FIG. 4 that the rising edge F1 of the first slope is steeper than the rising edge F2 of the second slope. S decays between these temporally consecutive edges F1, F2.

[0071] In order to estimate the model parameters pM of the welding cable model 13, it is particularly advantageous to select different gradients of at least two edges F1, F2 within the range of 100 A / ms to 10,000 A / ms. In particular, the gradient |Δi S / Δt| is 800 A / ms to 1000 A / ms, particularly preferably 900 A / ms, and the gradient |Δi S / Δt| is 200 A / ms to 400 A / ms, particularly preferably 300 A / ms. The gradient is expressed in absolute value. Because the welding current i S can be both positive (DC) and negative (DC) or alternate between positive and negative (AC).

[0072] In this case, the value of the input variable u, e.g. the welding current i S is the minimum input value u min , e.g., minimum welding current i S,min and the maximum input value u max , e.g., maximum welding current i S,maxThe minimum input value and the maximum input value may be set constant or may be different for different gradients. S When is used as the input variable u, the welding current i that can be supplied from welding power source 1 is S Especially 90% to 100% of the maximum welding current i S,max The minimum welding current i S is selected to be within a range of 0% to 40% of the welding current iS that can be supplied from welding power source 1.

[0073] However, the identification of the parameters may be performed in two stages. From the welding cable model 13 of FIG. 3, the input variables u, e.g. the welding current i S When a pulse is input, inductances L1, L2, and L m It can be seen that the voltage or current for L decays. If the pulse lasts long enough, the voltage or current decays completely. m acts like a short circuit, and the ohmic resistance R1 is given by, for example, the equation R1=u M / i S In the above, for example, the input welding current i S and the measured voltage u measured at measuring point 9, for example. M , e.g. socket voltage u B Therefore, since the resistance R1 is already known, the remaining model parameters p M The parameter estimation method for calculating θ can be simplified.

[0074] In this case, the measured voltage u M Until the inductances L1, L2, and L m The pulse is held long enough that the voltage on the measured voltage u decays. M If is constant, the measured voltage u M can be easily measured. The welding current i S In this case, the height of the pulse is the welding current i that can be supplied from the welding power source 1. S It is especially chosen by 40% to 60% of people.

[0075] It should be noted that the order of the edges F1, F2 and possibly the pulse for estimating the resistance R1 is not important for the identification of the parameters and can be chosen arbitrarily.

[0076] The input variables u selected for estimating the parameters, e.g., the welding current i S The time transition (discrete time transition) of is preset as a target variable SG in, for example, the welding process control unit 12. In order to obtain a number i=1,...,N of measured values, the associated output variable y is determined by the input variable u preset in advance, for example, the cable voltage drop u S is preferably measured in a time-discrete manner at predetermined sampling points Ta. The choice of the time course of the input variable u influences the characteristics of the identification of the parameters and is therefore preferably performed in a problem-oriented manner.

[0077] Estimated value u^ for arc voltage LB Let us consider an input variable u, e.g. (the measured socket current i B,ist (corresponding to the actual welding current i S,ist the measured variable or actual measured voltage u for measuring point 19 M,ist (For example, the actual socket voltage u B,ist ), for example, the model parameter p M After identifying the known model parameters p M This welding cable model 13 can be used by LB is the measured voltage u, optionally taking into account the known voltage drop across the non-fixed wire electrode. M,ist and the cable voltage drop u obtained from the welding cable model 13 S This is obtained from the known relationship between

[0078] In a simple case, the relationship is, for example, M,ist and the calculated cable voltage drop u S This is the difference between...

[0079] Model parameters of welding cable model 13 p M The calculation of is performed once before the start of welding. For this reason, the welding electrode 9 is short-circuited through the workpiece 7, and the input variable u (welding current i S ) is a preset or predetermined time progression of the model parameters p M In this case, the cable voltage drop u S The measurement voltage u corresponds to M is measured at measurement point 19. As a result, the model parameter p M can be calculated as above.

[0080] There are welding processes in which short circuit phases are regularly provided, for example so-called CMT (cold metal transfer) welding processes or short circuit arc welding processes. In this case, the welding electrode to be melted is moved towards the workpiece 7 until a short circuit occurs on contact with the weld pool. Afterwards, the welding electrode is moved back in the opposite direction. During the arc phase, a metal droplet is transferred to the welding electrode into the weld pool before the short circuit occurs. Arc phases and short circuit phases are alternated and performed cyclically.

[0081] Model parameters p M Such a short circuit phase may be utilized to continually update the welding cable model 13 during welding by newly determining as described above during the short circuit phase.

[0082] Model parameters p MSaid repeated new determination of can be useful for continuously adapting the welding cable model 13 to the actual situation during welding, for example when the welding torch 8 is guided by a welding robot and at least a part of the welding cable 4, the ground cable 6 and / or the welding current cable 5 is located on said welding robot. Due to the movements of the joints of the welding robot, alternating magnetic interactions occur between the welding cable 4 and the welding robot (conductive member 14), which can be taken into account as described above. However, due to the movements of said welding robot, the position of the welding cable 4 relative to the conductive member 14 can also change. As a result, interactions can occur as well. Naturally, interactions can also occur during manual welding.

[0083] FIG. 5 (upper part of FIG. 5) shows the arc voltage u generated by the conductive element 14 near the welding cable 4 in the case of a current impulse with a steeply rising current edge of the welding current iS, which is measured for comparison. LB The time transition of the arc voltage u estimated by the conventional simple R / L cable model is shown in Fig. LB,RL is shown by a dashed line. The arc voltage u^ calculated by the welding cable model 13 of the present invention having an order n>1 (specifically, third order as described above for FIG. 3). LB is shown by the dashed line. This arc voltage u^ LB is the measured arc voltage u LB It can be seen that the deviation between these two time series is smaller. For clarity, the R / L cable model LB,RL and the arc voltage u^ calculated according to the present invention LB and the measured arc voltage u LB The respective errors e between the arc voltage u^ and the arc voltage u^ calculated according to the present invention are shown in the lower part of FIG. LB The error e^ that occurs in is the arc voltage u calculated by the conventional R / L cable model. LB,RL is smaller than the error e that occurs in

[0084] It is clear from FIG. 5 that the conventional R / L cable model is less suitable than the inventive welding cable model 13 with n>1, especially in the case of a rapidly changing welding process (sharp current edges). LB It is clear that the arc voltage estimate u^ can be approximated more precisely if necessary. LB The error is calculated by dividing the inductance L1 of the welding cable model 13 by, for example, the welding current i S function of L1(i S )

[0085] As mentioned above, the model parameters p M The value of p is estimated by estimating the parameters. The estimation is repeated to estimate at least one model parameter p M When the time course of is recorded, at least one model parameter p M The state of the welding cable can also be estimated from the time progression of the value of the at least one model parameter p M If changes significantly during successive estimations (which may reach a corresponding limit), damage in the welding cable may also be estimated and a corresponding alarm, e.g. a visual indication and / or an acoustic alarm signal, may be issued at the welding power source 1.

[0086] The above model parameters p M or at least one model parameter p M can also be used to estimate the installation state of the welding cable 4. M (Coupling inductance L, such as in the welding cable model 13 according to FIG. 3, which is coupled to the conductive member 14) m ) becomes very large, which means that a corresponding limit may be reached, an alarm may be issued, for example acoustically or visually, that the position of the welding cable 4 must be checked.

[0087] As explained, multiple measurement locations 19 may be provided. This allows multiple model parameters pM are estimated for each of the measured values ​​for the different measurement points 19, respectively, thereby making it possible to calculate one welding cable model 13 for each of the different measurement points 19. Thus, at least two different estimated values ​​u^ of the arc voltage are calculated. LB can be calculated. Thus, for example, one estimate is checked for validity by the other estimate, by considering the two estimates as valid only if they do not differ too significantly (in which case the corresponding limit cannot be reached). However, if there are at least two different estimates of the arc voltage u^, LB is the estimated arc voltage u^ LB However, in order to limit locally the errors in the welding cable 4, the estimated value of the arc voltage u^ may be used to calculate the average value used as the LB A number of different estimates of the arc voltage u^ after one particular measurement point 19 (i.e. in the direction of the welding torch 8) may be used. LB However, the arc voltage estimates u^ are similar but different from the previous estimates for this particular measurement point 19. LB , this may indicate damage in the welding cable 4 before this particular measurement point 19.

[0088] As above for Fig. 1, the estimated arc voltage u^ LB can be used to control the welding process performed by the welding power source 1. However, the estimated arc voltage u^ LB can be used to calculate further characteristic values ​​important for the welding process. For example, the characteristic values ​​of the welding process, such as the heat input into the workpiece 7 to be welded and / or the section energy (also called arc energy), can be calculated using the arc voltage u^ LB and, if necessary, the welding current i S The interval energy may be calculated in terms of the heat input and / or the interval energy, which may be a measure for the energy supplied to the welding process, for example. The heat input and / or the interval energy may be recorded as log values ​​for quality assurance and for later tracking of errors in the weld. The estimated arc voltage u^ LBOther characteristic values ​​of the welding process that can be calculated based on and stored as recorded values ​​are resistance or power values ​​across the weld point. The recorded values ​​can be stored in the welding power source or elsewhere, or can be output (analog or digital) in real time (or periodically synchronized with the welding process) via an interface. The present application relates to the invention described in the claims, but may also include the following configurations as other aspects. 1. The estimated arc voltage (u^ LB ), comprising the steps of: Model parameters (p M ) is used to measure the cable voltage drop (u) of the welding cable (4) between a measuring point (19) on the welding cable (4) and the arc (10). S ) is calculated, Actual welding current (i S ) or the actual measured voltage (u M ), the actual estimate for the cable voltage drop (u^ LB In order to calculate the welding current (i S ) or the measurement voltage (u M ) and the cable voltage drop (u S ) and Cable voltage drop (u S ) and arc voltage (u LB ) the estimated arc voltage (u^ LB ) is calculated, The welding cable (4) is subjected to the arc voltage (u^ LB ) is modeled as a transfer system of second or higher order to estimate The estimated arc voltage (u^ LB ) is the measured voltage (u M ) and the calculated cable voltage drop (u S ) in the method. 2. Transfer function G(s), G Z 2. The method of claim 1, wherein the transfer system having (Z) is modeled as a division of an output variable y over an input variable u. 3. During a short circuit between the welding electrode (9) and the workpiece (7) or a molten pool on the workpiece (7), a number of associated output variables y are measured for N input variables u, and the model parameters (p M 3. The method according to claim 1 or 2, wherein the N input variables u and a plurality of output variables y are identified by a parameter estimation method based on the N input variables u and a plurality of output variables y. 4. The model parameters (p M ) is identified during the short circuit phase during welding. 5. The model parameters (p M 4. The method according to claim 3, wherein the input variable u and the associated output variable y are identified for a preset time course of the input variable u and the associated output variable y before welding. 6. 6. The method according to any one of 3 to 5 above, wherein a time transition of the input variable u having a plurality of rising edges (F1, F2) is generated, and at least two of the edges (F1, F2) have different gradients. 7. The welding current (i S 7. The method according to claim 6, wherein the different gradients of the at least two edges (F1, F2) relative to the input signal (F1, F2) are selected as input variables in the range of 100 A / ms to 10,000 A / ms. 8. 7. A method according to any one of 3 to 6 above, wherein, in order to determine model parameters of the welding cable model, the parameter estimation method minimizes the deviation between the measured values ​​of the output variables and the estimated values ​​of the output variables calculated by the welding cable model for the input variables. 9. 9. A method according to any one of claims 3 to 8, wherein a number of said model parameters are identified a number of times and the time course of the value of at least one of these model parameters is examined in order to estimate a defect in the welding cable or a poor installation of the welding cable. 10. The measured voltage (u M The welding current (i S ) is applied to the welding cable (4), and the ohmic resistance (R 1 ) is the measured voltage (u M ) and the applied welding current (i s 10. The method according to any one of 1 to 9 above, wherein the calculation is performed by dividing the amount of the 11. A welding power source for executing a welding process by a welding control device (11), Model parameters (p M a welding cable model (13) having a welding cable length (T1) and a welding cable length (T2) are implemented in an estimation unit (16) of the welding control device (11), This welding cable model (13) is a model of a welding current (i S ) or the measurement voltage (u M ) and the cable voltage drop (u S ) is expressed as a transmission system, The welding cable model (13) is a real welding current (i S ) or the actual measured voltage (u M ) versus the actual estimate for the cable voltage drop (u^ LB ) is calculated, The welding control device (11) is a cable voltage drop (u S ) and arc voltage (u LB ) from the known relationship between the arc voltage (u^ LB In the welding power source, The welding cable model (13) in the estimation unit (16) is modeled as a transmission system having a second or higher order; and The estimation unit (16) estimates the arc voltage (u^ LBw ) to the measurement point (19) as the measured voltage (u M ) and the calculated cable voltage drop (u S The welding power supply is configured to calculate the difference between the [Explanation of symbols]

[0089] 1 Welding power source 2 Power equipment 3 Current source 4 Welding Cable 5 Welding current cable 6 Ground Cable 7 Work 8. Welding Torch 9 Welding electrodes 10. Arc 11 Welding control device 12 Welding process control section 13 Welding Cable Model 14 Conductive materials 14a Conductive materials, shapes, robot arms 14b Conductive material, workpiece 14c Conductive materials, reinforced concrete materials 15 Loop 16 Estimation part 17 Welding wire feeder 18 Welding Wire 19 Measurement points A+ Connector Socket A- Connector socket u B Socket Voltage u S Cable Voltage Drop u M Measurement voltage i B Socket Current i S Welding current, current flow S Control Variables SG goal variable u LB Arc Voltage U^ LB Arc voltage (estimated value) y output variable Y^ Estimates of output variables u Input variable e error G transfer function R1 Ohmic resistor L1 Cable Inductance R2 ohmic resistor L2 inductance L m Coupling Inductance F1 1st edge, rising edge F2 2nd edge, rising edge

Claims

1. The estimated arc voltage (u^) of the arc (10) generated between the welding electrode (9) and the workpiece (7) LB ) comprising the steps of: Model parameters (p M ) is used to calculate the cable voltage drop (u) of the welding cable (4) between a measuring point (19) on the welding cable (4) and the arc (10). S ) is calculated, Actual welding current (i S ) or the actual measured voltage (u M ), the actual estimate for the cable voltage drop (u LB In order to calculate the welding current (i S ) or the measurement voltage (u M ) and the cable voltage drop (u S ) and Cable voltage drop (u S ) and arc voltage (u LB ) the estimated arc voltage (u^ LB ) is calculated, The welding cable (4) is modeled by the welding cable model (13) to calculate the arc voltage (u LB ) is modeled as a transfer system with second or higher order to estimate The estimated value of the arc voltage (u LB ) is the measured voltage (u M ) and the calculated cable voltage drop (u S ) is calculated as the difference between

2. Transfer functions G(s), G Z 2. The method of claim 1, wherein the transfer system having (Z) is modeled as a division of an output variable y over an input variable u.

3. During a short circuit between the welding electrode (9) and the workpiece (7) or a molten pool on the workpiece (7), a number of associated output variables y are measured for N input variables u, and the model parameters (p M 3. The method of claim 1, wherein the N input variables u and a number of output variables y are identified by a parameter estimation method based on the N input variables u and a number of output variables y.

4. The model parameters (p M 4. The method according to claim 3, characterized in that the short circuit is identified during the short circuit phase during welding.

5. The model parameters (p M 4. The method of claim 3, wherein the input variable u and the associated output variable y are identified for a preset time course of the input variable u and the associated output variable y before welding.

6. 6. The method according to claim 3, wherein a time course of the input variable u is generated having a number of rising edges (F1, F2), at least two edges (F1, F2) having different slopes.

7. The welding current (i S 7. The method according to claim 6, characterized in that the different slopes of the at least two edges (F1, F2) relative to the input voltage Vcc are selected as input variables in the range of 100 A / ms to 10000 A / ms.

8. 8. The method according to claim 3, characterized in that a plurality of the model parameters are identified a plurality of times and the time course of the value of at least one of these model parameters is examined in order to deduce defects in the welding cable or a poor installation of the welding cable.

9. The measured voltage (u M The welding current (i S ) is applied to the welding cable (4), and the ohmic resistance (R 1 ) is the measured voltage (u M ) and the applied welding current (i s 9. The method according to claim 1, wherein the calculation is performed by dividing the value of the positive integer multiple of the positive integer multiple by the positive integer multiple.

10. A welding power source for performing a welding process by a welding control device (11), comprising: A welding cable model (13) having model parameters (p M ) is implemented in an estimation unit (16) of the welding control device (11), The welding cable model (13) expresses the relationship between the welding current (i S ) flowing through the welding cable (4) of the welding power source (1) or the measured voltage (u M ) applied to the measurement point (19) of the welding cable (4) as an input variable u, and the cable voltage drop (u S ) across the welding cable (4) as an output variable y, as a transfer system; The welding cable model (13) calculates an actual estimate for the cable voltage drop (u^LB) for an actual welding current (iS) or an actual measured voltage (uM); The welding control device (11) calculates an estimated value of the arc voltage (u^LB) from a known relationship between a cable voltage drop (uS) and an arc voltage (uLB), The welding cable model (13) in the estimation unit (16) is modeled as a transmission system having a second or higher order; and The welding power source according to claim 1, wherein the estimation unit (16) is configured to calculate the estimated value of the arc voltage (u^LBw) as the difference between the measured voltage (uM) for the measurement point (19) and the calculated cable voltage drop (uS).

Citation Information

Patent Citations

  • Determination of the arc voltage in welding between a welding electrode and workpiece useful for direct measuement of the arc voltage with direct measurement of the arc voltage from the rectifer output side using an equation

    DE102005005771A1

  • JP1975127853A

  • Method of detecting a welding voltage

    US6710297B1

  • Method of detecting a welding voltage

    WO2000074888A1