JPH0259180A - Arc welding signal processor - Google Patents

Abstract

PURPOSE:To improve the accuracy of detecting the positional deviation of a work having a small shape change by subjecting the AC component of welding current or voltage obtd. by oscillation of a welding torch to Fourier series expansion and determining the ratio of the addition average value of respective groups of the signed numbers determined in such a manner. CONSTITUTION:The positional deviation between the oscillation center of the welding torch and weld line is determined from the change pattern of the welding current or welding voltage obtd. by oscillating the torch in the direction of crossing the weld line on the work. The DC component is removed from the welding current or welding voltage by a means I of this processor. The output signal of the means I is subjected to the Fourier series expansion up to the frequency of frequencies fXN with respect to the integer N larger than 2 and the sign and size of the coefft. of the terms of a sign wave and cosine wave at respective frequencies fXN are determined in a means II. The signed numbers are divided to integer M groups smaller than the predetermined integer N by a means III. The sets of the divided signed numbers are subjected to addition averaging by the multiplication rate predetermined for each of the groups by a means IV. Further, the ratio of the addition averaging results of the specific group predetermined among M pieces of the addition averaging results is determined in a means V.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an arc welding signal processing device for automatically tracking and correcting positional deviation of a workpiece in arc welding using an arc welding robot or the like.

[Conventional technology]

) It is a well-known fact that there is a certain relationship between arc length, welding current, and welding voltage from the capacitive arc phenomenon. The arc sensor is an application of this physical law to workpiece tracking during arc welding. In other words, when the arc welding torch is oscillated in a direction across the weld line on the workpiece, the arc length changes due to the shape of the workpiece, and the welding current and welding voltage near both ends of the oscillation change accordingly. By focusing on the fact that
This is a method for estimating changes in workpiece shape.

[Problem to be solved by the invention]

The above method of estimating changes in workpiece shape by comparing the welding current or welding voltage at both ends of the swing cannot sufficiently detect changes in workpiece shape in typical fillet welding, groove welding, etc. , based on this, the positional shift of the workpiece can be detected. However, with this method, for a workpiece with a small change in shape, it is not possible to detect a sufficient change by simply comparing the welding current or welding voltage at both ends of the swing, and the accuracy of detecting positional deviation of the workpiece is low.

Note that Japanese Patent Laid-Open No. 60-56483 discloses a method that focuses on the frequency and phase of a signal waveform.

However, the method disclosed in this publication uses frequency to integrate rather than sampling both ends, and uses phase information to synchronize the phase with the integrating means, so basically the torch This compares the currents on opposite sides of the swing.

Therefore, an object of the present invention is to provide an arc welding signal processing device that can accurately detect positional deviation of a workpiece whose shape changes are small.

[Means to solve the problem]

The first arc welding signal processing device of the present invention determines the center of swing of the arc welding torch from the change pattern of the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line on the workpiece. An arc welding signal processing device for determining a positional deviation from a welding line, the device comprising a DC component removal means for removing a DC component from a welding current or a welding voltage, and a DC component removal means when the oscillation frequency of an arc welding torch is f. Fourier series expansion is performed on the signal output from 2 to the frequency f x N for an integer N greater than 2, and the sign and magnitude of the coefficients of the sine wave and cosine wave terms are obtained at each frequency f x N. a dividing means for dividing each signed numerical value obtained by the Fourier series expansion means into groups of M integers smaller than a predetermined integer N; and a signed numerical value divided into each group by the dividing means. an averaging means for adding and averaging numerical values with a predetermined multiplication factor for each group; and ratio calculating means for calculating the ratio of the addition and average results of the remaining M-1 groups based on the addition and average results.

A second arc welding signal processing device of the present invention is an arc welding signal processing device for determining the positional deviation between the swing center of the arc welding torch and the welding line in the same manner as described above, and which calculates the DC component from the welding current or welding voltage. and a DC component removing means for removing the oscillating frequency of the arc welding torch, and a signal output from the DC component removing means when the oscillating frequency of the arc welding torch is f, and a term of the oscillating frequency f of the arc welding torch and a term twice that frequency. A Fourier series expansion means that expands a Fourier series up to a+nf obtained by this Fourier series expansion means,
anf.

31112 f, the coefficients of the four terms of ars2r are al
nf, cnsf group and s+n2 f, aa2
r groups; a first averaging means that adds and averages the coefficients of the terms aLnf and aaf divided by the dividing means by a predetermined multiplication factor; a second averaging means that adds and averages the coefficients of the terms aln2r and as2f with a predetermined multiplication factor; and a ratio that calculates the ratio of both averaging results obtained by the first and second averaging means. and calculation means.

A third arc welding signal processing device of the present invention is an arc welding signal processing device for determining the positional deviation between the swing center of the arc welding torch and the welding line in the same way as described above, A first bandpass filter has a center frequency to which welding current or welding voltage is applied, and a second bandpass filter has a center frequency twice the oscillation frequency of the arc welding torch and applies welding current or welding voltage.
a band-pass filter, an extreme value detection means for determining the magnitude and position on the time axis of the extreme values of the waveforms that have passed through the first and second band-pass filters, and extreme value detection means for determining the magnitude and position on the time axis of the extreme values of the waveforms that have passed through the first and second band-pass filters; and comparison means for comparing the magnitude and position of the values.

[For production]

In the configuration of the first arc welding signal processing device, the direct current component is removed from the welding current or the welding voltage by the direct current component removing means, and the signal output from the direct current component removing means is converted to an integer N larger than 2 by the Fourier series expansion means. is expanded into a Fourier series up to the frequency f×N, and the sign and magnitude of the coefficients of the sine wave and cosine wave terms are determined at each frequency f×N.

Each signed numerical value obtained by this Fourier series expansion means is divided into groups of M integers smaller than a predetermined integer N by a dividing means. The set of signed numerical values divided into each group by this dividing means is added and averaged by an adding and averaging means with a predetermined multiplication factor for each group, and the M number of sums obtained by this adding and averaging means is The ratio calculation means calculates the ratio of the average results of the remaining M-1 groups based on the average results of a predetermined specific group among the average results.

In this way, this arc welding signal processing device develops into a Fourier series the AC component of the welding current or welding voltage obtained by moving the arc welding torch in the direction across the welding line, and generates the sine wave and cosine wave of each frequency. The magnitude and sign of the coefficient of the wave term are determined, the determined signed numerical values are divided into groups and averaged, and the ratio of the averaged values of each group is determined. Can be detected well.

The reason why the positional shift of the workpiece can be detected using the method described above is that if there is a change in the shape of the workpiece, a waveform disturbance occurs in the welding current or welding voltage due to the change in the shape of the workpiece. This is because the waveform includes information on the positional relationship between the waveform and the arc welding torch, and the disturbance of this waveform can be expressed by the coefficients of each term when the waveform is expanded into a Fourier series.

In the configuration of the second arc welding signal processing device, the Fourier series expansion is limited to the oscillation frequency of the arc welding torch and a frequency twice that frequency, and other effects are handled by the first
This is similar to the arc welding signal processing device.

In the configuration of the third arc welding signal processing device, the welding current or the welding voltage is applied to the first bandpass filter having a center frequency of the t8 dynamic frequency of the arc welding torch, and It is applied to a second bandpass filter whose center frequency is twice the frequency. Then, the magnitude and position on the time axis of the extreme values of the waveforms that have passed through the first and second band-pass filters are respectively determined by the extreme value detection means. Furthermore, the comparison means compares the magnitude and position of the extreme values detected by the extreme value detection means.

In this way, this arc welding signal processing device extracts the swing frequency component of the arc welding torch from the welding current or welding voltage obtained by swinging the arc welding torch in the direction across the weld line, and The first arc welding process extracts components with a frequency twice the oscillation frequency of the torch, determines the extreme values of the waveform of these frequencies and their positions on the time axis, and compares the magnitude and position of the extreme values. Similar to the signal processing device, it is possible to accurately detect the positional deviation of a workpiece that has a small change in shape.

〔Example〕

This invention number! An embodiment will be described based on FIGS. 1 to 15. This arc welding signal processing device determines the position between the center of swing of the arc welding torch and the weld line based on the change pattern of welding current or welding voltage obtained by swinging the arc welding torch in a direction across the weld line on the workpiece. An arc welding signal processing device for determining the deviation, as shown in Fig. 1, includes a DC component removal means (■) for removing a DC component from the welding current or welding voltage, and where r is the oscillation frequency of the arc welding torch. Means for removing DC components! The signal output from the frequency f for an integer N greater than 2
Expand the Fourier series to ×N frequencies, and each frequency f×N
Fourier series expansion means (■) for determining the signs and magnitudes of the coefficients of the sine wave and cosine wave terms in dividing means (■) for dividing into groups; and averaging means (2) for adding and averaging the set of signed numerical values divided into each group by the dividing means (■) using a predetermined multiplication factor for each group.
Then, the ratio of the average results of the remaining M-1 groups is calculated based on the average result of a specific group predetermined among the M average results obtained by this averaging means (■). and a ratio calculating means V for calculating the ratio.

In this arc welding signal processing device, the DC component is removed from the welding current or welding voltage by the DC component removal means I,
The output of the DC component removing means 1 is subjected to Fourier series expansion by the Fourier series expansion means H for an integer N greater than 2 up to a frequency f×N, and the signs of the coefficients of the sine wave and cosine wave terms are determined at each frequency r×N. and the size are required.

Each signed numerical value obtained by the Fourier series expansion means (2) is divided into groups of M integers smaller than a predetermined integer N by the division means (2).

The set of signed numerical values divided into each group by this dividing means (■) is added and averaged by the adding and averaging means (■) using a predetermined multiplication factor for each group, and the M sets obtained by this adding and averaging means are The ratio calculation means 2 calculates the ratio of the average results of the remaining M-1 groups based on the average result of a predetermined specific group among the average results of .

In this way, this arc welding signal processing device detects the AC component of the welding current or welding voltage obtained by swinging the arc welding torch in the direction across the welding line, which changes the frequency component and phase according to changes in the shape of the workpiece. The AC component of this welding current or welding voltage is expanded into a Fourier series to find the magnitude and sign of the coefficients of the sine wave and cosine wave terms at each frequency, and the signed numerical values obtained are grouped. Since the values are divided and averaged and the ratio of the average values of each group is determined, it is possible to accurately detect the positional deviation of a workpiece with a small change in shape.

Hereinafter, this arc welding signal processing device will be explained in detail with reference to FIGS. 2 to 15.

FIG. 2 shows a perspective view of a stacked fillet work.

This stacked fillet work consists of overlapping the upper plate 2 on the lower plate 1,
The edge of the upper plate 2 is welded to the lower plate 1. In the case of small workpieces, the plate thickness is thin, so there are few steps at the welding point, and conventional arc sensor methods produce large noise components and are difficult to change shape. It is.

In FIG. 2, A indicates a welding line, and A2 indicates a swinging direction of the arc welding torch.

Fig. 3 (al shows a plan view of the overlapped fillet work, Fig. 3 (bl shows a cross-sectional view as well). In Fig. 3,
A shows the locus that the tip of the arc welding torch passes when the arc welding torch does not move in the direction of welding line A1 but swings in the direction of At, and A4 shows the center of swing of the tip of the arc welding torch. It shows.

In addition, this figure 3 shows a state in which the overlap fillet warp is shifted to the right in the figure from the set position in the teaching mode.
That is, the welding line A1 of the overlapped fillet work is deviated to the right from the swing center A4 of the arc welding torch.

Figure 4 shows the welding current in an ideal state with noise removed when the stacked fillet work shifts to the right as shown in Figure 3.The horizontal axis is time and the vertical axis is the magnitude of the current. to show the
Hereinafter, for convenience, the explanation will be based on welding current, but it is a well-known fact that the same tendency can be obtained with welding voltage, so even if welding voltage is used in the method described below, the same effect can be obtained.

Now, when the arc welding torch is to the right of welding line A1 in Fig. 3, the tip of the arc welding torch is on the lower plate 1 side of the overlap fillet work, and the tip of the arc welding torch and the overlap fillet work are on the lower plate 1 side. Since the distance is long, the current will be small. Meanwhile, the arc welding torch is welding &l! When it is to the left of AI, the tip of the arc welding torch is on the upper plate 2 side of the overlap fillet work, and the distance between the tip of the arc welding torch and the overlap fillet work is short, so the current increases.

If the position is shifted as shown in Figure 3, in the section where the arc welding torch is on the right side of welding line A1 and the welding current decreases, welding line A1 is shifted to the right compared to swing center A4. Therefore, the arc welding torch is on the left side of the welding line A1 and the welding current is shorter than the area where the welding current increases. In reality, the welding line AI of the lap fillet work is melted and the corners are rounded, so the welding current changes stepwise when the arc welding torch crosses the welding line A1. Instead, it will be blunted at the point of change. As a result, the welding current has a waveform as shown in FIG.

Fig. 5 (al is a state in which the overlap fillet work is shifted to the left in the figure from the set position in the teaching mode, that is, the welding line A1 of the overlap fillet work is the pivot center A of the arc welding torch. , shows a plan view of the overlapped fillet work in a state where the position is shifted further to the left, and FIG.

The rest is exactly the same as in FIG. 3.

Figure 6 shows the welding current in an ideal state with noise removed when the stacked fillet work shifts to the left as shown in Figure 5.The horizontal axis is time and the vertical axis is the magnitude of the current. Show that.

In the case of misalignment as shown in Figure 5, in the section where the arc welding torch is to the right of welding line A1 and the welding current decreases, as shown in Figure 6, welding line A1 is at the center of oscillation A4. Since the position is shifted to the left in comparison, the arc welding torch is longer than the welding WaA, which is longer than the section where the arc welding torch is further to the left and the welding current increases.

FIG. 7 shows what is called a ``groove work'', in which the abutting portion of two members 3.4, that is, the wide open portion in the shape of a 7, is usually filled with weld metal. Although this type of workpiece has a large change in shape and can be sufficiently detected by conventional methods, the present invention will be shown to be effective assuming a case where the V-groove is small.

FIG. 8 shows the case where ■ the groove work is shifted to the left from the position in the teaching mode on the diagram, that is, the welding line center position B1 of the V groove work is the swing center B of the arc welding torch,
In FIG. 8, which shows a cross-sectional view of the V-grooved workpiece in a state where the position is shifted further to the left, B indicates the left end of the swing;
indicates the right end of the swing, and B indicates the swing direction.

Figure 9 schematically shows the waveform of the welding current in an ideal state, excluding noise components, when welding is performed when the V-groove workpiece is displaced to the left as shown in Figure 8. . The horizontal axis represents time, and the vertical axis represents the magnitude of current.

This 9th waveform is drawn on the assumption that the swing of the arc welding torch in FIG. 8 starts from the right swing end B4. In FIG. 9, to indicates the time when the swing starts, that is, the time when the tip of the arc welding torch is at the right swing end B4, and t indicates the time when the tip of the V-groove workpiece is at the welding line center position B in FIG. tt indicates the time when the arc welding torch is at the left swing end B3. This waveform is periodically repeated with the oscillation period of the arc welding torch.

In Fig. 10, ■When the groove work is shifted to the right from the position in the teaching mode on the diagram, that is, ■The welding line center position B1 of the groove work is the swing center B of the arc welding torch.
, shows a cross-sectional view of the V-groove workpiece in a state where the position is shifted further to the right.

Figure 11 schematically shows the waveform of the welding current in an ideal state, excluding noise components, when welding is performed when the V-groove workpiece is displaced to the right as shown in Figure 10. . The horizontal axis represents time, and the vertical axis represents the magnitude of current.

In such a typical arc welding work, how to detect the deviation between the weld line of the work and the swing center of the arc welding torch will be explained. First, the main points will be explained with diagrams, and then detailed explanations will follow.

First, the case of overlap fillet work will be explained based on FIG. 12 and FIG. The waveforms are already shown in Figures 4 and 6, but they are shown again in Figures 12F and 13F.
It is shown together on the left and right as at.

From the waveforms of Fig. 12+al and Fig. 13 (al),
The oscillating frequency components are extracted, and the waveforms in Fig. 12 (mountain) and Fig. 13 (bl) are obtained. Similarly, the frequency component twice the oscillating frequency is extracted, and the waveforms in Fig. 12 (C1) are obtained.
and FIG. 13 (the waveform of C1 is obtained.

In Tol of FIG. 12, tlI+tt3 is the time at which the maximum value of the fundamental wave is given, and t, t, and C14 are the times at which the minimum value of the fundamental wave is given. Time t11 at which this fundamental wave reaches its maximum value. At tt3 and times t1□ and t14 when the fundamental wave takes a minimum value, the double frequency component of fcl in FIG. 12 takes a minimum value.

On the other hand, in Fig. 13 tb+, tll+ L 13
is the time when the maximum value of the fundamental wave is given as in Figure 12 (bl), and tt8.tt4 is the time when the minimum value of the fundamental wave is given as in Figure 12 (bl).The time when the maximum value is given to this fundamental wave. At tIl+ t+s and at time j+!, t14, when the fundamental wave takes a minimum value, the double frequency component of C1 takes a maximum value as shown in FIG.

Therefore, for example, the time t I that gives the maximum value to the fundamental wave
The direction of the positional shift can be determined based on the polarity of the double frequency component in l, tIff.

In addition, the magnitude of the deviation between the t-capacity tangent line A of the workpiece and the swing center A4 of the 8-point arc torch is shown in Figure 12.
The magnitude of the amplitude of the original tangential current waveform shown in Figure 12 (bl) and Figure 13 (kl) is related, and clearly it is the fundamental wave shown in Figure 12 (bl and Figure 13 (kl)) and Figure 12 [C1 and Figure 13 (kl)]. It is related to the magnitude of the second harmonic wave shown in Figure 13 (C1). In other words, the magnitude of the fundamental wave in Figure 12 (bl) and Figure 13 (b) depends on the welding conditions and the workpiece shape. Also, Figure 12 (C1) and Figure 13 (C1-2)
The maximum value of the amplitude of the harmonic wave is determined by the welding conditions, the workpiece shape, and the maximum deviation amount.

Therefore, based on the amplitude of the fundamental wave shown in FIG. 12(b) and FIG. 13[bl], FIG.
By determining the amplitude ratio of the second harmonic wave shown in Figure 3 (C1), the magnitude of the deviation between the welding line AI of the workpiece and the welding center A4 of the arc welding torch can be determined.

Another typical workpiece, a downward groove workpiece, will be explained with reference to FIGS. 14 and 15. The waveform of the welding current obtained as a result of the welding line center B+ of the V-groove work being displaced from the swing center B2 of the arc welding torch is already shown in Figs. 9 and 11, but it will be shown again. They are shown side by side as FIG. 14(a) and FIG. 15(a).

From the waveforms in FIG. 14 (al) and FIG. 15 (5),
The oscillating frequency components are extracted, respectively, to obtain the waveforms in Fig. 14 (mountain) and Fig. 15 (bl).Similarly, the frequency component twice the oscillating frequency is extracted, and the waveforms in Fig. 14 (cl.
and FIG. 15 (the waveform of C1 is obtained.

In FIGS. 14 and 15, Lo is the start time,
The position is the right swing end B4 as in FIGS. 9 and 11.

tel is the time when the arc welding torch is located at the swing center B2. tzz is the time when the arc welding torch is located at the left swing end B.

Now, assuming that the workpiece is not misaligned, in an idealized state, as can be seen from the explanation regarding FIGS.
) and a wave with twice the frequency of the fundamental wave in FIG. This is the content.

On the other hand, if a positional shift occurs, the fundamental wave component increases in proportion to the amount of shift, contrary to the case of the overlapped fillet work described in FIGS. 12 and 13.

Therefore, Fig. 14 tel and Fig. 15 (C1 2
Figures 14 fb) and 15 based on the magnitude of the harmonics.
By comparing the magnitudes of the fundamental waves in Figure (g)), the amount of deviation can be determined. The direction can be determined by checking the polarity of the fundamental wave component in FIG. 14 (bl) and FIG. 15 (bl) at a specific point of the rocking motion, in this case the right end Ba of rocking, that is, the rocking start point.

From the above explanation, you can understand the basic idea of finding the direction and magnitude of deviation.

Next, a specific method for determining the swing frequency component, the amplitude of the twice the frequency, etc. will be explained. This method uses Fourier series expansion.

If we expand the waves obtained in the lap fillet work shown in Figures 12 and 13 into a Fourier series, we get A, s+n (2tt f t) +A, sin (2・2πrt) +d.

becomes. It is assumed that the DC component has been removed in advance.

d is a triple or higher harmonic component, and is a sufficiently small value in the examples shown in FIGS. 12 and 13. Note that since the fundamental wave component is used as a phase reference, a cosine wave component that is a phase adjustment term does not appear with respect to the sine wave.

In this case, ρI wealth A + /A t is the value determined by the sensor, the sign represents the direction of deviation, and the magnitude represents the degree of deviation.

If we expand the waveform of the groove work in Figures 14 and 15 into a Fourier series, we get A, aln (2πrt+φ.) +A, aln (2・2 π ft +φ0
) ten d.

becomes. φ. is a term representing the phase difference between the swing and the signal, and since the right end of swing B4 is normally used as the reference point, it is either 0 degrees or 180 degrees. d,' are three or more harmonic components similar to the case of the overlapped fillet work described above, and are sufficiently small values. In this example as well, it is assumed that the DC component has been removed in advance, similar to the processing of the overlapped fillet work.

In this case, the sign represents the direction of the deviation, and the magnitude represents the degree of deviation, as in the case of the 0-fold fillet work, where ρ1 = (1φ.)·(AX /AI ) is the required value.

Although the typical case has been explained above, the signal waveform will be more distorted than in the above-mentioned typical example depending on the shape of the workpiece and the swing pattern.

In this case, detection accuracy can be improved by comparing not only the fundamental wave component and its second harmonic component but also higher harmonic components with respect to the oscillation frequency r. The waves are the same, and the number of waves to be compared increases. Regarding overlapped fillet work, the wave expanded into a Fourier series is A+aln(2πft) +A, 廁(2·2πft) +, a,, 5un(3·2πft) +d4. d4 is a harmonic component of 4f or more and has a sufficiently small value.

In this case, β1 = α1 ・ (A! /AI ) + α2 ・ (
A3/AI) is the value required for the sensor.

In the above, the term (Az/A+) is a high-order correction term determined by the workpiece shape and swing pattern.

Therefore, the coefficient α3. β2 depends on the shape of the workpiece and the swing pattern, and is determined experimentally.

In addition, when higher-order waves are included, Again(2yrft) +A, 廁(2・2π r【) ++Bzaa
(22πft) + ・・Conflict・・・＋A) 1 廁 (N・ 2πft) +B, I a+! (N·2πft)+d, and the coefficients of each term obtained by Fourier series expansion are first classified into M characteristic groups.

dH represents a higher-order term of Nth order or higher, and is ignored here.

Then, for the M classified groups, an average value of 1 to 1 is calculated using the following formula. To this average value, ~K1. l represents the feature of each group, and this feature is an extraction of significant changes appearing in the signal.

K, −β,・A +β2・A2+γ2・B.

+ . . . . . +β1 awakening ALI + r tl-Bt K2-βL1. I' At++++γLI41-Bt
+++βL1.8・ALI and 2+γL1.2・BL
I or 2+ ・Conflict − Medium −− +βLm°A14+γL! 'BL! 11=βLfM-11+I ・ALfM-n++γL
fM-11++ ・Bt+s-n++β, (,-mouth.t
' A L IH-11*t+γL ts-1+ *
z ・B L (s-n +10...spear +β8 ・A.

+7. - B8 Then, the relationship between these M additive average values 1 to KH and the positional deviation amount of the workpiece is given by the following equation.

β1=αx(KX/KA) + ・・ ◆・ ・・ +α*−+ CKwl−+ /Kw= )+α++4
+ (Km, + /niya) + αv (KY /Kl
) ＋ ・・・ ・・・ ・・・ . However, R represents a number from 1 to M.

X is 1 or 2 and Y is M or M-1. R
is 1 or 2, 8 does not exist, and R is M or M-
7 does not exist when it is 1. When R is 1, KR, □1・
does not exist, and when R is M, K*41 does not exist.

N1M, LJ, β1. γ4. α is a quantity determined experimentally and depends on the workpiece shape, swing pattern, and welding conditions. However, J represents a number from 1 to (M-1).

If expressed using Fourier series expansion, the relationship between signal waveform characteristics and positional deviation can be easily expressed numerically as described above.

Note that in the above, the phase during Fourier series expansion was set so that the coefficient of the cosine wave term of frequency f was O.
When Fourier series expansion is performed at an arbitrary phase, a cosine wave term of frequency f also appears, and a cosine wave term in the arithmetic expression for the average value Kl is added.

The arc welding signal processing device of the embodiment described above provides the following effects. For workpieces with little change in shape, such as workpieces made of stacked thin plates, the frequency components and their phases are
The signal is more stable than the current value at both ends of the torch swing.

The reason for this is that the extracted frequency 1 phase is originally not an instantaneous value of the signal, but is the most likely waveform estimated value estimated from local information of the shape change component.

When processed in this way, there is an inherent function to remove noise components, and it is possible to accurately detect workpieces with small changes in shape.

In addition, with this configuration, the shape of the waveform can be extracted from the changes that shape changes have on the signal waveform, so changes in the shape of the workpiece can be extracted more precisely than a method that simply compares the magnitude of the current on both sides of the torch oscillation. Can be done.

A second embodiment of this invention will be described based on FIG. 16. This arc welding signal processing device shows an example in which a filter is used to realize a case in which harmonic components of the third order or higher are not considered in stacked workpieces. This arc welding signal processing device is an arc welding signal processing device that calculates the positional deviation between the swing center of the arc welding torch and the weld line in the same way as the device using the Fourier series expansion method described above, and is shown in Fig. 16. As shown, the first bandpass filter BPF is applied with a welding current or welding voltage with the center frequency being the oscillation frequency of the arc welding torch.

and a second welding current or welding voltage is applied with a center frequency that is twice the oscillation frequency of the arc welding torch.
a bandpass filter BPFt, and first and second bandpass filters BPF.

an extreme value detection means KK for determining the magnitude and position on the time axis of the extreme value of the waveform that has passed through the BPF, and a comparison means for comparing the magnitude and position of the extreme value detected by the extreme value detection means KK. It is equipped with HK.

In this arc welding signal processing device, a welding current or a welding voltage is applied to a first bandpass filter BPF whose center frequency is the oscillation frequency r of the arc welding torch, and the welding current or welding voltage is twice the oscillation frequency r of the arc welding torch. A second bandpass filter BPF! whose center frequency is the frequency of BPF! added to. Then, the magnitude and position on the time axis of the extreme value of the waveform that has passed through the first band-pass filter BPF, and the second
The magnitude and position on the time axis of the extreme value of the waveform that has passed through the band pass filter BPFz are determined by the extreme value detection means KK. Furthermore, the comparison means HK compares the extreme value detected by the extreme value detection means KK, and the size and position.

In this way, this arc welding signal processing device extracts the swing frequency component of the arc welding torch from the welding current or welding voltage obtained by swinging the arc welding torch in the direction across the weld line, and The first embodiment extracts components with a frequency twice the oscillation frequency of the torch, determines the extreme values of the waveform of those frequencies and their positions on the time axis, and compares the extreme values, sizes, and positions. Similarly, positional deviations of workpieces with small changes in shape can be detected with high accuracy.

This will be explained in more detail below. IN is an input signal, which corresponds to, for example, a welding current. The same concept can be used when using welding voltage. The input signal IN is input to first and second band pass filters BPF+ and BPFi having different pass bands. The first band pass filter BPF has a center frequency of the oscillation frequency f of the arc welding torch, and a front and rear one
It has a characteristic of sufficient attenuation at /2f. The second band-pass filter BPF is 2 of the oscillation frequency of the arc welding torch.
The center frequency is double the frequency 2f, and the front and rear 1/2r
It has characteristics of sufficient attenuation.

One of the signals that has passed through the first band pass filter BPF is input to the peak detector PDT. This peak detector PDT determines the maximum value of the input signal. The signal that has passed through the peak detector PDT is input to the pulse generator PG. As a result, the pulse generator IPG outputs a voltage of "1" level when the signal input to the peak detector PDT takes a large value, and when this output is in the "1" state, the switch SW is turned on. open.

The signal that has passed through the second band pass filter BPFz is input to the averaging circuit AVR in order to obtain its average value. The signal passed through the averaging circuit AVR and averaged becomes the reference input of the comparator COMP, and one input of the comparator COMP (the output of the second band-pass filter BPFt) is
Value.

That is, the output of the comparator COMP outputs "1" or "0" depending on whether the output of the second band pass filter BPFz is larger or smaller than the average value.

From the above, the switch SW samples the input indicating whether the wave twice the oscillation frequency f is displaced to the positive side or the negative side at the maximum value of the oscillation frequency f. It turns out. Then, the polarity of the direction signal DIR output from the switch SW has information on the direction of deviation.

This is clear from what has been explained with reference to FIGS. 12 and 13.

Now, the first and second bandpass filters BPF, ,B
The output of PF1 is further converted into absolute value circuits ABS, , ABS
! The ratio of the amplitude of the wave having the oscillating frequency f and the wave having the frequency 2f, which is twice the absolute value, is obtained by the divider DIV. The output SL of this divider DIV has information on the magnitude of work shift. This is clear from the explanation regarding FIGS. 12 and 13 as well as above.

Note that the first and second band pass filters BPF+, BPFz, peak detector PDT, pulse generator PC, and averaging circuit AVR used in FIG.

Comparator GOMP, switch SW, absolute value conversion circuit AB
SI, ABS2. Since the divider DIV is well known in circuit technology, we will not discuss its configuration means, but the first
It is clear that Figure 6 is achievable.

Each of the above embodiments has a configuration in which positional deviation is detected by detecting either the welding current or welding voltage, but both the welding current and welding voltage are detected and positional deviation is detected based on each. You can do it like this.

〔Effect of the invention〕

According to the arc welding signal processing device of the first invention, the alternating current component of the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line is expanded into a Fourier series, and a sine wave of each frequency is generated. The magnitude and sign of the coefficient of the cosine wave term are determined, and the signed numerical values obtained are divided into groups and averaged, and the ratio of the averaged values of each group is determined. It can be detected with high accuracy.

Since the arc welding signal processing device of the second invention is a simplified version of the configuration of the first invention, the accuracy is slightly lower than that of the first invention, but the positional deviation is significantly lower than that of the conventional example. High detection accuracy.

The arc welding signal processing device of the third invention extracts the swinging frequency component of the arc welding torch from the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line, and The first arc welding process extracts components with a frequency twice the oscillation frequency of the torch, determines the extreme values of the waveform of these frequencies and their positions on the time axis, and compares the magnitude and position of the extreme values. Similar to the signal processing device, it is possible to accurately detect the positional deviation of a workpiece that has a small change in shape.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing the configuration of an arc welding signal processing device according to an embodiment of the present invention, Fig. 2 is a perspective view of a stacked fillet work, and Fig. 3 is a diagram showing a rightward position shift in the stacked fillet work. An explanatory diagram of the swinging of the arc welding torch, Fig. 4 is a time chart of welding current at 18 welding of the overlapped fillet work shown in Fig. 3, and Fig. 5 is arc welding when the overlap fillet work is shifted in the left direction. An explanatory diagram of the swinging of the torch, Figure 6 is a time chart of the welding current during welding of the overlapped fillet work in Figure 5,
Figure 7 is a perspective view of the V-groove workpiece, Figure 8 is an explanatory diagram of the swinging of the arc welding torch when the V-groove workpiece is deviated in the left direction, and Figure 9 is a perspective view of the V-groove workpiece. A time chart of the welding current during welding, Fig. 10 is an explanatory diagram of the swinging of the arc welding torch when the V-groove work is shifted in the right direction, and Fig. 11 is the welding of the V-groove work in Fig. 1θ. Time charts of welding current, Figures 12 and 13 are time charts for detecting positional deviation of overlapped fillet workpieces, and Figures 14 and 15 are time charts for detecting positional deviation of groove workpieces, respectively. FIG. 16 is a block diagram showing the configuration of a second embodiment of the present invention. ■...DC component removal means, ■...Fourier series expansion means, ■...Dividing means, ■...Additional averaging means, ■...
...Ratio calculation means, BPF,, BPF! ...bandpass filter, KK...extreme value detection means, HK...comparison means Fig. 12 Fig. 13 Fig. 15 figure

Claims (3)

[Claims] (1) An arc that determines the positional deviation between the center of swing of the arc welding torch and the welding line from the change pattern of the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line on the workpiece. A welding signal processing device, comprising: DC component removing means for removing a DC component from the welding current or welding voltage; and a signal output from the DC component removing means when the oscillation frequency of the arc welding torch is f. Fourier series expansion means for calculating the signs and magnitudes of the coefficients of the sine wave and cosine wave terms at each frequency f×N by performing Fourier series expansion on an integer N larger than 2 up to a frequency f×N; dividing means for dividing each signed numerical value obtained by the series expansion means into groups of M integers smaller than the predetermined integer N; and a set of signed numerical values divided into each group by the dividing means. an arithmetic averaging means that adds and averages each group with a predetermined multiplication factor for each group, and an arithmetic average result of a predetermined specific group among the M arithmetic and average results obtained by this arithmetic and averaging means. an arc welding signal processing device, comprising: ratio calculation means for calculating the ratio of the average results of the remaining M-1 groups with reference to . (2) Arc that determines the positional deviation between the center of swing of the arc welding torch and the welding line from the change pattern of the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line on the workpiece. A welding signal processing device, comprising: DC component removing means for removing a DC component from the welding current or welding voltage; and a signal output from the DC component removing means when the oscillation frequency of the arc welding torch is f. A Fourier series expansion means for expanding into a Fourier series a term of the frequency f of the oscillation of the arc welding torch and a term twice that frequency, and sinf, cosf, sin2f, cos2f obtained by this Fourier series expansion means.
The coefficients of the four terms of sinf, cosf groups and s
a dividing means for dividing into groups of in2f and cos2f; and a dividing means for dividing said sinf and cos2f groups by said dividing means.
a first averaging means that adds and averages the coefficients of the sf terms by a predetermined multiplication factor; and the coefficients of the sin2f and cos2f terms divided by the dividing means are added by a predetermined multiplication factor. An arc welding signal processing device comprising: second averaging means for averaging; and ratio calculation means for calculating the ratio of the averaging results obtained by the first and second averaging means. (3) Arc that determines the positional deviation between the center of swing of the arc welding torch and the welding line from the change pattern of the welding current or welding voltage obtained by swinging the arc welding torch in a direction across the welding line on the workpiece. A welding signal processing device, comprising: a first bandpass filter having a center frequency set at the oscillation frequency of the arc welding torch and to which the welding current or welding voltage is applied; and a frequency twice the oscillation frequency of the arc welding torch. a second band-pass filter to which the welding current or welding voltage is applied with a center frequency of An arc welding signal processing device comprising an extreme value detection means and a comparison means for comparing the magnitude and position of each extreme value detected by the extreme value detection means.