JPH0630823B2 - High-frequency electric resistance welding automatic identification and control device - Google Patents

Description

TECHNICAL FIELD The present invention relates to a welding state automatic identification and control device in high frequency electric resistance welding.

As is well known, an electric resistance welded pipe is manufactured by forming a strip into a cylindrical shape by a molding machine, applying a high-frequency current to both V-shaped edges of the strip to heat and melt, pressurizing and welding with a squeeze roll. . Further, the spiral steel pipe is manufactured by spirally winding a strip into a tubular shape and applying a high-frequency current to the overlapping edges to heat and weld the strip.

In such electric resistance welding, a phenomenon in which the welding state has a characteristic is seen depending on the amount of heat input. For more details, see
-33784 and Japanese Patent Publication No. 54-40454,
Briefly explaining in FIG. 9, 10 is a strip which is welded to form an electric resistance welded pipe 11, 12 is a squeeze roll,
Reference numeral 13 is a contact tip for feeding, and 14 is a V convergence point. FIG. 9 (a) shows a state in which the heat input is not so much applied, the welding is performed at the V convergence point 14, and the position of the welding point is almost unchanged.
This is called a type 1 welding phenomenon.

When the heat input is increased, the state shown in Fig. 9 (b) is reached and the molten metal becomes
Eliminated by the electromagnetic force due to the flowing current (the wavy line in the figure shows these), it is not welded at the V convergence point 14, but a parallel slit is made from the point 14 toward the center of the squeeze and welded at the slit end point 15. It The strip is moving, and in the slit part, the phenomenon that the droplets bridge the slits frequently occurs, and when the bridges bridge, the repulsive force disappears because the current does not flow on the squeeze roll side, and the molten metal fills the slits. Therefore, the welding point 15 is not constant and repeats microvibration. This state is called a type 2 welding phenomenon.

When the heat input is further increased, the state becomes as shown in Fig. 9 (c), the slit width is wide and long, but at a certain length, a short circuit due to the droplet occurs at point 14, the slit is filled with molten metal, and the squeeze center End point 15 of the slit extending toward
Is returned near the V convergence point 14 and leaves again at the strip moving speed toward the center of the squeeze. Repeat the above.
This state is called a type 3 welding phenomenon.

Depending on the heat input, an intermediate state between these first, second, and third welding phenomena may occur. In particular, the present invention relates to a means for automatically identifying which of the first type, the second type, ... Welding phenomenon the current welding state corresponds to.

[Conventional technology]

As described above, there are welding phenomena of type 1, type 2, type 3 and intermediates between these in ERW pipe welding. However, which welding phenomenon is optimum cannot be unequivocally determined. , The capacity of the welding machine and the molding machine. However, if these are determined, the optimum welding phenomenon is also determined, and if welding is performed by the determined optimum welding phenomenon (heat input is controlled so that the phenomenon is maintained), the best welding quality can be maintained. Therefore, it is important to know whether the current welding state corresponds to the first type, the second type, ..., Welding phenomenon, and to identify and control automatically in the case of automatic control.

In Japanese Patent Publication No. 54-33784, the welding phenomenon is identified by the changing cycle. That is, the impedance between the contact tips 13, 13 changes due to the movement of the welding point, and accordingly, the welding current, the anode voltage / current of the high-frequency oscillation tube, the grid voltage / current, the phase difference of the high-frequency voltage / current, the squeeze roll pressure, etc. Welding characteristic value) changes. The fluctuating frequency of this welding characteristic value is over 300Hz in the first type (however, this should be regarded as noise), in the second type 40Hz ~ 300Hz,
The third type is 1 Hz to 40 Hz. Therefore, the above Japanese Patent Publication No. 54-33784
Then, extract the frequency fluctuation component of the high frequency voltage from the voltage between the contact tips, calculate the frequency by multiplying the wave number of the component for a certain period of time, and if it is outside the setting range, adjust the welding machine current voltage and set the frequency of the fluctuation component. Be within range.

Although FIG. 11 also shows a circuit of the same type, in this case, fluctuations in the cycle are detected. Ie input signal (eg contact tip voltage, which is frequency modulated by welding point movement)
Is added to the bandpass filter and the Schmitt circuit 21 to form a rectangular wave, which is then added to the counter 22. The counter has a preset switch 23 for setting a numerical value of 0 to 25, for example, 2
5 is preset and subtracted each time a pulse is input from the circuit 21, and when it becomes 0, a stop signal is output. Count 2
Reference numeral 4 receives a clock of 250 MHz, for example, from the clock generator 25, and also receives the clear and start signals from the stop signal and the function (control) circuit 26 to count the clocks between the start and stop signals.
This count value indicates the time during which the counter 22 receives 25 pulses from the circuit 21. Counting is repeated every 25 pulses, and the count value (ti) is set in the register 27. A predetermined value (t _o ) is set in the register 28 by the digital switch 29, and the subtractor 30 sets the difference (t _i −t) between them.
_o ) is output and set in the register 31.

The value of register 31 is the time of 25 waves of the above input signal.
(t _i : sum of each frequency of 25 waves) is a difference from a predetermined value (t _o ), and a change in t _i indicates a welding point fluctuation. The reason for taking the difference from the predetermined value is to take out only the change, excluding the base, and this makes it possible to reduce the absolute value of the difference (takes a positive or negative value). The value of the register 31 is used for control through the band cut filter 32, and the D / A
It is supplied to the display 34 through the converter 33 and displayed. It is also possible to perform manual control by seeing this display.

[Problems to be Solved by the Invention]

When the welding phenomenon is identified by the change in the frequency or the cycle of the voltage between the contact tips associated with such a change in the welding point, there is a risk of making an incorrect judgment. The period (t _i −t
_o ) shows changes with time, (a) is the first kind, (b) is the second kind, (c)
Is the case of the third type, but it varies quite complicatedly.
The period (the reciprocal of which is the frequency) is the amplitude of the waveforms in these figures (a) to (c), but the amplitude changes depending on the gain of the circuit, and unless the gain is adjusted properly, the second type becomes the first type. Or (the amplitude is equivalent to the first type, the same applies hereinafter), or the third type. If the input signal contains noise,
The circuit 21 in the figure is erroneously pulsed, which in turn causes an error in the period measurement. In this way, the mere periodical monitoring may result in erroneous determination of the type 3 welding phenomenon as the type 2 welding phenomenon.

As is clear from FIG. 10, the time-dependent change in cycle is
The species, the second species, and the third species are quite different. That is, the first
In the species, the period width is small and relatively smooth, in the species 2, it becomes large and a sawtooth-like waveform is observed, and in the species 3, the amplitude is considerably large and the sawtooth wave appears clearly. . If the waveform of the change of this cycle with time can be automatically recognized, it is possible to automatically detect the welding phenomenon type that is not confused by gain or noise, and it is possible to perform electric resistance welding with a desired welding phenomenon.

The present invention focuses on these points, and it is an object of the present invention to provide an apparatus capable of automatically identifying a welding phenomenon type from a temporal change pattern of a cycle and maintaining an optimum welding state.

[Means for Solving the Problems]

As shown in FIG. 1, in the present invention, for controlling high-frequency electric resistance welding, a waveform analysis device 50 for recognizing whether the current welding state is a first type, a second type, ... Welding phenomenon is provided. To use.

In FIG. 1, 40 is an AC power supply, 41 is an AC voltage regulator,
42 is a boosting / rectifying circuit, 43 is a high-frequency oscillation circuit, and 10 is a strip to be electro-welded. Further, 44 is a welding characteristic measuring instrument, which in this example takes in the voltage between the contact tips,
A low pass filter, a digital counter, etc. are provided to output the count value t _i . 45 is a subtracted signal memory, 46
Is a subtracting device, 47 is a D / A converter, and 48 is a display. Further, 51 is a voltage command device, and 49 is an analog control unit for heat input control. The AC voltage regulator 41 and the analog control unit 49 configure a heat input control unit, and the welding characteristic measuring instrument 44, the subtracted signal memory 45, and the subtraction device 46 configure a feature amount calculation device.

As shown in FIG. 2, the waveform analysis device 50 takes in the output S ₁ of the subtraction device 46 into a memory, a variation amount Pi, a variation time Qi, a variation frequency Ri, and a disappearance rate Si, and an averaging process. Means, probability of each feature Pr (*, A
It has a means for calculating i), a means for obtaining the total of the probabilities for each welding inoculation, and a means for making the welding type having the maximum total the determination result.

[Action]

The strip 10 is formed into a tubular shape while being conveyed, and a high frequency current is supplied to the V converging portion by the contact tip 13, heated and welded, and formed into an electric resistance welded pipe 11. The high frequency current is generated and supplied by the oscillation circuit 43, and is adjusted by the voltage adjuster 41. That is, the voltage regulator 41 includes a pair of thyristors connected in anti-parallel, and the thyristor is controlled by the output signal S ₄ of the control unit 49 to change the output voltage, so that the power supply voltage of the oscillation circuit 43 changes. Then, the high frequency current changes.

Since the oscillation circuit 43 includes the V converging portion in the circuit, if the welding point changes due to heat input and the impedance changes, the oscillation frequency changes. This is detected by the welding characteristic measuring instrument 44, and the output signal S ₁ (the difference signal (t _i
-T _o) cause.

The change over time of the signal S ₁ is determined by the welding type (first type, second type ...
A characteristic pattern is shown for each welding phenomenon.

The waveform analysis device 50 stores in advance the pattern of the signal S ₁ for each welding type, specifically, the membership function. When the signal S ₁ is input from the subtraction device 46, the waveform analysis device 50 and the stored membership function It recognizes which welding type the current welding state is, and outputs a signal S ₂ representing the recognized welding type.

The signal S ₂ is input to the voltage command device 51 and compared with the desired (set) welding type. If there is a difference, the voltage command device outputs a voltage command signal S ₃ for making it the same. The welding type corresponds to heat input (= voltage x current), and therefore the signal necessary to set the current welding type to the set welding type also indicates the heat input, but the pipe diameter, material, V convergence angle Since the current corresponds to the voltage when the welding conditions such as ... are determined, voltage is used here instead of heat input. The voltage command signal S ₃ enters the control unit 49 to change the firing control signal S ₄ and change the output voltage of the voltage regulator 41. Here, the current also corresponds to the voltage according to the circuit conditions, so the voltage control is heat input control. In this way, the correction is performed so that welding is performed in the desired welding phenomenon. The heat input VI is also returned to the analog control unit 49, which is for monitoring.

Instead of such automatic adjustment, manual adjustment is also possible.
That is, in this case, the waveform analysis device 50 simply outputs the welding type to the display device 52 to display which welding phenomenon is currently being performed by the first type, the second type, ... Upon seeing this, the operator inputs a signal to the control unit 49 (not shown), changes the output signal S ₄ to change the welding phenomenon, and inputs the signal when the desired welding type is displayed on the display device 52. Just stop.

The output signal S ₁ of the subtraction device 46 is D / A converted by the converter 47 and then supplied to the display unit 48, which then supplies the waveform to the display unit 48.
Displays a waveform as shown in FIGS. 10 (a) to 10 (c). This is used for registering the membership function in the waveform analyzer.

The waveform analysis device 50 calculates the characteristic amount, that is, the variation amount Pi, the variation time Qi, ... From the output signal S ₁ , which is performed as follows. That is, as shown in FIG. 10 (d), in a certain time width in this example, the signal S ₁ during 300 mS is forward and backward (that is, the difference signals (t _i −t _o ) and (t _i−1 −t _o). )) Predetermined value In the present example, the one having a difference of 1 V or more (a, b, c, etc.) is obtained, and the one having the largest difference is the predetermined number A from the larger one, and five in this example. Then, the fluctuation amount Pi (i =
₁ , ₂ , ...). The variation amount Pi is averaged to be one of the feature amounts.

Further, the fluctuation time Qi is the time from the value of the trough to the time when the value of the trough recovers 50% of a predetermined% of the fluctuation amount in this example. The variation time Qi is also averaged to be one of the feature quantities.

The number of fluctuations Ri is obtained by obtaining the number of fluctuations Ni in each time zone by dividing the time width (300 mS) into n equal parts in this example,
Let it be the average. Here, the fluctuation means the peak / valley change obtained when the fluctuation amount Qi is calculated, and in this case, all of the changes are counted.

The extinction rate Si (which is also the average extinction rate ▲ ▼) is a small time δ (for example, 5) at which there is a valley-peak-valley change of the signal S _1.
mS), and the height of the mountain is equal to or higher than a predetermined value (1V) within the time width (300mS), and the number M of appearances is calculated.
It is assumed that the ratio with the number of fluctuations is taken.

▲ ▼ = M / ▲ ▼ These ▲ ▼, ▲ ▼, ▲ ▼, and ▲ ▼ are obtained for each of the time widths (300 mS), and these are used as characteristic quantities of electric resistance welding.

This is up to the steps in Fig. 2. That is, in the step, the signal S _{1 is} specifically t _i −t _o (i = ₁ , ₂ , ...
Each of the above) is stored in the data memory and the time width (300 m
For each S _{1 of} S), Pi, Qi, ... Are calculated in steps, and ▲ ▼, ▲ ▼ ,. As is clear from the above description, some steps are mixed. In addition, for processing after the step, for example, 7
Since it is 00 mS, if data is sampled into the memory for 300 mS, data is not sampled for 700 mS, and data is sampled again after 700 mS.

After the feature quantities ▲ ▼, ▲ ▼, ... Are obtained in the step, the process proceeds to the next step, and the welding type probability Pr (*, Ai) estimated from these feature quantities is obtained. To do this, use the membership function.

FIG. 4 shows an example of the membership function of the number of fluctuations and the welding type. Here, the welding types are divided into six types, namely, the first type 1, the second type lower limit 2 ⁻ , the second type 2, the second type upper limit 2 ⁺ , the third type lower limit 3 ⁻ , and the third type 3, a, b, c, d, e
Is a membership function for each type, which indicates the feature quantity and the probability of occurrence of the welding phenomenon. It should be noted that the first three lower limit 3 ^- the membership function of the (not shown). The first type in FIGS. 3 (a) to (f),
... The pattern of the signal S ₁ of the welding phenomenon is shown. This membership function uses a trapezoidal approximation type. The upper side shows a probability of 1, the lower side shows a probability of 0, and the hypotenuse shows an intermediate probability.
Therefore, for example, if the number of fluctuations ▲ ▼ is 6.7, the probability of welding type 3 is 1, the probability of ²⁺ is 1, and the probability of 2 is 0.
7, the same 2 ^- a and 1 probability is 0.

Similar membership functions are prepared for the fluctuation amount ▲ ▼, the fluctuation time ▲ ▼, and the disappearance rate ▲ ▼, and the functions Pr are read by the measured ▲ ▼, ▲ ▼, and ▲ ▼, and the probability Pr for each welding type is read. Find (*, Ai). This is the step of FIG. Where * is welding type 3,3
Any one of ⁻ , 2 ⁺ , 2, 2 ⁻ and 1, Ai is ▲ ▼ if i = 1, ▲ ▼ if i = 2, ▲ ▼ if i = 3.
If i = 4, ▲ ▼ is shown. When the probability Pr (*, Ai) is obtained, the process proceeds to the next step, and the sum of probabilities C and C = Σa _i Pr (*, Ai) is calculated for each welding type. Here, a _i is a weighting coefficient, which is used to distinguish what is important for determining the welding type and what is not so important.

Next, go to step, find the welding type with the highest sum C,
Let it be the recognition output S ₂ .

In other words, the waveform analysis device 50 uses the feature quantities ▲ ▼, ▲
▼, …… The membership function for each is stored in advance,
The input signal S _{1 is used} to calculate ▲ ▼, ▲ ▼, ..., The value of the member shop function is read therefrom to obtain the probability Pr (*, Ai), and the probability Pr (*, Ai) for each welding type is calculated.
The sum C of Ai) is obtained and the maximum welding type is output.

〔Example〕

In the following table, the feature amount A ₁ (variation amount) obtained from the signal S ₁ ,
A ₂ (Variation time), A ₃ (Number of changes), A ₄ (Disappearance rate)
, The probability value calculated from these, and the sum C of the probabilities.

As is clear from the above table, the maximum value of sum C is 3.9,
Therefore, the current welding type is determined to be 2 ⁺ (type 2 upper limit).

The membership function is created with the dose shown in FIG. That is, the feature value Ai is calculated from the signal S ₁ and is calculated as i
= ₁ , ₂ , ... Plotting results in Fig. 5 (a). Obtain the average value and standard deviation σ,
(b) Obtain a normal distribution function and approximate the data Ai with this. Next, as shown in (c), the upper side is 2σ and the lower side is
Approximate with a trapezoid from _max to _min . This trapezoid is the membership function. Instead of a trapezoidal approximation, a triangle may be used as shown by the dotted line, and the result does not change much.

4 × 6 = 24 of the waveform analyzer 50 when the system is started up
It is necessary to memorize the membership function of the seed, but this means that the signal S ₁ is actually obtained by electric resistance welding, the waveform analyzer 50 calculates the characteristic amount Ai, and the operator looks at the display 48. It can be performed by determining the welding type, inputting the type into the waveform analysis device, and repeating such processing while changing the heat input and causing various welding phenomena to appear (this processing program can be executed by the waveform analysis device). Store it).

When the operator inputs the welding type by looking at the characteristic amount displayed on the display, the waveform analyzer (computer) 50 creates (Ai, X, Y) and stores it in a data file. Here, X is the welding type entered by the operator,
Y is a serial number of pipe making conditions (this is entered in advance). Data files are classified by serial number Y of pipe making conditions, and each data is in the form of D (Ai, x, i). Here, i is a data number, for example, 1 to 100.

FIG. 6 shows the procedure for estimating the membership function. Further, FIGS. 7 and 8 show the measurement results of the fluctuation amount and the fluctuation frequency and the membership function obtained from them.

Welding phenomena were classified into six types, but of course the invention is not limited to this. In electric resistance welding, there are many cases where the upper limit of 2 types is appropriate,
It may be a plurality of types with this as the center.

Further, although there are four types of feature amounts, the most effective amount is the amount of variation, and then the number of variations. Although the weighting coefficient a _i corresponds to this, the determination algorithm may be changed to perform the primary determination based on the variation amount and the secondary determination based on the other feature amount.

〔The invention's effect〕

As described above, in the present invention, the characteristic amount representing the molten steel moving state of the welding point is obtained, and the current welding type is determined from the characteristic amount and the membership function created in advance for each welding type. Since a signal indicating that is output, accurate welding state information can be provided, and automatic control of the welding type and automatic holding of the optimum electric resistance welding state are possible. A signal indicating the welding type output from the waveform analyzer can be used for a monitor, and an alarm can be issued when the current welding phenomenon largely deviates from the set welding phenomenon.

[Brief description of drawings]

FIG. 1 is a block diagram showing the basic configuration of the present invention, FIG. 2 is a flow chart showing the processing contents of a waveform analysis device, FIG. 3 is a waveform diagram of various welding phenomena, and FIG. 4 is an example of a membership function. Chart, Fig. 5 is an explanatory diagram showing the procedure for creating a membership function, Fig. 6 is a flow chart showing the procedure for creating a membership function, and Figs. 7 and 8 are examples of correlation between online data and membership function. The table shown in FIG. 9, FIG. 9 is an explanatory diagram of the welding phenomena of the first to third types, FIG. 10 is a table showing an example of the change of the cycle with time, and FIG. In FIG. 1, reference numeral 41 is an AC voltage regulator, 42 is a boosting / rectifying circuit, 43 is a high frequency oscillation circuit, 10 is a strip to be electro-welded, and 13 is a contact tip.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroyoshi Majima 1 Kimitsu, Kimitsu-shi, Chiba Shin-Nippon Steel Co., Ltd. Kimitsu Steel Co., Ltd. (56) References Japanese Patent Publication No. 57-20075 (JP, B2)

Claims (1)

[Claims] 1. A device for automatically identifying and controlling the welding state of high frequency electric resistance welding performed by the output of a high frequency oscillation circuit (43) having a heat input control unit (41, 49), and supplying a signal extracted from the welding unit. A calculation device (44 to 46) that outputs a signal (S ₁ ) indicating the temporal change of the moving state of the welding point, and the relationship between the feature amount created for each welding type and the welding type occurrence probability is shown. The membership function is stored, and the feature quantity (A) is calculated from the signal (S ₁ ) output from the calculation device.
_i )), and a waveform analysis device (50) for outputting a welding type signal (S ₂ ) indicating the current welding state from the obtained feature amount and the membership function.
High-frequency electric resistance welding automatic identification and control device.