JPH11104833A - Weld line tracking method in welding - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a weld line tracking method during welding without an inclination against the case that the influence of an arc deflection and a welding bead substrate exists and is capable of welding to uniformly fuse a groove wall. SOLUTION: An image in a molten pool including a welding wire inserted position during welding is fetched by a CCD camera and is image-processed, a welding wire insertion position deviation ΔX for groove left/right ends is calculated from its molten pool shape and further, from the correlation signals such as a difference ΔH in the left/right wet heights of the molten pool, a difference Δρ in the curvatures of the left/right molten pools, a difference ΔA in the areas of clearances in the corner parts of the left/right molten pool ends and a difference ΔG in the amt. of the clearances in the left/right molten pools, the molten state discrimination signals of left/right groove walls are obtained. Based on a welding wire inserted position deviation signal and the molten state discrimination signal of the left/right groove walls, left/right tracking axis center positions are controlled in a real time by using an arithmetic means of fuzzy logic, etc., and thus, weld line tracking is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of automatically following a welding line without welding by monitoring welding conditions immediately below an arc during welding. The present invention relates to a welding line profiling method during welding in which a situation is monitored by a CCD camera, a feature amount such as a molten pool shape being welded is recognized by an image processing device, and welding line profiling is performed in real time.

[0002]

2. Description of the Related Art In conventional automatic welding, a welder performs welding while monitoring the welding condition, and the right and left scanning is adjusted by the welder, so that labor cannot be saved. As the automatic scanning method of the welding line, there are a mechanical scanning method of a contact type and an electric scanning method of a non-contact type. The former is affected by the finishing accuracy of the groove because it is a contact type, especially on surfaces where the groove is mainly gas cut, such as welding, depending on the surface shape of the surface, accuracy and durability There was a problem. In the latter non-contact method, a copying method using an electromagnetic sensor and a non-contact copying method called an arc sensor using a change in current or voltage have been put into practical use. Not only is accuracy and durability inferior, but also the sensor is large, which makes it difficult to install it in a groove.
Further, the arc sensor is most frequently used among these, but it is easily affected by the welding posture and it is difficult to obtain the accuracy.

[0003]

Therefore, in order to carry out a high-precision welding line copying without being affected by the welding posture and noise during welding, the welded portion is directly viewed and imaged by a CCD camera. A method of controlling the image by image processing has been considered. For image processing, the following deviation ΔX of the wire insertion position X _W relative to generally the center position of left and right edge positions X _L, X _R of the molten pool image as shown in FIG. 5, i.e., the wire insertion position deviation ΔX (1) determined by the _{equation, ΔX = X W - (X} L + X R) / 2 ...... (1) welding line is moved right and left scanning axis 17 shown in FIG. 1 as the shift amount [Delta] X is 0 A method of copying is adopted.

However, in such a method, when the deflection of the arc 3 or the influence of the base of the welding bead 4 is affected, the bead shape is slightly biased in that direction, and it is not possible to perform welding without inclination and uniformly melting the groove wall. Was. The present invention has been made in view of the above technical problems, and has a welding line profile during welding capable of performing welding without tilting and evenly melting a groove wall even when there is an influence of arc deflection or a base of a welding bead. The aim is to provide a method.

[0005]

According to the first aspect of the present invention,
In the welding line profiling method, which monitors the welding status immediately below the arc during welding with a CCD camera and recognizes features such as the shape of the molten pool being welded with an image processing device and performs welding line profiling in real time, An image of the weld pool including the welding wire insertion position is captured by a CCD camera, image-processed, and a welding wire insertion position shift ΔX with respect to the left and right ends of the groove is calculated from the shape of the weld pool. The difference between the wetting height ΔH, the curvature difference Δρ between the left and right weld pools, the area difference ΔA between the corners at the edges of the left and right weld pools, the difference ΔG in the amount of gap between the left and right weld pools, etc. A melting state discrimination signal for the front wall is obtained, and the center position of the left and right scanning shafts is calculated using fuzzy inference and other arithmetic means based on the welding wire insertion position shift signal and the melting state discrimination signals for the left and right groove walls. Controls Im, and performs the copying weld line.

According to this invention, the welding line is traced during welding based on the two signals of the melting state determination signals of the left and right groove walls together with the displacement amount ΔX of the welding wire insertion position. Even when the base of the weld bead is affected, it is possible to perform welding line profiling without inclination and capable of performing welding for melting the groove wall evenly.

According to a second aspect of the present invention, a welding condition immediately below an arc during welding with weaving is monitored by a CCD camera, and a feature amount such as a molten pool shape being welded is recognized by an image processing device in real time. In the welding line profiling method of performing welding line profiling, the detection means detects that the weaver has reached the center of the weaving during welding, captures an image of the weld pool at the center of the weaving with a CCD camera, processes the image, From the shape of the molten pool, the welding wire insertion position deviation ΔX with respect to the left and right ends of the groove is calculated, and further, the deviation ΔH of the left and right wet heights of the molten pool, the difference Δρ in the curvature of the left and right molten pools, From the correlation signals such as the area difference ΔA of the gap at the corner portion and the difference ΔG of the amount of the gap between the left and right molten pools, the melting state determination signals of the left and right groove walls are obtained, and the welding wire insertion position displacement signal is obtained. And by a melting state determination signal of the left and right GMA wall to control in real time the center position of left and right scanning axis by using an arithmetic means of fuzzy inference or the like, and performs the copying weld line.

That is, in the case of welding involving weaving,
Since it is necessary to capture the image at the center position of the weaving, the detection signal from the encoder attached to the weaving shaft, or the detection signal from the limit switch attached to the left and right ends in the welding method with a constant weaving width, detects the weaving left and right ends. It is preferable to capture an image by using a CCD camera with a random shutter trigger function that can capture in synchronization with the weaving center position obtained by calculation (randomly synchronized with the center position signal) at the moment while detecting the position. In a normal CCD camera, synchronization is performed at a constant cycle every 1/30 seconds or 1/60 seconds. Until this synchronization is achieved, a maximum of 1/30 or 1/6 is obtained.
After waiting for 0 seconds, a shifted image is taken. According to a CCD camera that synchronizes this randomly, an image can be captured at that moment, and an image without any shift can be obtained.

According to the third aspect of the present invention, the welding situation immediately below the arc during welding is monitored by the CCD camera, and the image processing device recognizes the characteristic amount such as the shape of the molten pool being welded and welds in real time. In the welding line profiling method that performs wire profiling, the short circuit of the welding wire is detected by a detecting means by a detection means, an image of the molten pool after the short circuit is taken into a CCD camera, and the image is processed, and the molten pool shape is processed. The welding wire insertion position deviation ΔX with respect to the left and right ends of the groove is calculated, and the deviation ΔH of the left and right wet heights of the molten pool, the difference Δρ in the curvature of the left and right molten pools, and the gap between the corners of the left and right molten pools are further calculated. The melting state determination signal of the left and right groove walls is obtained from the correlation signal such as the area difference ΔA of the gap between the left and right molten pools and the difference ΔG of the gap amount between the left and right molten pools. Left by traffic light The center position of the scanning axis is controlled in real time by using an arithmetic means of fuzzy inference or the like, and performs the copying weld line.

That is, in order to realize monitoring with high accuracy and high reliability, it is necessary to obtain a clearer image. Therefore, the present invention is configured to take in an image when the welding wire is short-circuited to the molten pool, thereby excluding welding arc light and inputting an image showing only the molten pool. Thus the corner curvature of the molten pool right ρ _L, ρ _R and molten pool left and right clearance area A _L, A _R, and molten pool left and right gap amounts G _L, G _R
Can be measured with high accuracy. The detection of whether or not the welding wire has short-circuited to the molten pool is performed by monitoring the welding voltage V during welding, when the welding voltage is low, or when the short-circuit signal coming out of the welding machine is ON, the welding wire is turned on. Accurate measurement can be achieved by capturing an image using a CCD camera with a random shutter trigger function in which the shutter is randomly released as a short circuit.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will now be described in detail with reference to the drawings. However, unless otherwise specified, the configuration circuits, relative arrangements, types, and the like of the constituent circuits and components described in this embodiment are not intended to limit the scope of the present invention thereto, but are merely illustrative examples. .

FIG. 1 is a control block diagram of a welding copying apparatus according to an embodiment of the present invention. FIGS. 2 and 3 are explanatory views showing processing images of a molten pool accompanied by arc deflection.
Are the wetting heights H _L and H _R of the weld pool 14 at the left and right corners from the weld pool tip, and FIG. 3 is the corner curvature ρ _L of the weld pool right and left.
[rho _R and molten pool left and right clearance area A _L, A _R, and molten pool left and right gap amounts G _L, showing a G _R. In order to explain the outline of the present apparatus based on FIG. 1, a welding situation is imaged through a camera control panel 8 by a CCD camera 6 having a light-shielding filter 7 disposed on the front side, and the captured image is taken into an image processing apparatus 10 and given a predetermined image. The image processing is performed, and the result is transmitted to the overall control device 11 to perform control.

The imaging timing of the CCD camera 6 for taking in the molten pool image is controlled by the overall control device 11 via the image processing device 10 in accordance with a signal from the welding machine 13. Images captured by the CCD camera 6 and image processing results are displayed on the TV monitor 12. When the groove position shift is detected, the overall control device 11 operates the right and left scanning shaft 17 via the servo motor control panel 9 to move the welding torch 1 attached to the tip of the shaft stage such as the vertical shaft 16 and the weaving shaft 15. Perform left-right copying.

That is, the welding monitoring device according to the present embodiment
An image of the welding status is obtained from the image processing device 10 and monitored by an overall control device 11 which performs overall control, and welding line profiling and welding condition adaptive control are automatically performed. 6, a camera control panel 8, an image processing device 10, a control device 11, and a welding machine 13. Here, the TV monitor 12 is a monitor device for checking a welding situation, and is installed as needed. For example, TIG welding or gas shielded arc welding is applied to the welding method to which the present embodiment is applied. In order to perform these weldings with high quality, the overall control device 11 controls the vertical and horizontal positions of the welding torch 1, current control, welding speed control, filler wire supply speed control, weaving speed control, weaving both end stop time control, and the like. . This is traditionally
This operation is performed by an operator, and by automatically controlling the operation, it is possible to reduce man-hours in unmanned welding and provide high-quality welding.

In the present embodiment, in order to perform this automatically, the image of the welding situation is dimmed by a light-shielding filter 7 such as an interference filter, and the image is input by a CCD camera 6 via a camera control panel 8 to perform image processing. The image is analyzed by the device 10 and the position of the welding torch 1 in FIG. 2, that is, the insertion position X _W at the center position of the weaving shaft 15 of the welding torch 1,
The weld pool left end X _L and the weld pool right end X _R , which are the left and right ends of the groove, and the wet heights H _L and H _R of the weld pools 14 at the left and right corners from the weld pool tip are obtained and welded to the control device 11. Output as information. In the case of not performing weaving of the welding torch 1, a welding torch 1 situ is the inserting position X _W.

The reason for obtaining the wet heights H _L and H _R of the molten pool 14 is, for example, as shown in FIG. 2, the lower the wet height H of the molten pool 14, the more the molten pool 14 melts and hangs. If the wet height H of the molten pool 14 is high, it is correlated (indirectly) that the molten pool 14 has receded and has not melted.
It will be represented in. Accordingly, wetting height H _L of the molten pool 14, can determine the molten state of the GMA walls of the left and right from the H _R, which the left and right height H _L, as H _R is equal, if words Kaere, [Delta] H ( ΔH: H _L ~H _R: it is possible to dissolve the lateral evenly groove wall by the difference between the right and left height) is controlled to be zero.

That is, the molten pool portion 14 is imaged by the CCD camera 6, and both ends of the groove 5 and the insertion position of the welding wire 2 are image-processed by the image processing device 10 based on the shape of the molten pool 14. X _L, determine the insertion position X _W of the current of the welding wire 2 against X _R, these from the following (1) obtaining a wire insertion position deviation ΔX by the formula. _{ΔX = X W - (X L} + X R) / 2 ... (1) further obtains a difference [Delta] H of the molten pool wetting height positions of the left and right corner portions in the following formula _{(2), ΔH = H L -H R ...} ( 2) From these deviations ΔX and ΔH, fuzzy inference is performed according to the fuzzy rule shown in FIG. 4, and the right and left scanning shaft 17 is moved based on the fuzzy inference result to perform welding line scanning. The control of each axis including the right and left scanning axes 17 is controlled by a servo motor control panel 9 including a driver and an amplifier for each axis.

At this time, in the case of welding involving weaving, since it is necessary to capture the image at the center position of the weaving, a detection signal from an encoder attached to the weaving shaft 15 or a welding method with a constant weaving width is used. A random shutter that can be synchronized with the weaving center position obtained by calculation (randomly synchronized with the center position signal) at the moment while detecting the weaving left and right ends by the detection signal from the limit switches attached to the left and right ends. An image is captured by a CCD camera 6 with a trigger function.

That is, 1/30 in the ordinary CCD camera 6
Synchronization is performed at a constant cycle every second or every 1/60 second. Until this synchronization is achieved, a maximum of 1/30 or 1/60 second is waited and a shifted image is taken.
According to the CCD camera 6 that synchronizes this randomly, an image can be captured at that moment, and an image without any shift can be obtained. In the case of welding without weaving, since the molten pool and the arc do not move,
It is possible to take an image with a normal CCD camera. In addition, in order to obtain a clear image that can be easily processed, the shutter speed is automatically set in accordance with the welding condition (current or groove shape). However, at this time, if the welding conditions are constant, the shutter speed may be constant.

The image can also be adjusted by operating the aperture, but the aperture is mechanically moved, which slows down processing. However, since the shutter speed is electrically controlled by an electronic shutter,
Advantageously, it can be performed at high speed.

As shown in FIG. 4, the fuzzy inference is based on the amount of left and right positional deviation ΔX, the moving speed V ′ with the previous torch moving direction, and the difference ΔH in the wet height of the molten pool, or
The difference Δρ in the curvature of the left and right corners of the weld pool, the difference ΔA in the gap area between the left and right weld pools, and the difference ΔG in the gap amount between the left and right weld pools were input, with the direction of the right and left scanning shaft 17 to which the welding torch 1 was attached. The moving speed V of the right and left scanning shaft 17 is output.

The fuzzy inference will be described with reference to FIG. 4A from which the difference ΔH in the wet height of the molten pool 14 is first removed. In FIG. 4 (A), 1) if the shift amount ΔX is ZR (ZERO: zero) and the moving speed V ′ with the previous torch moving direction is ZR (zero), the moving speed V with the torch moving direction is VR. Is ZR (zero)
To stop the torch 1. 2) Next, the left and right positional deviation amount ΔX of the torch 1 is PL (Posi
tive Large: large in the positive direction), and if the previous moving speed V ′ with the torch moving direction is ZR (zero), the moving speed V with the torch moving direction is set to NM (Negative Medium:
The movement of the torch 1 is controlled by setting an intermediate value in the minus direction). 3) It should be noted that the left and right positional deviation amount ΔX of the torch 1 is PL (Positi
ve Large: large in the plus direction), and if the previous moving speed V 'with the torch moving direction is PL (Positive Large: large in the plus direction), the moving speed V with the torch moving direction is NL (NegativeLarge: minus) And the torch 1 is controlled to move. As described above, the fact that the torch axis is moving in the plus direction even if the previous moving speed is PM although the torch axis is shifted in the plus direction means that the difference may be large, Command NL to move at a high speed in the minus direction.

Next, the wetting height deviation ΔH of the molten pool 14 is fuzzy inferred in addition to the above. FIG. 4 (B)
In, for example, the lateral shift ΔX is ZR (ZERO: zero)
, The wet height deviation ΔH of the molten pool 14 is NL
(Negative Large: large in the negative direction) because the left wet height is low, it melts better than the right, and PM (Positive Medium: an intermediate value in the positive direction) so that the welding torch 1 moves to the right. And the moving speed V with the torch moving direction.

Therefore, in the case where the deflection of the arc 3 and the influence of the base of the weld bead 4 are affected, it has been described that the wet height ΔH of the molten pool 14 is obtained and controlled by fuzzy inference. Also, as shown in FIG. 3, the curvatures ρ _L and ρ _{R of the} left and right corners of the molten pool and the gap areas A _L , A _R ,
And molten pool left and right gap amounts G _L, it is also possible to control by fuzzy inference even G _R, the same effect can be obtained. That is, when the curvature ρ _L of the left corner of the weld pool is smaller than the right curvature ρ _R , when the gap area A _{L on the} left side of the weld pool is larger than the gap area A _R on the right, and when the weld pool left the gap amount G _L of indicates that the undissolved compared both greater in comparison to the right of the gap amount G _R is retracts molten pool 14 of the left side to the right side of the molten pool.

In this case, it is necessary to obtain a clearer image in order to realize monitoring with high accuracy and high reliability. For this purpose, when the welding wire 2 is short-circuited to the molten pool 14, an image thereof is taken in, so that the welding arc 3 can be excluded, and an image showing only the molten pool 14 can be input. Thereby, the left and right corner curvatures ρ _L and ρ _{R of the} molten pool, the gap areas A _L and A _{R on} the left and right of the molten pool, and the gap amount G on the left and right of the molten pool.
_L, local images, such as G _R also accurately be measured. The detection of whether or not the welding wire 2 is short-circuited to the molten pool 14 is performed by monitoring the welding voltage V during welding and when the welding voltage V is small.
Or, when the short-circuit signal coming out of the welding machine 13 is ON,
The measurement can be accurately performed by capturing an image using a “CCD camera 6 with a random shutter trigger function” in which the shutter is randomly released as a short circuit of the welding wire 2.

Further, by determining the height Y _W of the welding wire insertion position, the position of the vertical shaft 16 of the welding torch 1 can be controlled.

[0027]

As described above, according to the present invention, welding can be performed without inclination and capable of uniformly melting the groove wall even when there is an influence of the deflection of the arc or the base of the welding bead. Can be obtained.

[Brief description of the drawings]

FIG. 1 is a control block diagram of a welding copying apparatus according to an embodiment of the present invention.

FIG. 2 is an explanatory view showing a processed image of a molten pool accompanied by arc deflection, and particularly shows the wet heights H _L and H _R of the molten pool at the left and right corners from the molten pool tip.

FIG. 3 is an explanatory view showing a processing image of a molten pool accompanied by arc deflection. In particular, corner curvatures ρ _L , ρ _{R on the} left and right of the molten pool, gap areas A _L , A _{R on} the left and right of the molten pool, and the molten pool. left and right gap amount G _L, showing a G _R.

FIGS. 4A and 4B are examples of tables showing rules for performing fuzzy inference regarding groove line tracing.

FIG. 5 is an explanatory diagram showing a weld pool following a groove line by a conventional image processing method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 welding torch 2 welding wire 3 arc 4 welding bead 5 groove 6 CCD camera 7 light shielding filter 8 camera control panel 9 servo motor control panel 10 image processing device 11 overall control device 12 TV monitor 13 welding machine 14 weld pool 15 weaving shaft 16 Vertical axis 17 Horizontal scanning axis

Claims (3)

[Claims] 1. A CCD camera for detecting a welding situation immediately below an arc during welding.
In a welding line profiling method that monitors a camera and recognizes features such as the shape of the weld pool during welding with an image processing device and performs welding line profiling in real time, an image of the weld pool including the welding wire insertion position during welding Is captured by a CCD camera, and this is image-processed.
From the shape of the weld pool, a welding wire insertion position shift ΔX with respect to the left and right edges of the groove is calculated, and further, a difference ΔH between the left and right wet heights of the weld pool, a difference Δρ between the curvatures of the left and right weld pools, From the correlation signals such as the area difference ΔA of the gap at the corner portion and the difference ΔG of the amount of the gap between the left and right molten pools, a melting state determination signal of the left and right groove walls is obtained, and the welding wire insertion position shift signal and the left and right groove walls are obtained. A welding line profiling method in which the center position of the left and right tracing axes is controlled in real time using arithmetic means such as fuzzy inference or the like in accordance with the molten state discrimination signal. 2. A welding process in which a welding condition immediately below an arc during welding with weaving is monitored by a CCD camera, a feature amount such as a molten pool shape being welded is recognized by an image processing device, and a welding line is traced in real time. In the line scanning method, the detection means detects that the weaver has reached the center of the weaving during welding, captures an image of the weld pool at the center of the weaving with a CCD camera, and processes this image.
From the shape of the molten pool, the welding wire insertion position deviation ΔX with respect to the left and right ends of the groove is calculated, and further, the deviation ΔH of the left and right wet heights of the molten pool, the difference Δρ in the curvature of the left and right molten pools, From the correlation signals such as the area difference ΔA of the gap at the corner portion and the difference ΔG of the amount of the gap between the left and right molten pools, a melting state determination signal of the left and right groove walls is obtained. A welding line profiling method characterized in that the center position of the left and right tracing axes is controlled in real time using arithmetic means such as fuzzy inference or the like to perform welding line profiling in accordance with the molten state discrimination signal. 3. The state of welding immediately below the arc during welding is measured by a CCD.
In the welding line profiling method, which monitors with a camera and recognizes features such as the shape of the weld pool during welding with an image processing device and performs welding line profiling in real time, the detection means detects that the welding wire has short-circuited to the weld pool. The image of the weld pool after the short circuit is detected and taken into a CCD camera, and the image processing is performed to calculate the welding wire insertion position deviation ΔX with respect to the left and right ends of the groove from the shape of the weld pool. The difference between the wetness height ΔH, the curvature difference Δρ between the left and right weld pools, the area difference ΔA between the corners at the ends of the left and right weld pools, the difference ΔG in the gap amount between the left and right weld pools, etc. The melting position determination signal of the front wall is obtained, and the center position of the left and right scanning axes is controlled in real time by using a calculating means such as fuzzy inference based on the welding wire insertion position shift signal and the melting state determination signal of the left and right groove walls. And melt A welding line copying method characterized by performing tangential copying.