JPH04266480A - Method and device for detecting welding position - Google Patents

Abstract

PURPOSE:To furnish a welding position detecting method and its device which are applied to weld line profile work, enables to perform weaving welding with excellent profile accuracy and further, enables to realize automatically profile welding with high accuracy without being affected on weaving operation even for complicated multilayer welding. CONSTITUTION:This welding position detecting method and its device are characterized by providing a groove position detector 19 provided with a light source to irradiate a weld line with light and a photodetecting means to detect reflected light from the direction to form a specified angle with respect to the surface formed by irradiation light from the light source, a picture processor 20 which executes picture processing from a video signal observed by the photodetecting means and detects the groove position, a torch position controller 17 to move a welding torch and a laminating condition generating device 21 which connects the interval between the above-mentioned picture processor and torch position controller by a communication control circuit and communicates the control data and processing data.

Description

[Detailed description of the invention]

0001

TECHNICAL FIELD The present invention relates to a welding position detection method and apparatus, and more particularly to a welding position detection method and apparatus suitable for welding line tracing work used in fully automatic welding robots, for example. .

0002

2. Description of the Related Art An example of a conventional welding method will be described with reference to FIGS. 14 and 15. Figure 14 shows
An explanatory diagram showing a conventional weaving welding method,
FIG. 15 is a perspective view showing a conventional optical groove detection method. In FIG. 14, 8 is the lower plate of the fillet joint, 9 is the vertical plate, A is the welding start point, B is the welding end point, and the line segment connecting AB is the welding line.

[0003]Currently, when welding line AB is to be welded by teaching playback using a welding robot, the welding start point A point, the welding width points H1 and H2, and the welding end point B point are taught in advance by teaching. I'll keep it. After teaching in this manner, a playback operation is started, and at this time, weaving path calculation is executed using the four teaching data, and weaving welding is performed along the path shown by the broken line in the figure. As described above, normal welding work is performed only when the actual welding workpiece is accurately placed at the taught position, but otherwise welding is not performed correctly. For this reason, a method is used in which the actual welding line is detected by a sensor and the teaching line is corrected using the signal.

[0004] Various types of sensors have been developed so far, but a typical conventional sensor for optically detecting a welding line is, for example, a slit light and two-way sensor as shown in FIG. There is one based on dimensional light receiving means. Light projecting means 2
A laser beam 11 sharply focused into a slit shape is irradiated onto the grooves preceding the arc point of the lower plate 8 and vertical plate 9, which are the members to be welded, and the reflected light 12 from the grooves is captured by an ITV camera, etc. The position Q2 to be welded is determined by detecting the light with the two-dimensional light receiving means 3 and processing and analyzing Q1, Q2, and Q3 from the obtained light section line image.

[0005] Applying such a device to a welding robot,
When controlling a welding torch, for example, as shown in FIG. 16, a welding torch 1, a light projecting means 2, and a light receiving means 3 are integrated into a torch equipped with a detection device, which is attached to the tip of an arm of a robot 5. FIG. 16 is an external view of a general device in which a detection section is attached to a welding robot. In FIG. 16, 4 is an image processing device, 6 is a robot control device, and 7 is an image processing device.
indicates a welding machine. Regarding such technology, please refer to the Welding Society, 128th Welding Methods Committee Materials (August 7, 1990).
(Daily) ``About the automation of thick plate welding using visual sensing''.

[0006]

Problems to be solved by the above-mentioned prior art will be explained with reference to FIG. 17. FIG. 17 is an explanatory diagram showing a case where the positional relationship between the taught work and the actual work is deviated. In the above conventional method, the actual welding line A with respect to the teaching line A-B of the welding workpiece shown in FIG.
'-B' can be detected, but it is not possible to know how much the actual workpiece is rotating around A'-B'. As a result, if weaving welding was performed by simply correcting the position of only the welding teaching line A-B to A'-B' using the welding robot, the target positions on the vertical plate and lower plate side would shift and accurate welding would not be possible. As a result, there was a problem that welding quality deteriorated.

Furthermore, when the weld bead is multi-layered and weaving welding is applied, the groove position detection method using the optical detection method described above is used to detect each welded layer when detecting the second and subsequent layers. Since the obtained groove light cutting image changes, the groove position to be welded cannot be detected with a single process, and there is a problem in that it is not possible to accurately aim at the position during weaving and perform welding.

The present invention has been made in order to solve the problems of the prior art described above, and is capable of controlling the weaving by sensing not only the groove center position but also the weaving position (amplitude and direction), and has good profiling accuracy. To provide a welding position detection method and device that enable weaving welding,
This is the first purpose. The second object of the present invention is to provide a welding position detection method that can automatically realize high-precision profile welding without being affected by weaving operations even in complex multi-layer welding. The goal is to provide equipment.

[0009]

[Means for Solving the Problems] In order to achieve the above object, a welding position detection method according to the present invention includes a light projecting means equipped on a welding torch, and a two-dimensional light receiving means for detecting reflected light. It has a groove position detection device that irradiates the welding line of the workpiece with slit-shaped light from the light projecting means, and detects the welding line from a direction that makes a certain angle with respect to the plane formed by the irradiated light. In the welding position detection method, which detects the position of the weld line by receiving reflected light by a light receiving means and calculating the groove light cutting image detected by the light receiving means, the groove of the workpiece is detected at the welding start point. Any 3 including welding lines for each constituent surface
The three-dimensional positions of the points are determined by the groove position detection device, and the weaving vectors of weaving welding regarding the groove surface are calculated from the three-dimensional positions of the three points,
This allows the start point position teaching data and weaving vector teaching data to be corrected.

[0010] In order to achieve the above object, the welding position detection device according to the present invention has a structure including a light projector that is attached integrally with a welding torch and irradiates a slit-shaped light onto the welding line of the workpiece. and a two-dimensional light receiving means for detecting reflected light from a direction forming a constant angle with respect to the plane formed by the irradiated light from the light source. , an image processing device that performs image processing to detect the groove position from the video signal observed by the light receiving means, a torch position control device that moves the welding torch, and the image processing device and the torch position control device, respectively. and a lamination condition generating device that communicates control data and processing data by connecting the two with a communication control circuit.

Furthermore, in order to achieve the second object, the lamination condition generation device in the welding position detection device according to the present invention includes a knowledge base loader that inputs work management data of multilayer welding via a storage medium. , a welding condition table that stores data related to welding conditions among the welding management data, a lamination condition table that stores data related to lamination conditions, and a CPU that creates lamination conditions for multilayer welding.
and a working memory that stores the created lamination conditions.

0012

[Operation] The operation of the present invention will be explained using a fillet joint as an example. First, in order to obtain the equation of the plane of the surface of the vertical plate forming the fillet joint, the workpiece positions at two arbitrary points on the vertical plate are detected by operating the above-mentioned optical detection device. Next, the position of the workpiece at two arbitrary points on the lower plate, which is another member forming the fillet joint, is detected by operating the optical detection device. Thereafter, the detection device is guided to the welding start point taught in advance to detect the groove position. The detection position here corresponds to one measurement point on each surface of the vertical plate and the lower plate forming the fillet joint. As a result of the above, a total of three positions can be detected on each of the vertical plate and the lower plate. If the positions (coordinates) of any three points are known, the equations for the surfaces of each of the vertical plate and the lower plate can be determined, although the details will be described later.

In the welding work shown in FIG. 11, which will be described later, if the plane equations of the surfaces of the vertical plate and the lower plate are known, it is possible to find the vectors of weaving in the vertical plate and lower plate directions when weaving is performed. This allows the previously taught aiming position to be corrected. In addition, for multi-layer welding using weaving welding, welding can be performed by detecting the groove center position and the first and second corner points that indicate the weaving amplitude, and performing tracing control using these results. This makes it possible to accurately follow the desired position. The image processing software (position to be detected, number of detection points, etc.) for each layer weld line in multi-layer welding is determined by referring to work data in a table (memory) prepared in advance, so complex multi-layer welding can be automated. This makes it possible to perform control based on the model.

[0014]

[Embodiment] An embodiment of the present invention will be described below with reference to FIGS. 1 to 13.
Explain with reference to. FIG. 1 is a block diagram of a welding position detection device according to an embodiment of the present invention, FIG. 2 is a perspective view showing an example of a welding method, and FIG. 3 is an embodiment of a welding method to which the present invention is applied. FIG. In FIG. 1, 15 is a welding workpiece, 16 is a torch position control mechanism, and 17 is a welding workpiece.
1 is a torch position control device, 18 is a welding machine, 19 is a groove position detection device, 20 is an image processing device, and 21 is a lamination condition generation device. Among these, the torch position control mechanism 16 is, for example, a welding robot shown in FIG. 16. The groove position detection device 19 is composed of a slit light projecting means 2 and a two-dimensional light receiving means 3 shown in FIG. There is. The image processing device 20 processes an image obtained by converting the video signal observed by the light receiving means 3 into a digital signal to detect a desired groove position.

Further, the torch position control device 17 and the image processing device 20, the welding machine 18 and the lamination condition generation device 21, and the lamination condition generation device 21 and the image processing device 20 are connected to each other via communication control circuits (28a to 28f). and communicate control data, processing data, etc. In FIG. 1, 22 to 27 are constituent elements of the lamination condition generating device 21, which will be described later.

Among the embodiments of the present invention, a method of performing welding tracing by correcting the teaching data of the weaving center position and amplitude when performing weaving welding will be described first. FIG. 2 shows weaving welding of a fillet joint as an example of a welding method. In this figure, parts equivalent to those in FIG. 14 described above are designated by the same reference numerals. That is, A indicates the welding start point, B indicates the welding end point, the line connecting AB indicates the welding line, and H1 and H2 indicate the weaving width, respectively. The points A, B, H1, and H2 are taught in advance by teaching, and during playback operation, welding is performed along the path shown by the broken line in the same figure by calculating the weaving path using the teaching data of these four points. Ru. The present invention allows a normal welding operation to be carried out even when the actual welding work is not placed accurately at the taught position, and this will be explained using FIG. 3.

In FIG. 3, the work indicated by the two-dot chain line is the work at the teaching position shown in FIG. here,
This will be explained using an example in which the workpiece is shifted to the position shown by the solid line. First, equations of the planes of the surfaces of the vertical plate 9 and lower plate 8 forming the fillet joint are determined. For this purpose, the vertical plate 9
For this, the actual workpiece position at any two points (S1, S2) in FIG. 3 is detected by operating the optical detection device described above (points S1, S2 are taught in advance). FIG. 3 shows the appearance when the three-dimensional position of point S1 is detected. That is, a slit light is irradiated near S1, and a light sectioned image obtained at this time is observed and image processed to recognize the bending point of the light sectioned image and detect the three-dimensional position of point S1. After detecting two points on the vertical plate 9, we next detect two arbitrary points (S4, S5) on the lower plate 8, which is another member forming the fillet joint.
), the optical position detection device is operated in the same manner as described above to detect the three-dimensional position. Thereafter, the detection device is guided to the welding start point taught in advance to detect the groove position. The detection position (A') here corresponds to one measurement point on each surface of the vertical plate 9 and the lower plate 8 forming the fillet joint. As described above, a total of three positions were detected on each of the vertical plate and the lower plate.

Now, the three-dimensional position coordinates of the three points S1, S2, A' on the vertical plate 9 are (X1, Y1, Z1), (X2, Y
2, Z2), (X3, Y3, Z3), the equation (P1) of a plane passing through these three points is expressed by the following equation.

[Math 1]

[Math. 2] Or aX+bY+cZ+d=0...
...(2) Here a=Y1Z2+Y2Z3+Y3Z1-(Y1Z
3+Y2Z1+Y3Z3)……(3) b=X1Z3+X2Z1+X3Z2−(X1Z
2+X2Z3+X3Z1)……(4) c=X1Y2+X2Y3+X3Y1−(X1Y
3+X2Y1+X3Y3)……(5) d=X1Y3Z2+X2Y1Z3+X3Y2Z
1-(X1Y2Z3+X2Y 3Z1+
X3Y1Z2)……(6)

Similarly, the three-dimensional position coordinates of the three points S4, S5, and A' on the lower plate 8 are expressed as (X
4, Y4Z4), (X5, Y5, Z5), (X3, Y3
, Z3), the equation of the plane of the lower plate 8 passing through these three points (
P2) is (X1, Y1, Z1), (X2,
The detected values of Y2, Z2) are (X4, Y4, Z4), (X5
, Y5, Z5). Next, when performing weaving welding by executing the playback operation, the welding start point (and weaving center position)
The position of the teaching data A is corrected to the position of the detection point A', and the weaving amplitudes H1 and H2 are calculated using the coordinate position of A' and the plane formulas P1 and P2, respectively.
Calculate 2'. In this way, it is possible to perform copy welding on the vertical plate and lower plate sides by matching the target weaving position to the actually installed workpiece position.

Next, FIG. 4 shows a method of performing weld tracing by correcting the teaching data of the weaving center position and amplitude when performing weaving welding in multi-layer welding, etc.
Explain with reference to. FIG. 4 is an explanatory diagram showing an example of a welding construction method of multilayer welding, and is a cross section of a bead in a state where the final weld bead has been formed. In Figure 4, 8A
, 9A is the part to be welded, 35 is the backing metal, and the numbers ■～■
is the welding pass number. Further, Table 1 shows an example of welding conditions in this welding process. In this example of welding conditions, as shown in Table 1, passes (■) and (■) form the fifth layer. Furthermore, regarding the presence or absence of weaving, welding is not performed in the ■ and ■ passes, and welding is performed by weaving from the ■ pass to the final ■ pass.

[0021] The case of automatically copying multilayer welding will be described below with reference to FIGS. 5 to 13 in conjunction with FIGS. 1, 4, and Table 1. FIG. 5 is an image processing flowchart in the image processing device, FIG. 6 is a schematic diagram showing the external appearance at the time of detecting the groove position, FIGS. 7 to 10 are explanatory diagrams showing the groove image and the image after image processing, and FIG. 12 and 13 are explanatory diagrams showing a method for welding the fourth layer of multilayer welding, and FIGS. 12 and 13 are explanatory diagrams showing a method of processing images for the fourth layer of multilayer welding. In this embodiment, the above-mentioned groove position detection method is determined in advance according to each welding pass according to the multilayer welding conditions shown in FIG. 4 and Table 1, and recognition and welding tracing control are executed in accordance with this method.

First, the processing flow will be described with reference to FIG. 5, taking as an example the case where the welding tracing control for the first layer is executed. Further, the external appearance at the time of detecting the groove position is schematically shown in FIG. 6 (the welding torch is omitted from the illustration), and an example of a light section image in this case is shown in FIG. As shown in step F1 in FIG. 5, when receiving the detection ON instruction from the torch position control device 17, the image processing device 20 responds that it has received the detection ON instruction (step F3), and first detects the The presence or absence of weaving is determined from the welding condition data sent with the command (step F4). Since the first layer pass (2) is a condition in which there is no weaving, the process proceeds to step F6, and the image processing device 20 captures a groove image and stores the image. The outline of the processing from step F7 to step F10 will be described below with reference to the images shown in FIGS. 7, 8, 9, and 10.

In step F7, optical noise (Hume image H1, sputter images S1, S2, S4 scattered in the air) other than the optically sectioned images L1, L2, and L3 included in the stored image shown in FIG. , a process for removing the sputter image SP attached to the work surface and a binarization process are performed, and only the optically sectioned images L1, L2, and L3 are extracted as shown in FIG. Next, in step F8, the horizontal center line of the light section line is determined, and this data is used to determine the vertical continuity of the screen to determine whether the light section line is continuous or discontinuous. , and the resulting collinear elements are classified and labeled as shown in FIG. 9. next,
In step F9, the labeled line element data is used to open the welding pass number with reference to the data regarding this welding pass number in the image recognition condition table 33 created in advance in correspondence with the multilayer welding conditions in Table 1. Integrate the base lines of the previous image. The results are shown in FIG. In this case, there are three straight portions L1, L2, and L3.

Furthermore, in step F10, corner points W1 and W2 of the groove are detected with reference to a processing procedure created in advance in the image recognition condition table 33. Since the detection result of this groove corner point is the coordinate data on the imaging plane of the two-dimensional photodetector, this data is used to determine various constants of the optical system (the optical The coordinate system is converted into the coordinate system of the torch position control mechanism using coordinate conversion parameters determined by the system arrangement, imaging magnification, etc. (step F11). In the next step F12, referring to the calculation procedure created in advance in the recognition condition table 33,
Calculate the target position of the welding torch with this welding pass number. Then, the calculation result of the target position of the welding torch is transferred to the torch position control device 17 in step F13. The torch position control device 17 uses this data to control the torch position. After the calculation result of the target position of the welding torch is transferred to the torch position control device 17 in step F13,
Check whether the program step being played back has switched to the next step number, and if it has switched to a new step number, step F
1, and if not, return to step F4 and repeat the same processing operations as described above. The above is a case where the welding tracing control for the first layer (pass number ■) in FIG. 4 is performed.

Next, the fourth layer (pass number ■) in FIG.
Taking as an example, a case where tracing control is performed using weaving welding will be described with reference to FIG. As shown in FIG. 11, in this case, the welding progresses as shown by arrow A, with the target position being the corner C1 of the workpiece 9A and the bead of the third layer, and the vicinity of the corner C2 of the workpiece 8A and the bead of the third layer. The torch is swung perpendicular to the direction to copy. In this embodiment, as shown in the flowchart of FIG. 5, detection of a groove image in a welding pass with weaving involves determining where during weaving the groove image is input from the torch position control device 17 side. This is carried out in response to a detection command issued from the control device 17 side. That is, in this case, the torch position control device 17 issues a detection command when the welding torch reaches the maximum amplitude position of weaving welding (positions corresponding to the two corner points C1 and C2). In response to this detection command, the process proceeds to step F6 and starts the operation of capturing a groove image. Figure 12
, 13 respectively show schematic diagrams (disturbing light is omitted) of images obtained near the corner points C1 and C2.

In the processing in the above-described flowchart, the integration of the base line (step F9) and the detection of the groove position (step F10) are performed using this pass number created in advance in the image recognition condition table 33. Detect the corner points W1 and W2 of the groove with reference to the processing method for (2) above. For example, the corner point W1 is detected from the groove image in FIG. 12, and the corner point W2 is detected from the groove image in FIG. Therefore, in the basic line integration (step F9), only L1 and L2 are extracted for the groove image in Fig. 12, and only L2 and L3 are extracted for the groove image in Fig. 13. The processing method for the pass number ■ may be created in the image recognition condition table 33 in advance. There is also a limit to the size of the field of view of the imaging section of the light receiving means 3, so when the bottom portion L2 of the groove becomes wide (the length B in FIG. 13 is large), the light cut image is Since the entire image cannot be stored in one screen, it is essential that the two images be input and processed separately. This makes it possible to trace the weld line even on a workpiece with weaving welding.

Next, another method of copy welding using weaving will be explained. This is the case when performing weaving welding in a bead shape as shown by pass number ■ or ■ in FIG. In this case, in the process of the pass number ■ described above, point C1 (if it corresponds to the image, W
Point C1' or point C1'' corresponding to 1) can be detected, but the point corresponding to point C2 on the opposite side is (a) out of the field of view (in the case of point C2') (b) when the point on the workpiece surface is detected. If the step is smooth and cannot be detected by image processing (C2'
´ point) etc. In such a case, one side (W
From the detection data of 1) and the weaving amplitude length (previously created in the image recognition condition table 33), the other position that cannot be detected is also calculated (step F12).
) Perform welding. As described above, by creating each processing method in processing steps F9, F10, and F12 in advance in the image recognition condition table 33 for each pass number, copy welding becomes possible even in complicated multilayer welding. Furthermore, even if the welding workpiece deviates from the taught position, welding can be performed with high accuracy.

Next, the lamination condition generator 21 for determining the lamination conditions for multilayer welding according to the present invention will be explained with reference to FIG. The lamination condition generator 21 is a CP as shown in the figure.
U22, working memory 23, welding condition table 25
, a stacking condition table 26, a knowledge base loader 27, and an input/output interface 29. The operator first inputs work management data for multilayer welding into the knowledge base loader 27 via the lamination condition generator 21 via a storage medium such as a floppy disk. The work management data includes, for example, the following contents. (1) Total number of multilayer welds (2) Number of passes in each layer (3) Presence or absence of weaving (4) Groove image processing software for each welding pass (groove shape and detection location) (5) Aim for each welding pass Position correction amount (6) Of the work management data above the welding current and arc voltage in each welding pass, (1) to (5) are stored in the lamination condition table 26, and the welding machine 18 in (6) Related data is stored in the welding condition table 25.

The CPU 22 creates lamination welding conditions for multilayer welding with reference to the contents of the work management data, and further transmits the lamination welding conditions to the working memory 23 to be stored therein. During playback operation, the image processing device 20 and the lamination condition generation device 21
The welding machine 18,
Communication control circuits 28a and 28f are connected between the lamination condition generator 21.
Communicate necessary control data etc. with each other via
Perform weld copying work.

As shown in FIG. 1, the apparatus of this embodiment has the following features:
Conventional torch position control mechanism 16, torch position control device 1
7 and a welding machine 18, a groove position detection device 19, an image processing device 20, and a lamination condition generation device 21 are added. Therefore, as long as the function of detecting and copying the welding position and the function of determining the lamination conditions for multilayer welding are not activated, the conventional teaching and playback methods can be used as is.

Furthermore, in order to determine the lamination conditions for multilayer welding, the operator inputs work management data for multilayer welding via a storage medium such as a floppy disk. On the other hand, during teaching, work management data for multi-layer welding is stored in a working memory (not shown) in the torch position control device 17, and during playback, this data can be used as it is for execution, or a part of it can be input. It is also possible to use it as data to determine the lamination conditions for multilayer welding.

[0032]

[Effects of the Invention] As explained in detail above, according to the present invention, for weaving welding, there is a detection function for determining the plane equation representing each surface of the vertical plate and the lower plate, and a function for detecting the groove position. We provide a welding position detection method and device that enables welding welding with good tracing accuracy, since it is possible to perform welding control by sensing not only the groove center position but also the weaving position (amplitude and direction). can do.

In addition, for multi-layer welding using weaving welding, the welding center position, the upper side of the weaving, and the lower side of the weaving can be synchronized with the detection command from the torch control mechanism without being affected by the weaving operation. Since the welding line position result can be obtained by processing the groove image, by performing tracing control using this result, it becomes possible to accurately trace and control the position to be welded. Furthermore, the present invention stores work management data such as the number of multilayer welding, the number of passes in each layer, presence or absence of weaving in each layer, and the position to be welded in a table (memory) provided in advance in multilayer welding, This is referenced during the playback operation, and the welding position is detected based on the same groove detection results as described above, so this welding position detection method can automatically realize high-precision profile welding even for complex multi-layer welding. and the equipment thereof.

[0034]

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a welding position detection device according to an embodiment of the present invention.

FIG. 2 is a perspective view showing an example of a welding method.

FIG. 3 is a perspective view showing an embodiment of a welding method to which the present invention is applied.

FIG. 4 is an explanatory diagram showing an example of a welding method of multilayer welding.

FIG. 5 is an image processing flowchart in the image processing device.

FIG. 6 is a schematic diagram showing the appearance when detecting the groove position.

FIG. 7 is an explanatory diagram showing a groove image in the image processing device.

8 is an explanatory diagram showing an image obtained by removing optical noise from the image of FIG. 7. FIG.

9 is an explanatory diagram of an image showing line elements from the image of FIG. 8; FIG.

10 is an explanatory diagram of an image obtained by integrating the basic lines of the groove image from the image of FIG. 9; FIG.

FIG. 11 is an explanatory diagram showing a welding method for the fourth layer of multilayer welding.

FIG. 12 is an explanatory diagram showing a method of processing an image in the fourth layer of multilayer welding.

FIG. 13 is an explanatory diagram showing a method of processing an image in the fourth layer of multilayer welding.

FIG. 14 is an explanatory diagram showing a conventional welding method.

FIG. 15 is a perspective view showing a conventional optical groove detection method.

FIG. 16 is an external view of a general device in which a detection section is attached to a welding robot.

FIG. 17 is an explanatory diagram showing a case where the positional relationship between the taught work and the actual work is deviated.

[Explanation of symbols]

1 Welding torch 2 Light projecting means 3 Light receiving means 4 Image processing device 15 Welding work 16 Torch position control mechanism 17 Torch position control device 18 Welding machine 19 Bevel position detection device 20 Image processing device 21 Lamination condition generator 22 CPU 23 Working memory 25 Welding condition table 26 Lamination condition table 27 Knowledge base loader 33 Image recognition condition table

Claims (6)

[Claims] [Claim 1] Light projecting means equipped on a welding torch;
It has a groove position detection device equipped with a two-dimensional light receiving means for detecting reflected light, and irradiates a slit-shaped light from a light projecting means to the welding line of the workpiece, and detects the surface formed by the irradiated light. In a welding position detection method, a light receiving means receives reflected light from a direction forming a certain angle to The three-dimensional positions of arbitrary three points including the weld line are determined by the groove position detection device on each surface constituting the groove of the workpiece at the starting point, and the groove surface is determined from the three-dimensional positions of the three points. A method for detecting a welding position, characterized in that the weaving vectors of each weaving weld are calculated, and the starting point position teaching data and the weaving vector teaching data are corrected based on the weaving vectors. 2. A device for moving a welding torch, the device issuing a command to detect a groove image at the center or maximum amplitude position of weaving, and actuating a groove position detection device based on the image detection command. 2. The welding position detection method according to claim 1, further comprising detecting a welding line. 3. The welding position detection method according to claim 1, wherein an image processing recognition condition table is created according to each welding pass in multilayer welding, and image processing recognition is executed in accordance with the table. 4. The image processing recognition condition table includes the basic line of the groove light cutting image, its integration method, the detected location, the number of detected points, and the target position of the welding torch relative to the detected location. Welding position detection method. [Claim 5] A light projecting means that is attached integrally with the welding torch to the welding line of the workpiece and irradiates a slit-shaped light, and a light projecting means that forms a certain angle with respect to the plane formed by the irradiated light from the light source. In a welding position detection device having a groove position detection device equipped with a two-dimensional light receiving means for detecting reflected light from a direction, the groove position is detected by image processing from a video signal observed by the light receiving means. A stacking condition in which an image processing device for detecting, a torch position control device for moving the welding torch, and a communication control circuit connects each of the image processing device and the torch position control device to communicate control data and processing data. A welding position detection device comprising a generator. 6. The lamination condition generation device comprises: a knowledge base loader that inputs work management data of multilayer welding via a storage medium; a welding condition table that stores data related to welding conditions among the welding management data; 6. The method according to claim 5, comprising a lamination condition table that stores data related to lamination conditions, a CPU that creates lamination conditions for multilayer welding, and a working memory that stores the created lamination conditions. Welding position detection device.