JPS6320180A - Specific point detecting method on weld line - Google Patents

Abstract

PURPOSE:To detect the specific point on a weld line with a simple composition by making the shape of an arc light binary by picking it up from the upper part of the joint part of the member to be welded and analyzing the pattern in the arc light shape. CONSTITUTION:Prior to welding a weld line is detected from the picture image picked up by a pickup device 1 and the position of the above of a picture plane of a weld line is stored. A window with the weld line as the reference is set. This window has a weld line as Y axis and is set so as to include the place showing a peculiar shape. When a welding start command is placed by a robot control device, a detection control device takes a picture image data in, latching the picked up image on a video RAM and performing the bivalence of making the arc light part white and others black. A scanning is then performed horizontally from the left up for the window and the variation in the bivalence level is detected.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention is a method for directly viewing welding arc light and analyzing its shape pattern in arc welding, particularly lap joint welding, fillet joint welding, and butt joint welding. The present invention relates to a method for detecting a singular point of a weld line, that is, a welding end point, an outer corner point, or an inner corner point.

[Conventional technology]

In arc welding, the welding part, including the molten pool, current-carrying tip, wire, welding groove, etc., is imaged with an industrial television camera (ITV), and the resulting images are analyzed.
Methods of controlling welding positions, welding conditions, etc. have been known for a long time.

However, arc light is more intense than the brightness of the surrounding area (for example, the molten pool) and contains large amounts of infrared and ultraviolet rays, making it difficult to detect welded parts. .

Therefore, conventionally, the following solutions have been taken.

That is, one method is to add some innovation to the imaging device or the optical filter. For example, JP-A-51-7293
In Publication No. 8, in addition to providing filters to limit infrared rays, heat rays, color, and light intensity, a silicon vidicon image pickup tube is used in the television camera to pass radiation from the welding area only in the visible light region and near-infrared region. I try to let them do it.

Furthermore, in Japanese Patent Application Laid-Open No. 59-212172, an infrared television camera captures an image through a filter that allows only infrared rays to pass through. Furthermore, in JP-A-58-159980,
By using a CCD (solid-state image sensor) as the imaging device, no filter is required.

Another way to overcome the effects of arc light is to equip industrial television cameras with high speed shutters. for example,
In Japanese Unexamined Patent Publication No. 58-7776, the shutter is opened only when the short circuit occurs, taking advantage of the opportunity for the arc to disappear when the arc short circuits. Also, JP-A-59-202178
In this publication, in pulse welding, the shutter is opened only when the arc light during the base current period is attenuated.

As a method of imaging a welded part while distinguishing it from arc light, there is a method of providing a light source for imaging. For example, JP-A-5
In Japanese Patent No. 5-42185, a light source such as arc light that has a different spectral distribution from the disturbance light is provided, and a wavelength band in which the amount of the disturbance light is smaller than the amount of light from the light source is set as the detection wavelength. Further, in Japanese Patent Application Laid-Open No. 59-191574, a light source is provided for projecting light forward in the traveling direction of the welding torch to represent an image of the welding groove as a contour divided by shading.

[Problem that the invention seeks to solve]

What is common in the various conventional examples mentioned above is that
The object to be imaged is, for example, a welded part such as a molten pool, and the arc light is disturbance light that disturbs it, so it can be said that the object to be imaged is to be separated and removed from the arc light.

However, since the intensity of arc light is intense and its wavelength covers almost the entire range, it is difficult to completely remove it, and the equipment must be complicated and expensive. Furthermore, even if it could be removed, it would be extremely difficult to capture an image using a general industrial television camera. Even if an attempt is made to detect a welding end point or part of a corner by analyzing a captured image, for example, an image of a molten pool, complicated analysis must be performed for detection. In particular, a distinction has to be made whether the welding torch has reached an end point or a corner point, or whether it has shifted position.

Additionally, most sensors that use an external light source such as a laser beam currently detect steps in the welding groove using line information (slit light), which makes it difficult to stably detect welding end points and corner points. cannot be done. The laser beam method and wire contact method cannot detect errors caused by thermal distortion caused by welding.

The present invention was made in view of the above circumstances, and does not require an external light source such as a laser, nor does it require any special equipment for separating or removing arc light, and can be applied to a general industrial television. The purpose is to detect the welding end point of lap joint welding and fillet joint welding, and the inner and outer corner points by detecting arc light in the visible light range with a camera and analyzing the pattern of the arc light. .

[Means for solving problems]

In order to achieve this object, the method for detecting a singular point in a welding line of the present invention detects the welding groove position and performs arc welding along the welding groove position. Specifically, the shape of the light is captured by an imaging device, the captured arc light shape is binarized, and the pattern of the arc light shape is analyzed to detect a singular point in the weld line.

In the case of lap joint welding, the inventors first repeated a simple experiment in which a TV camera was installed above the weld joint to take images of the welding situation. In the image, only the outline of the arc light could be discerned.

However, because the weld line has a step in the joint, in the weld line in front of the torch that is not filled by the bead, the arc itself also has a step, which causes a part of the contour of the arc light imaging diagnosis to have a characteristic shape. It was found that

Furthermore, the present inventors have confirmed that this phenomenon also occurs in fillet joint welding and butt joint welding.

The present invention is based on the above knowledge, and focuses on the arc light itself, which was conventionally thought to be removed as disturbance light, and images the shape of the arc light seen from above at the joint. This method analyzes the shape and finds singular points in the weld line.

[Effect]

In the present invention, as shown in FIG. 1, an imaging device 1 such as a television camera is set on the joint. The arc 3 is emitted in a substantially conical shape from the tip of the wire 6, but since there is a step in the welding line 7, a step in the arc 3 occurs in the welding line in front of the torch that is not filled with beads. When this is imaged by the imaging device l on the joint, the stepped portion of the arc 3 can be detected as a characteristic portion. FIG. 2 shows the arc shape when the arc is progressing along the welding line 7, and FIG. 3 shows the arc shape when the arc is separated from the welding line.

For example, as shown in FIG. 4, when the welding end point is reached, the wall blocking the arc light disappears in front of the arc as it approaches the welding end point 8, so the arc extends in a direction around the wall of the workpiece. By detecting the characteristic portion where the arc extends only in front of the arc, the welding end point can be detected.

In addition, as shown in Figure 5, when reaching the outer corner point 9 of the joint, the wall blocking the arc light bends at a right angle in front of the arc, so the arc looks like a part of a circle cut out at a right angle. It becomes a shape. Thereby, the outer corner point can be detected.

The arc shapes in Figures 4 and 5 are almost indistinguishable, but whether the end point of the weld line is the end point of welding or the outer corner point from which welding continues can be determined in advance by welding. Since it is set in the control program,
The distinction can be made clearly.

Furthermore, as shown in FIG. 6, when the joint reaches the inner corner point, the arc light that has spread in front of the arc is blocked by the front wall of the work, resulting in a blunt shape at the front and sides.

Thereby, the inner corner point can be detected.

〔Example〕

Hereinafter, the present invention will be specifically explained based on Examples.

FIG. 1 shows an example of lap joint welding, where 1 is a torch, 2 is a wire, 3 is an imaging device, 4 and 5 are members to be welded, and 6 is an arc light. It is assumed that the torch l is moving in the advancing direction (direction toward the back of the drawing) while flying an arc.

At this time, the image captured by the imaging device 3 is binarized such that the arc portion is "white" and the other portions are "black" (shaded area), as shown in FIG.

The imaging device 3 is composed of a CCD camera, etc., and the image taken by it is sent to the detection control device, where it is image-processed according to a start command from the robot control device, and is also transferred to a CRT.
It is also sent to a display device and displayed.

Prior to welding, the weld line is first detected (101 in Fig. 1).
). This is detected from the image captured by the imaging device 3. FIG. 1 (al is an image thereof, in which a welding line 31, a wire 32, a wire shadow 33, and a torch 34 are shown.The welding line 31 is a line where two members to be welded come into contact with each other.

Next, the position of the welding line on the screen (on the coordinates set on the screen) is stored (FIG. 1 102).

Also, a window (area of the image to be processed) is set based on the welding line (103 in FIG. 1).

As shown in FIG. 1 (bl), this window is set so that the welding line is the Y-axis and a portion exhibiting a unique shape is included.

The above is the processing that should be performed in advance.

Now, when the robot control device 22 issues a command to start welding, the detection control device 23 takes in image data (Fig.
04).

Next, the captured image is latched onto the video RAM and binarized to make the arc light part white and the other parts black (see Fig. 10).
5).

Next, the window is scanned horizontally from the upper left to detect changes in the binarization level (Fig.
6).

Methods for clearly distinguishing the welding end point and the outer corner point include a method of detecting an inflection point A of a binarized arc pattern, a method of detecting an inflection point B, as shown in FIGS. 4 and 5. There are a method of detecting the inflection point C, a method of detecting the distance l of the extended arc light, etc.

In addition, as a method for clearly distinguishing the inner corner points, the second method is
A method of detection based on the disappearance of the characteristic part E in the figure, a method of detection from the characteristic part F or G in Figure 6, or a detection method based on the vertical distance e and horizontal distance d of the arc light and their ratio. There are ways to do this.

〔Effect of the invention〕

As explained above, in the present invention, arc light itself is used as a target for detecting a singular point in a weld line, and the singular point is detected by analyzing the shape of the arc light. Therefore, an expensive external light source such as a laser is not required, and the configuration is extremely simplified. In addition, since arc light itself is an intense light source, it is unaffected by other disturbance light and does not require special cut-off filters, band-pass filters, etc., and welding filters etc. to reduce the light intensity can be used. You only need to use . Furthermore, there is no need for a shutter or the like for synchronization during the base current of pulse welding used in a method in which arc light is regarded as disturbance light or during a short circuit in a short arc state. As a result, there are no restrictions on welding machines, types of wire, welding conditions, etc. Since a special wavelength of the arc light is not required, a general COD camera (sensitivity in the visible light range of 300 to 1000 nll) can be used without being restricted by the specifications of the television camera. Furthermore, since it is based on pattern analysis of arc light, it is also possible to detect welding of thin plate lap joints of one dragon or less.

Incidentally, the diameter of the arc light is about 5 fi, and in the past there has been no attempt to focus on such a minute image, and of course it was completely unknown that it had such a unique shape.

The present invention can be said to be an epoch-making invention that focuses on the fact that a captured image of arc light exhibits a shape having a specific relationship with the weld line, and finds a singular point in the weld line.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIGS. 2 to 6 each show the shape of a binarized pattern of arc light.
Figure 2 shows the case when there is an arc on the weld line, Figure 3 shows the case when it is off the weld line, Figure 4 shows the case when there is an arc at the welding end point, and Figure 5 shows the case when there is an arc at the outer corner point. Figure 6 shows the case where there is an arc at the inner corner point.

Claims (1)

[Claims] 1. When performing arc welding along the welding groove position while detecting the welding groove position, the shape of the arc light is imaged by an imaging device from above the joint part of the workpiece to be welded; 1. A method for detecting a singular point in a weld line, the method comprising detecting a singular point in the weld line by binarizing the imaged arc light shape and analyzing the pattern of the arc light shape. 2. A method for detecting a singular point in a weld line according to claim 1, characterized in that the imaging device is integrated with the welding torch, and the viewpoint of the imaging device follows the movement of the welding torch to capture the image.