JP5158406B2 - Welding situation analysis apparatus and method - Google Patents

Description

  The present invention relates to a welding condition analysis apparatus and method for determining in real time whether or not a welding state is good by using an image obtained by a weld visualization device.

  Patent Documents 1, 2 and the like have already been proposed as techniques for determining a welding state after welding when welding a metal plate by welding.

  In the apparatus and method of Patent Document 1, as shown in FIG. 10, the weld visualization device 60 includes a high-pixel CCD camera equipped with an electronic shutter and a high-intensity flash lamp. Synchronize camera shooting. The welding control device 62 includes an image capturing device that captures image data output from the weld visualization device 60, an image processing device that performs image processing on the image data to determine a welding state, and an output device that outputs the welding state. Is.

  The “laser welding inspection system” in Patent Document 2 aims to inspect the laser welding bead in real time, captures the image of the welding bead following the laser welding head, and the image processing is compared with the acceptable reference value. When it is determined that the weld is within a predetermined tolerance range, the weld is determined to be acceptable.

Japanese Patent Application Laid-Open No. 2004-74264, “Welding Portion Visualizing Device and Welding Control Device and Method Using the Same” JP 2006-198682 A, “Laser Weld Inspection System”

  The means of Patent Document 1 was invented and filed by the inventors of the present application, and enables visualization of images in the vicinity of a molten pool with a large amount of information, which can be used to analyze welding conditions with high accuracy. Has characteristics.

  The system of Patent Document 2 also aims to inspect the laser weld bead in real time by following the laser welding head and capturing an image of the weld bead.

However, the means of Patent Documents 1 and 2 have the following problems.
Welding may not be performed normally due to lens cracking of the welding laser, and the weld may be insufficiently melted. However, with conventional means, it is not possible to detect insufficient penetration during welding. Therefore, after the welding operation is completed, the weld penetration can be detected by inspecting the width of the bead and the lining of the welded portion by hand. Checking for deficiencies.
Therefore, it is necessary to perform the weld inspection work separately from the welding work, and there is a problem that the real-time property is lacking and the productivity is lowered.

  Further, even if an image of a weld bead is captured during welding by means of Patent Documents 1 and 2, it is difficult to determine in real time whether the weld state is good or bad.

  The present invention has been developed to solve the above-described problems. That is, an object of the present invention is to provide a welding state analyzing apparatus and method that can inspect a welded part in real time during welding and determine in real time whether or not a welding state including a defective weld due to insufficient penetration is good.

According to the present invention, the illumination of the welded portion by the pulse light source and the photographing of the welded portion by the CCD camera or the C-MOS camera are synchronized, and the image of the welded portion including the molten pool is continuously photographed to obtain a plurality of image data. A weld visualization device to obtain
There is provided a welding condition analyzing apparatus comprising: an image processing device that detects a bead width of a welded portion included in each image data and analyzes a welding condition from the bead width.

According to a preferred embodiment of the present invention, the image processing device uses a threshold value for a pixel having a certain luminance value or more to perform rounding to keep the luminance value low,
The edge extending along the bead is emphasized as a boundary between the dark part and the bright part, and a bead wrinkle is extracted from the edge to detect the bead width.

Further, the image processing device divides the bead extraction area into finer areas, dynamically determines the rounding threshold for each area,
In addition, bead wrinkle candidate points are extracted for each area, and bead wrinkle is extracted from the relationship with candidate points in other areas.

Moreover, according to the present invention, a weld visualization step for continuously capturing images of a weld including a weld pool and obtaining a plurality of image data;
An image processing step of detecting a bead width of a welded portion included in each image data and analyzing a welding state from the bead width is provided.

According to a preferred embodiment of the present invention, in the image processing step, for a pixel having a certain luminance value or more, rounding is performed to suppress the luminance value using a threshold value,
The edge extending along the bead is emphasized as a boundary between the dark part and the bright part, and a bead wrinkle is extracted from the edge to detect the bead width.

In the image processing step, the bead extraction area is divided into finer areas, and a rounding threshold is dynamically determined for each area.
In addition, it is preferable to extract a bead wrinkle candidate point for each area and extract a bead wrinkle from a relationship with a candidate point in another area.

  In a welding apparatus, when welding is not normally performed due to a lens crack of a welding laser, the bead width of the welding tends to increase. Therefore, it is possible to estimate the occurrence of poor welding from the bead width on the image. The present invention utilizes this characteristic to detect lack of penetration based on the weld bead width.

That is, according to the above-described apparatus and method of the present invention, the welded portion visualization device synchronizes the illumination of the welded portion with the pulsed light source and the photographing with the CCD camera or the C-MOS camera, and images the welded portion including the molten pool. A plurality of image data can be obtained by continuously photographing.
Further, the image processing apparatus can detect the bead width of the welded portion included in each image data, and analyze the welding situation from the bead width.
Therefore, it is possible to determine the quality of the welding state including a welding failure due to insufficient penetration from the obtained image by using a welding portion visualization device that visualizes an image near the molten pool at the time of welding.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. In each figure, common portions are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is an overall configuration diagram of a welding condition analysis apparatus according to the present invention. In this figure, 1 is a workpiece (material to be welded), 2 is a groove, 3 is a molten pool, 4 is a bead, 5 is a welding fume, 6 is a welding head, and 8 is a welding control device.

  The bead means a part where a part of the base material is melted and solidified by welding. As described above, in a welding apparatus, when welding is not normally performed due to a lens crack of a welding laser, the bead width of the welding tends to be widened. From this, it is possible to estimate the occurrence of welding defects.

However, the beads photographed by the welding visualization device have the following characteristics, and there are cases where stable bead width cannot be detected.
(1) It is basically darker than the portion (work) other than the bead on the image.
(2) There is a case where halation occurs.
(3) Basically, the image has a horizontal band shape in the horizontal direction.
(4) The bead boundary may be uneven due to the effect of soot or the like.
(5) It may appear white due to fumes (fine oxide particles formed when the component of the base material rises as vapor and cools in the air).

  The present invention uses these characteristics to detect a stable bead width and to detect insufficient penetration based on this.

  The workpiece 1 is a thin plate having a thickness of 1 to 3 mm, and shows a case where butt welding is performed via a groove 2. The welding head 6 is for laser welding and is controlled by a welding control device 8. In addition, this invention is not limited to these, Other material and thickness, lap welding, an arc, and a plasma welding head may be sufficient.

  In FIG. 1, the welding state analysis apparatus of the present invention includes a welded portion visualization device 10 and an image processing device 20.

  The welding part visualization device 10 includes a pulse light source (in this example, a high-intensity flash lamp 11), a lamp power supply unit 12, a lamp light quantity control unit 13, a camera head unit 14, a camera body 15, a control device 16, and a monitor 17.

The high-intensity flash lamp 11 is supplied with electric power from the lamp power supply unit 12 in accordance with a strobe pulse signal from the camera body 15 and illuminates the welded portion. The lamp light quantity control unit 13 applies a light quantity control voltage to the lamp power supply unit 12.
The pulse light source is not limited to a high-intensity flash lamp, and may be a laser light source such as a YAG laser harmonic.
The high pixel CCD camera includes a camera head unit 14 and a camera body 15. The camera head unit 14 incorporates a lens and a CCD (for example, 1.4 million pixels), and takes an image of a welded portion including the molten pool 3. The camera body 15 controls the camera head unit 14 and inputs captured image data (for example, 1320 × 1040 Pixel) to the control device 16.
A C-MOS camera or the like may be used instead of the CCD camera.
The control device 16 is a personal computer main body, for example, and controls the lamp light quantity control unit 13 and the camera main body 15 to output image data to the image processing device 20 and display it on the monitor 17.

  With the above-described configuration, the weld visualization device 10 continuously synchronizes the illumination of the weld by the high-intensity flash lamp 11 and the photographing by the high pixel CCD cameras 14 and 15, and continuously displays the image of the weld including the weld pool 3. A plurality of image data are obtained by photographing.

  The image processing device 20 is a personal computer main body, for example, and has a function of detecting a bead width of a welded portion included in each image data photographed by the welded portion visualizing device 10 and analyzing a welding situation in real time from the bead width. The control device 16 and the image processing device 20 may be configured by a single personal computer.

The image processing apparatus 20 uses a threshold value for pixels with a certain luminance value or more to perform rounding that keeps the luminance value low, and emphasizes an edge extending along the bead as a boundary between a dark portion and a bright portion. Then, bead wrinkles are extracted from the edges, and the bead width is detected.
Further, the image processing apparatus 20 divides the bead extraction area into finer areas, dynamically determines a rounding threshold value for each area, extracts bead wrinkle candidate points for each area, The bead trap is extracted from the relationship with the area candidate points.
Furthermore, the determination result by the image processing device 20 is input to the welding control device 8 and reflected in the control.

  FIG. 2 is an explanatory diagram of a monitor image of the weld visualization device described above. As shown in this figure, the monitor 17 of the weld visualization device 10 includes an image of a weld taken by a high pixel CCD camera, that is, a workpiece 1, a groove 2 and a seam 2a therebetween, a molten pool 3, and Bead 4 is displayed.

FIG. 3 is a pass / fail judgment flowchart in the image processing apparatus of the present invention. The image processing apparatus 20 according to the present invention automatically performs image processing on image data of a welded portion photographed by a high pixel CCD camera.
This image processing includes determination steps S1 to S9.

In step S1, a bead reference position is set by the user. That is, the user sets a bead reference position for ensuring consistency with the extracted bead width before starting welding.
(1) Find data whose bead width can be visually recognized from past welding data taken under the same conditions as the current welding conditions.
(2) The bead reference position frame 21 as shown in FIG. 4A is moved to an appropriate position on the horizontal axis so as to follow the bead of the past data searched in (1).
(3) The bead reference position is set by aligning the upper and lower sides of the bead reference position frame 21 with the upper and lower positions of the bead cage.

Next, welding is started (S12), and image data is acquired in step S2. That is, image data for performing image processing is acquired from an image captured by the weld visualization device 10.
(1) Taking a bead reference position 21 set by the user for each of the upper and lower bead ridges, taking a position at an appropriate length up and down with respect to that position, and obtaining a quadrilateral image with these sides as the upper and lower sides To do.
(2) In order to extract the candidate points for the bead wrinkles, as shown in FIG. 4B, the upper and lower acquired images are divided into a plurality of areas 22 divided into a plurality of pieces in the horizontal direction to set candidate point extraction areas ( S23) is performed.

In step S3, high luminance pixels on the image are rounded. That is, rounding of high-luminance pixels on the image is performed in order to prevent bead wrinkle candidate points from appearing unexpectedly due to the influence of halation.
(1) A threshold value for performing rounding by taking an average value of luminance values of bead portions in the area for each area 22 and multiplying that value by a constant k (for example, a constant of 0.8 to 1.0). Ask for.
(2) Using the threshold value of (1) for each area, a pixel whose luminance is higher than the threshold value is rounded to the threshold value.

In step S4, edge enhancement is performed. That is, the edge on the acquired image is emphasized.
(1) Using a predetermined filter that emphasizes a white to black portion in the vertical direction from the top to the bottom of the paper in the area on the upper side of the image in which the high luminance pixels are rounded in step S3, the white to black Emphasize changing edges.
(2) The area on the lower side of the image in which the high-luminance pixels are rounded in step S3 is displayed on the lower side of the screen by using another filter that emphasizes the black to white portion from the top to the bottom in the vertical direction. Emphasize edges that change to.

In step S5, the sum of luminance values is calculated. That is, the sum of luminance values is taken for each area.
(1) The luminance value of each pixel after edge enhancement is acquired.
(2) In each area, the luminance values of pixels having the same ordinate are added as shown in FIG.
In this figure, the area position is the first area from the right of the upper bead 、, the origin of coordinates is the upper left of the area, the horizontal line of 350 is the threshold value, and three circles are candidate points for the bead 淵It is.

In step S6, bead wrinkle candidate points are extracted. That is, a candidate point for bead cocoon for bead cocoon extraction is extracted.
(1) For each area, a portion (convex portion) larger than the threshold value is extracted from the sum of luminance values.
(2) The coordinate at which the sum of the luminance values is maximum in the extracted convex portion is set as the ordinate of the bead hail candidate point in the area. The abscissa of the candidate point is the median value of the abscissa of the area.
(3) (1) and (2) are repeated as long as there are convex portions. Therefore, a plurality of candidate points may be extracted.

It is confirmed whether the processing has been completed for all the extraction areas (S67), and when it is confirmed, bead soot is extracted in step S7. That is, a bead wrinkle for detecting the bead width is extracted.
(1) For each area, candidate points that are close to the reference position of the bead bottle set by the user are extracted.
(2) A straight line is fitted to these points using the least square method.
(3) In the straight line, when the least square error is larger than a certain threshold value or when the slope is larger than a certain threshold value, it is assumed in step 78 that the bead haze extraction has failed.

If the bead defect has been successfully extracted (step 78), the bead width is checked in step S8. That is, the bead width is collated, and the number of images with an abnormal bead width is counted.
(1) The set bead width on the image is obtained from the bead reference position set in step S1.
(2) The detected bead width (bead width at the median value of the abscissa of the bead reference position) is obtained from the bead cage extracted in step S7.
(3) An allowable bead width is obtained from the set bead width in (1). If the detected bead width in (2) exceeds the bead width, in step 81, the bead width is determined as an abnormal bead width. The number of abnormal images is counted (step 82).

If the bead width is normal in step 81, a welding abnormality is detected in step S9.
That is, at the end of welding, when the abnormal number count in step S8 is a certain ratio or more with respect to the number of acquired images in one welding, it is determined that the welding is abnormal and a warning is displayed (S91). After there is no welding abnormality or a warning is displayed, the detection process ends (S92).

A detection result when the detection algorithm of the present invention is implemented using actual welding-time image data photographed by the above-described weld visualization device 10 is shown.
In the detection, a bead wrinkle is extracted from the above-described image data and the process until the bead width can be detected is shown.

  As shown in FIG. 6, a bead position is set for the original image in step S1 for the above-described image data.

  As shown in FIG. 7, in step S23, the bead candidate point extraction area is set by dividing the area using the bead reference position.

  As shown in FIG. 8, steps S <b> 3 to S <b> 6 are repeated in each area to extract bead wrinkle candidate points.

  As shown in FIG. 9, in step S <b> 7, a bead cocoon is extracted from the bead candy candidate points.

  Thereby, it can be confirmed that the bead wrinkles are correctly extracted, and the bead width can be detected.

  As described above, in a welding apparatus, when welding is not normally performed due to a lens crack of a welding laser or the like, there is a tendency that the bead width of the welding is widened. Insufficient penetration is detected based on the bead width.

That is, according to the apparatus and method of the present invention described above, an image of a weld including a weld pool is obtained by synchronizing the illumination of the weld with a pulsed light source and photographing with a CCD camera or C-MOS camera, etc. Can be obtained continuously to obtain a plurality of image data.
Further, the image processing apparatus can detect the bead width of the welded portion included in each image data, and analyze the welding situation from the bead width.
Therefore, it is possible to determine the quality of the welding state including a welding failure due to insufficient penetration from the obtained image by using a welding portion visualization device that visualizes an image near the molten pool at the time of welding.

  In addition, this invention is not limited to embodiment mentioned above, Of course, it can change variously in the range which does not deviate from the summary of this invention.

It is a whole lineblock diagram of the welding situation analysis device of the present invention. It is explanatory drawing of the monitor image of a welding part visualization apparatus. It is a quality determination flowchart in the image processing apparatus of the present invention. It is explanatory drawing of step S2, S3 of this invention. It is explanatory drawing of step S5 of this invention. In this embodiment, the reference position is set on the original image. It is the image which set the area in the Example. It is an extraction image of a candidate point of a bead bowl in an example. It is the image of the bead basket obtained in the Example. It is a schematic diagram of the apparatus of patent document 1. FIG.

Explanation of symbols

1 workpiece (material to be welded), 2 groove, 2a seam,
3 molten pool, 4 beads, 5 welding fume (fume),
6 welding heads, 8 welding control devices,
10 weld visualization device, 11 pulse light source (high brightness flash lamp),
12 lamp power supply unit, 13 lamp light quantity control unit,
14 camera head, 15 camera body,
16 control device, 17 monitor, 20 image processing device

Claims (4)

A weld visualization device that synchronizes illumination of a weld with a pulsed light source and imaging of the weld with a CCD camera or C-MOS camera, and continuously captures images of the weld including the weld pool to obtain a plurality of image data; ,
An image processing device that detects a bead width of a weld included in each image data and analyzes a welding situation from the bead width;
The image processing apparatus includes:
(A) Emphasize the edge extending along the bead as a boundary between the dark part and the bright part,
(B) For each of a plurality of areas divided from the bead extraction area, the luminance value of a pixel having the same vertical coordinate is defined as a direction in which the plurality of areas are divided and along the bead. In addition, for each area, a portion larger than the threshold value is extracted from the sum of the luminance values, and the vertical coordinate at which the sum of the luminance values within the extracted portion becomes the maximum A welding condition analyzing apparatus characterized by extracting bead wrinkles from a relationship with candidate points in other areas, and detecting bead widths. Before the image processing apparatus (a),
Divide the bead extraction area into smaller areas,
For each area, take the average value of the luminance value of the bead part in the area, multiply that value by a constant, and dynamically determine the rounding threshold,
2. The welding state analyzing apparatus according to claim 1, wherein , for each area, for a pixel having a certain luminance value or more, rounding is performed using the threshold value to reduce the luminance value low. A weld visualization step for continuously capturing images of a weld including a weld pool and obtaining a plurality of image data;
An image processing step of detecting a bead width of a weld included in each image data, and analyzing a welding situation from the bead width;
In the image processing step,
(A) Emphasize the edge extending along the bead as a boundary between the dark part and the bright part,
(B) For each of a plurality of areas divided from the bead extraction area, the luminance value of a pixel having the same vertical coordinate is defined as a direction in which the plurality of areas are divided and along the bead. In addition, for each area, a portion larger than the threshold value is extracted from the sum of the luminance values, and the vertical coordinate at which the sum of the luminance values within the extracted portion becomes the maximum A welding situation analysis method characterized by extracting bead wrinkles from a relationship with candidate points in other areas, and detecting a bead width. Before (a)
Divide the bead extraction area into smaller areas,
For each area, take the average value of the luminance value of the bead part in the area, multiply that value by a constant, and dynamically determine the rounding threshold,
For each area, with respect to certain luminance value or more pixels, by using the threshold value, it performs a rounding kept low luminance value, a welding condition analysis method according to claim 3, characterized in that.

Images