JP2017217662A - Diagnostic system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a diagnostic system capable of diagnosing surely a weld state of a welding object member by reducing deficiency of a photographing field caused by arc light, without enlarging a welding torch part.SOLUTION: A diagnostic system has: a front light receiver for acquiring light from the front of a moving welding torch; a rear light receiver for acquiring light from the rear of the welding torch; an image generation part for generating an image from light acquired by the front light receiver and light acquired by the rear light receiver; and a diagnosis part for diagnosing a weld state of a weld object member from the image generated by the image generation part.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a diagnostic apparatus for diagnosing a welding state.

  Arc welding is a technique for welding two metal members using an arc discharge phenomenon, and is used in many industrial fields. In arc welding, there are many parameters such as a welding current and a welding voltage, a flow rate of a shielding gas, a shape of a discharge electrode, and a distance between a metal member and an electrode. In order to form a sound weld, it is necessary to adjust these parameters according to the material and shape of the metal member.

  Patent Document 1 describes an apparatus that facilitates this parameter adjustment by photographing an arc welded portion and visualizing a welding situation. This Patent Document 1 states that “the welding site visualization device of the present invention is configured to receive light incident from a welding arc formed between a site to be welded and a welding torch part and a scene of a peripheral region heated by the welding arc. A light guiding means for guiding, a first adjusting section for adjusting a brightness suitable for photographing with respect to a welding arc light guided by the light guiding means, and a light incident from a scene of a peripheral area guided by the light guiding means. A light adjusting unit having a second adjusting unit for adjusting to a luminance suitable for photographing, an image signal recording unit for recording light adjusted by the light adjusting unit as an image signal, and an image signal output by the image signal recording unit. Image combining processing means for combining, and a plurality of light guiding means and light adjusting means constitute a plurality of optical paths. ” According to Patent Document 1, “the light incident from the scene of the welding arc light and the peripheral area heated by the welding arc has a large luminance difference. By providing the first adjustment unit and the second adjustment unit with different light intensity adjustment performance, it is possible to set the optimum observation conditions for each incident light, and at the same time visualize the situation of the welding arc and the surrounding area. be able to. According to Patent Document 1, “Because the light guiding means is provided, the photographing torch such as a camera is not directly installed at the tip of the welding torch, so that the welding torch can be further downsized. Is possible. "

JP 2007-1108055 A

  Patent Document 1 describes a welding processing apparatus that visualizes the state of a welding arc and a peripheral region while reducing the size of a welding torch part. However, the welding processing apparatus described in Patent Document 1 is limited to visualizing the state of the welding arc and the surrounding area, and describes a technique for improving the soundness of the arc welded portion using an image obtained by visualization. Not.

  In addition, the welding processing apparatus described in Patent Document 1 optimizes the first optical path disposed in the vicinity of the wire (welding electrode) for imaging of light incident from the welding arc light. For example, a problem remains in photographing a molten pool formed under a wire. In particular, it is difficult to photograph the molten pool and the welding member on the opposite side of the first optical path from the arc light because the adjustment unit is optimized for photographing the welding arc light having a strong intensity. In the visualization result lacking a part of the molten pool and the welded member, there is a possibility that the determination accuracy is lowered in the above-described soundness determination.

  Furthermore, since the welding processing apparatus described in Patent Document 1 captures the vicinity of the wire and the back of the welding torch, the metal member before welding in front of the welding direction can be captured even though the arc light and the state of the molten pool can be captured. There is still a problem in shooting the state. In particular, the state of a gap (hereinafter referred to as a gap) between metal members joined by welding has a great influence on the soundness of an arc welded portion, but a welding processing apparatus described in Patent Document 1 in which an optical path is arranged on the side surface of a welding torch. Therefore, it is difficult to photograph a gap state taking a long and narrow shape along the welding progress direction. In the visualization result lacking the state of the metal member before welding in front of the welding progress direction, there is a possibility that the determination accuracy is lowered in the above-described soundness determination.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a diagnostic device that reduces the loss of the field of view due to arc light and reliably diagnoses the welding state of a member to be welded without increasing the size of the welding torch.

  In order to solve the above problems, a diagnostic apparatus according to the present invention includes a front light receiving unit that acquires light from the front of a moving welding torch, a rear light receiving unit that acquires light from the rear of the welding torch, and An image generation unit that generates an image from the light acquired by the front light receiving unit, the light acquired by the rear light reception unit, and a diagnosis unit that diagnoses the welding state of the welding target member from the image generated by the image generation unit.

  According to the present invention, it is possible to provide a diagnostic device that can reduce the loss of a field of view due to arc light and reliably diagnose the welding state of a member to be welded without increasing the size of the welding torch.

  Problems, configurations, and effects other than those described above will become apparent from the following description of embodiments.

1 is a configuration diagram of a diagnostic apparatus according to a first embodiment. Schematic which shows the effect | action of the light-receiving part based on Example 1, an optical circuit, and an optical circuit combiner. Schematic diagram showing an example of an image formed on the image detection unit An example of an image of arc welding taken on the spot with a photographing device 1 is a configuration diagram illustrating configurations of an image diagnostic apparatus, a reference data storage apparatus, and a diagnostic result storage apparatus in Embodiment 1. FIG. FIG. 6 is a process diagram illustrating operations of the imaging device, the image diagnostic device, the reference data storage device, and the diagnostic result storage device in the first embodiment. The figure which shows the structure of the data preserve | saved at the reference data preservation | save apparatus in Example 1. The graph which shows the example of the soundness parameter | index output to a control part from a front image diagnostic part in Example 1. The graph which shows the example of the soundness parameter | index output to a control part from a back image diagnostic part in Example 1. The block diagram which shows the structure of the welding spot image diagnostic apparatus of the arc welding part which concerns on Example 2 with a welding part. The block diagram which shows the structure of the welding spot image diagnostic apparatus of the arc welding part which concerns on Example 3 with a welding part

  Hereinafter, examples will be described with reference to the drawings.

  An embodiment of the present invention will be described below.

  FIG. 1 is a configuration diagram illustrating a configuration of the diagnostic apparatus according to the first embodiment together with a welded portion.

  First, arc welding will be described. A gap 102 exists between the metal member 101a and the metal member 101b before welding. In arc welding, an arc 203 is generated from the electrode 202 by applying a voltage between the electrode 202 provided at one end of the welding torch 201 and the metal member 101a or the metal member 101b. The metal member 101 a and the metal member 101 b are heated by the arc 203, and a part of the metal member 101 a and the metal member 101 b is melted to form a molten pool 103. When the temperature of the molten pool 103 decreases due to the movement of the welding torch 201 in the welding direction 204 or a voltage decrease, the molten pool 103 is solidified to form the weld bead 104.

  The diagnostic device according to the first embodiment includes a front light receiving unit 310 in front of the welding torch 201 along the welding direction 204 and a rear light receiving unit 320 in the rear of the welding torch 201 along the welding direction 204. Each of the front light receiving unit 310 and the rear light receiving unit 320 includes one or a plurality of optical lenses and a mirror. The front light receiving unit 310 includes light emitted from the molten pool 103 on the front side along the welding direction 204 from the electrode 202, light emitted from the metal member 101a and the metal member 101b before welding, and a gap. The light emitted from a part of 102 is received simultaneously. Further, the rear light receiving unit 320 receives light emitted from the molten pool 103 on the rear side of the molten pool 103 along the welding direction 204 from the electrode 202 and light emitted from the weld bead 104. The above mirror can be replaced with a prism, an optical fiber, or other optical element that changes the traveling direction of light.

  Light received by the front light receiving unit 310 and the rear light receiving unit 320 is guided to the front / rear optical circuit combiner 500 via the front light circuit 410 and the rear light circuit 420, respectively. To achieve this function, the front light circuit 410 and the back light circuit 420 each have one or more optical lenses and mirrors. The mirror here can be replaced by a prism or optical fiber or other optical element that changes the direction of travel of light. Using these optical elements, for example, the front optical circuit 410 and the rear optical circuit 420 can be realized by utilizing the configuration of an optical system known as a borescope, a fiberscope, or a periscope. FIG. 1 shows an example in which the front optical circuit 410 and the rear optical circuit 420 are each composed of two components, a vertical component and a horizontal component.

  The front / rear optical circuit synthesizer 500 includes one or a plurality of optical lenses and mirrors, and outputs two types of light incident from the front light circuit 410 and the rear light circuit 420 to the photographing apparatus 600. The imaging apparatus 600 converts (shoots) the light incident from the front and rear optical circuit combiner 500 into a moving image divided into frames according to the incident time, and outputs the moving image to the image diagnostic apparatus 700.

  Here, the imaging apparatus 600 of the present invention images both the light emitted from the molten pool 103 at a temperature equal to or higher than the metal melting point and the light emitted from the relatively low temperature gap 102 and the weld bead 104. An imaging apparatus 600 that can capture visible light is used. Specifically, the visible light is light having a wavelength of 600 to 1400 nm. In addition, in order to prevent problems caused by shooting or transmission of moving images or other processing speed differences, a buffer memory that temporarily holds a part or all of the moving images may be provided in the shooting apparatus 600.

  Note that the imaging apparatus 600 is also referred to as an image generation unit, and includes images of the surfaces of the metal members 101a and 101b before welding, an image of the gap 102 before welding, and images of the surfaces of the metal members 101a and 101b after welding. Is used for diagnosis of the welding state.

  The image diagnostic apparatus 700 analyzes a moving image input from the imaging apparatus 600 based on the reference data stored in the reference data storage apparatus 710, and determines one or a plurality of soundness indices indicating the soundness of the arc welded portion. The data is output to the user and the diagnostic result storage device 720. The configurations and operations of the image diagnosis apparatus 700, the reference data storage apparatus 710, and the diagnosis result storage apparatus 720 will be described later with reference to FIGS.

  FIG. 2 is a schematic diagram illustrating the operation of the front light receiving unit 310, the rear light receiving unit 320, the front optical circuit 410, the rear optical circuit 420, and the front / rear optical circuit combiner 500 according to the first embodiment of the present invention. In the molten pool 103, the molten pool 103 on the front side from the electrode 202 along the welding direction 204 and the light R <b> 1 emitted from the gap 102 enter the front light receiving unit 310. The light R <b> 1 is converged by the lens 311 in the front light receiving unit 310 and forms an image on the mirror 312. Although only one lens 311 is shown in FIG. 2 for simplicity, it is a well-known technique to appropriately arrange a plurality of lenses. The light R1 changes its direction by the mirror 312 and enters the front optical circuit 410, and further changes the direction by the mirror 411 and enters the front and rear optical circuit combiner 500. The light R1 is further changed in direction by the mirror 511 and is incident on the photographing apparatus 600. At this time, the light R <b> 1 is converged by the lens 512 and forms an image on the image detection unit 601 in the photographing apparatus 600.

  Light R2 emitted from the molten pool 103 and the weld bead 104 on the rear side along the welding direction 204 from the electrode 202 is incident on the rear light receiving unit 320 and passes through the rear light circuit 420 to generate the front / rear optical circuit combiner 500. And an image is formed on the image detection unit 601 together with R1.

  At this time, in order for the light R 1 and the light R 2 to form a clear image on the image detection unit 601, the optical distance until the light R 1 and the light R 2 are emitted from the molten pool 103 and the like and reach the image detection unit 601 is almost the same. It is desirable that Accordingly, when the physical distance traveled until the light R1 and the light R2 are emitted from the molten pool 103 and the like and reach the image detection unit 601 is different, the front light receiving unit 310 and the rear light receiving unit 320, or the front light circuit 410 and the rear light circuit 420 or the front / rear light circuit synthesizer 500 adjust lenses or prisms or other optical elements at one or more locations. In the apparatus of FIG. 2, adjustment is performed using a lens 512 and a lens 522 arranged in the front / rear optical circuit combiner 500. In FIG. 2, the lens 512 and the lens 522 are shown as one lens for simplicity, but it is a well-known technique to adjust the optical distance by appropriately arranging a plurality of lenses.

  FIG. 3 is a schematic diagram illustrating an example of an image formed on the image detection unit 601. The image I1 includes a front image I10 derived from the light R1 and a rear image I20 derived from the light R2. The front image I10 includes an image I102 of the gap 102 and an image I103f of the front portion of the weld pool 103. On the other hand, the rear image I20 includes an image I103r of the rear portion of the weld pool 103 and an image I104 of the weld bead 104. In addition, both I10 and I20 include I101a and I metal member 101b which are images of the metal member 101a and the metal member 101b. Furthermore, it is desirable for the reason described later that both I10 and I20 include either or both of the image I201 of the welding torch 201 and the image I202 of the electrode 202. In the example of FIG. 3, I10 and I20 include both I201 and I202.

  FIG. 4 shows an example of an image obtained by photographing arc welding on the spot with the photographing apparatus 600. Compared to the schematic diagram of FIG. 3, the image of FIG. 4 does not include the image I202 of the electrode 202. The rest of the configuration is the same as that in FIG.

  In FIG. 4, even in the same molten pool 103 image, I 103 f shown in the front image I 10 is larger than I 103 r shown in the rear image I 20. Although it is not clearer than the image of the weld pool, other objects shown in I10 and I20 are also shown in I10 larger than I20. This difference in size is caused by a so-called difference in zoom magnification. Specifically, the light, the light R1 that forms the front image I10, and the light R2 that forms the rear image I20 are emitted from the molten pool and incident on the image detection unit 601, and a lens, mirror, This is due to the difference in optical elements. Therefore, the size shown in the image can be larger than I20 as shown in FIG. 4, or conversely, I20 can be larger than I10 depending on the specific device configuration.

  The so-called difference in zoom magnification between I10 and I20 does not cause a problem when I10 and I20 are separately used for diagnosis. On the other hand, for example, when the images I10 and I20 are analyzed together, such as when evaluating the maximum dimension along the welding direction 204 of the weld pool 103, there is a problem of correction of the difference in zoom magnification. This problem is solved by the fact that both I10 and I20 described above include either or both of the image I201 of the welding torch 201 and the image I202 of the electrode 202. Since the outline of the welding torch 201 and the electrode 202 is clear as shown in FIG. 4, it is easy to identify the outline from the image and evaluate the size on the image. Accordingly, the diagnostic imaging apparatus 700 evaluates the size of the welding torch image I201 or the electrode image I202 shown in I10 and I20 and is evaluated at I20 with respect to the size S201f of the welding torch image I201 evaluated at I10, for example. The ratio of the size S201r of the welding torch image I201, S201r / S201f, may be multiplied by the size of the front molten pool image I103f. In addition, S201f / S201r is multiplied by the size of the rear molten pool image I103r, or the actual size RS201 of the welding torch 201 is used to multiply RS201 / S201f by the size of the front molten pool image I103f. The image of the welding torch 201 or the electrode 202 whose outline can be easily identified, such as multiplying the size of the image I103r of the rear molten pool by RS201 / S201r, is included in I10 and I20 in common, and the size is used. Any method for correcting the difference in so-called zoom magnification between I10 and I20 may be used.

  FIG. 5 is a configuration diagram illustrating configurations of the image diagnostic apparatus 700, the reference data storage apparatus 710, and the diagnostic result storage apparatus 720 according to the first embodiment. The image diagnostic apparatus 700 analyzes the moving image input from the imaging apparatus 600 based on the reference data stored in the reference data storage apparatus 710 and determines one or a plurality of soundness indices indicating the soundness of the arc welded portion. , A device for outputting to the user and diagnostic result storage device 720. Note that the start and end of the operation of the photographing apparatus 600 are synchronized with the start and end of the operation of the welding apparatus 200 via a trigger signal.

  In order to realize this function, the image diagnosis apparatus 700 includes a control unit 701, a front image diagnosis unit 703, a rear image diagnosis unit 704, a main storage unit 705, an imaging device transmission / reception interface 706, and a data bus 707. Have.

  FIG. 6 is a processing diagram illustrating operations of the imaging apparatus 600, the image diagnostic apparatus 700, the reference data storage apparatus 710, and the diagnostic result storage apparatus 720 according to the first embodiment. Each of 720, 701, and the like at the upper end of FIG. 6 corresponds to the label of FIG. Users indicated with symbols instead of labels. The vertical lines below each label and user in FIG. 6 indicate the passage of time, and the processes are arranged from top to bottom in order of time.

  The operation starts from a diagnosis preparation instruction to the user's control unit 701 shown in the upper left side of FIG. 6 (processing S001). At this time, the user also inputs a welding operation condition to be diagnosed together with the instruction to the control unit 701. The control unit 701 requests the reference data storage device 710 to output reference data corresponding to the input welding conditions (S011). The reference data storage device 710 retrieves reference data corresponding to the welding conditions from the stored reference data, and outputs the reference data to the main storage unit 705 in the image diagnostic apparatus 700 (S101). When the reference data input from the reference data storage device 710 is completed, the reference data is transmitted from the main storage unit 705 to the front image diagnosis unit 703 and the rear image diagnosis unit 704 (S051). The front image diagnosis unit 703 and the rear image diagnosis unit 704 store the reference data in the sub storage unit provided therein, and then output a storage end signal to the control unit 701 (S031, S041). When the storage end signal is input, the control unit 701 reports the end of diagnosis preparation to the user (S012).

  After confirming the report of the control unit 701, the user inputs an instruction to start diagnosis to the control unit 701 (S002). The control unit 701 outputs a diagnosis start instruction to the imaging apparatus transmission / reception interface 706 (S013), and the imaging apparatus transmission / reception interface 706 outputs a diagnosis start instruction to the imaging apparatus 600 (S061). The imaging apparatus 600 is changed from the standby state to the imaging preparation state by the instruction S061. Here, the standby state is a state in which shooting is not started even when a so-called trigger signal is input from the outside, and the shooting preparation state is a state in which shooting is started immediately when a trigger signal is input.

  The welding apparatus 200 inputs a trigger signal to the imaging apparatus 600 simultaneously with starting welding. When the trigger signal is input, the imaging device 600 starts imaging immediately, and sequentially outputs the captured moving images to the imaging device transmission / reception interface 706 in the image diagnostic device 700. The output may be performed for every so-called frame constituting the moving image, or a plurality of frames may be output together. In order for the image capturing apparatus 600 to output a plurality of frames collectively, a buffer memory that temporarily stores data for a plurality of frames is required inside the image capturing apparatus 600, but this is due to a difference between the moving image capturing speed and the transmission speed. Can be prevented.

  The imaging device transmission / reception interface 706 outputs the moving image input from the imaging device 600 to the main storage unit 705 (S062). The main storage unit 705 stores the input moving image (S052). The main storage unit 705 is located between the imaging device transmission / reception interface 706, the front image diagnosis unit 703, and the rear image diagnosis unit 704, and stores moving image data. A problem due to a difference in processing speed between the front image diagnosis unit 703 and the rear image diagnosis unit 704 is prevented.

  The front image diagnosis unit 703 and the rear image diagnosis unit 704 read the moving image stored in the main storage unit 705, and based on the reference data stored in the internal sub storage units in steps S031 and S041, The moving image read from 705 is diagnosed, and one or a plurality of soundness indices indicating the soundness of the arc welded part are output to the control part 701 (S032, S042). A specific example of the moving image diagnosis will be described later with reference to FIGS.

  The control unit 701 outputs the soundness index input from the front image diagnosis unit 703 and the rear image diagnosis unit 704 to the user and the diagnosis result storage device 720 (S014). The diagnostic result storage device 720 stores the input result (S201). The user confirms the output of the control unit 701 (S003).

  The series of processing from the photographing and output by the photographing apparatus described above to the user's soundness index confirmation is repeated until the welding end trigger signal from the welding apparatus 200 is input to the photographing apparatus 600.

  When a welding end trigger signal is input from the welding apparatus 200, the imaging apparatus 600 outputs an imaging end signal to the imaging apparatus transmission / reception interface 706 in the diagnostic imaging apparatus 700 together with a moving image. After this output, the photographing apparatus 600 shifts to the photographing preparation state. For moving images, each component in the image diagnostic apparatus 700 including the imaging apparatus transmission / reception interface 706 performs the series of processes described above. On the other hand, when an imaging end signal is input, the imaging apparatus transmission / reception interface 706 outputs it to the control unit 701. The control unit 701 also outputs the end of photographing together with the soundness index to the user (S01 (M + 3)). When the user instructs the control unit 701 to end diagnosis, the control unit 701 outputs an end instruction to the imaging device transmission / reception interface 706 (S01 (M + 4)), and the imaging device transmission / reception interface 706 outputs an end signal to the imaging device 600. (S06 (M + 2)). When the end signal is input, the photographing apparatus 600 shifts from the photographing preparation state to the standby state.

  Hereinafter, examples of the moving image diagnosis processes S032 to S03 (M + 1) and S042 to S04 (M + 1) in the front image diagnosis unit 703 and the rear image diagnosis unit 704 according to the first embodiment will be described using mathematical expressions. The following processing is an example of applying an image pattern recognition method known as the Recognition Taguchi method (hereinafter, RT method). It goes without saying that other image discriminating methods such as a supporting vector machine can be applied to the present invention.

  Since the moving image is composed of a plurality of still images taken at different times, the diagnosis of the moving image is realized by repeating the diagnosis on each still image (hereinafter simply referred to as an image). When diagnosing the nth image, the target image is composed of a plurality of pixels.

  Hereinafter, a square image in which I pixels are arranged in the horizontal direction and J pixels in the vertical direction will be described as an example. Similar processing can be applied to images having other shapes such as a circle or a bar.

If the luminance of the pixel i in the horizontal direction and the pixel j in the vertical direction in the nth image is expressed as x ⁽ⁿ⁾ _ij , I × J from x ⁽ⁿ⁾ ₁₁ to x ⁽ⁿ⁾ _IJ in the nth image. Data is included.

In the first step of the image diagnosis in the first embodiment, feature quantities Y1 ⁽ⁿ⁾ and Y2 ⁽ⁿ⁾ are calculated from x ⁽ⁿ⁾ ₁₁ to x ⁽ⁿ⁾ _IJ of the nth image by the following mathematical formula.

The first feature amount Y1 ⁽ⁿ⁾ is calculated using (Equation 1).

(Equation 1)
Sxy ⁽ⁿ⁾ = <x ⁽ⁿ⁾ ₁₁ > x ⁽ⁿ⁾ ₁₁ + <x ⁽ⁿ⁾ ₁₂ > x ⁽ⁿ⁾ ₁₂ +… + <x ⁽ⁿ⁾ _IJ > x ⁽ⁿ⁾ _IJ
Y1 ⁽ⁿ⁾ = Sxy ⁽ⁿ⁾ / <Sxx ⁽ⁿ⁾ >
In the RT method, Sxy ⁽ⁿ⁾ , <Sxx ⁽ⁿ⁾ >, and Y1 ⁽ⁿ⁾ may be referred to as a linear form, an effective ordinal number, and a slope, respectively. <X ⁽ⁿ⁾ _ij > and <Sxx ⁽ⁿ⁾ > are data stored as reference data, which will be described later.

The second feature amount Y2 ⁽ⁿ⁾ is calculated using (Equation 2).

(Equation 2)
ST1 ⁽ⁿ⁾ = x ⁽ⁿ⁾ ₁₁ ² + x ⁽ⁿ⁾ ₁₂ ² +… + x ⁽ⁿ⁾ _IJ ²
Se1 ⁽ⁿ⁾ = ST1 ⁽ⁿ⁾ -(Sxy ⁽ⁿ⁾ ) ² / <Sxx ⁽ⁿ⁾ >
Y2 ⁽ⁿ⁾ = √ {Se1 ⁽ⁿ⁾ / (I × J-1)}
In the RT method, the above ST1 ⁽ⁿ⁾ , Se1 ^(n), and Y2 ⁽ⁿ⁾ are the total sum of squares, error sum of squares, and standard signal-to-noise ratio (hereinafter, standard SN ratio ⁾ , respectively. ) May be called.

The second step of the image diagnosis in the first embodiment is an average value of Y1 ⁽ⁿ⁾ and Y2 ⁽ⁿ⁾ of the nth image and Y1 ⁽ⁿ⁾ and Y2 ⁽ⁿ⁾ acquired at the time of welding the non-defective product < The Mahalanobis' distance D ⁽ⁿ⁾ between Y1 ⁽ⁿ⁾ > and <Y2 ⁽ⁿ⁾ > is calculated according to (Equation 3).

(Equation 3)
D ⁽ⁿ⁾ = {<V22 ⁽ⁿ⁾ > [Y1 ⁽ⁿ⁾ -<Y1 ⁽ⁿ⁾ >] ²
-2 <V12 ⁽ⁿ⁾ > [Y1 ⁽ⁿ⁾ -<Y1 ⁽ⁿ⁾ >] [Y2 ⁽ⁿ⁾ -<Y2 ⁽ⁿ⁾ >]
+ <V11 ⁽ⁿ⁾ > [Y2 ⁽ⁿ⁾ -<Y2 ⁽ⁿ⁾ >] ² } / 2
In the RT method, the above <V11 ⁽ⁿ⁾ >, <V12 ⁽ⁿ⁾ >, and <V22 ⁽ⁿ⁾ > may be collectively referred to as a correlation coefficient. <V11 ⁽ⁿ⁾ >, <V12 ⁽ⁿ⁾ >, and <V22 ⁽ⁿ⁾ > are data stored as reference data and will be described later.

In the third step of the image diagnosis in the first embodiment, the soundness index I ⁽ⁿ⁾ of the image is calculated according to (Equation 4) from the Mahalanobis distance D ^{(n) of} the nth image calculated in the second step. Output to the unit 701.

(Equation 4)
I ⁽ⁿ⁾ = D ⁽ⁿ⁾ / <s ⁽ⁿ⁾ >
Here, <s ⁽ⁿ⁾ > is a normalization constant. <s ⁽ⁿ⁾ > is data stored as reference data, which will be described later. The soundness index I ^{(n) obtained} by ⁽ Equation 4) represents that the smaller the value, the more similar to the image during non-defective welding. The definition of the soundness index I ⁽ⁿ⁾ is not limited to (Equation 4). For example, I ⁽ⁿ⁾ can be defined as in ^(Equation 5) using the soundness threshold value T ⁽ⁿ⁾ .

(Equation 5)
I ⁽ⁿ⁾ = T ⁽ⁿ⁾ -D ⁽ⁿ⁾ / <s ⁽ⁿ⁾ >
If the definition of (Equation 5) is used, a failure determination can be made when I ⁽ⁿ⁾ is smaller than 0.

  FIG. 7 is a diagram illustrating the structure of data stored in the reference data storage device 710 in the first embodiment. Each reference data is given an identification number ID and is largely composed of a welding specification part and a reference data part. The weld specification part is DSspec., Which is the design drawing number of the weld part. And MSspec., Which is the procurement specification table number of metal parts. OSpec., Which is the welding work specification number. including. These design documents and specifications are created by the design department, procurement department, manufacturing or production engineering department, etc., but if the geometric shape, material, and welding conditions of the weld are specified, other Reference numbers for materials may be used.

The reference data portion is data calculated from moving images during welding of L (L ≧ 1) non-defective products, and includes nine data files D1 to D9 described below. D1 is a moving image obtained by arithmetically averaging the pixel values of each moving image at the time of non-defective welding with respect to L non-defective moving images, and ^{expressed by} <x ⁽ⁿ⁾ _ij > in the above (Equation 1). is there. D2 is a file in which the effective ordinal number of each moving image during non-defective welding is stored for all n constituting the moving image, and is <Sxx ⁽ⁿ⁾ > in ⁽ Equation 1). <Sxx ⁽ⁿ⁾ > is calculated by (Equation 6).

(Equation 6)
<Sxx ⁽ⁿ⁾ > = <x ⁽ⁿ⁾ ₁₁ ² > + <x ⁽ⁿ⁾ ₁₂ ² > +… + <x ⁽ⁿ⁾ _IJ ² >
D3 is the <Y1 ⁽ⁿ⁾ > and <Y2 ⁽ⁿ⁾ > obtained by arithmetically averaging the feature values Y1 ⁽ⁿ⁾ and Y2 ⁽ⁿ⁾ of the moving images during non-defective welding with respect to the L non-defective movies. Is a file stored over all the constituent images of the moving image, and is used in the calculation of (Equation 3). D4 is <V11 ⁽ⁿ⁾ > obtained by arithmetically averaging the correlation coefficients V11 ⁽ⁿ⁾ , V12 ^(n), and V22 ⁽ⁿ⁾ of each image of moving images during non-defective welding with respect to L good quality movies. A file in which V12 ⁽ⁿ⁾ > and <V22 ⁽ⁿ⁾ > are stored over all constituent images of a moving image, and is used in the calculation of (Equation 3). <V11 ⁽ⁿ⁾ >, <V12 ⁽ⁿ⁾ >, and <V22 ⁽ⁿ⁾ > are calculated using (Expression 7).

(Equation 7)
<Vab ⁽ⁿ⁾ > = {[Ya ₁ ⁽ⁿ⁾ -<Ya ⁽ⁿ⁾ >] [Yb ₁ ⁽ⁿ⁾ -<Yb ⁽ⁿ⁾ >]
+ [Ya ₂ ⁽ⁿ⁾ -<Ya ⁽ⁿ⁾ >] [Yb ₂ ⁽ⁿ⁾ -<Yb ⁽ⁿ⁾ >
+ ...
+ [Ya _L ⁽ⁿ⁾ -<Ya ⁽ⁿ⁾ >] [Yb _L ⁽ⁿ⁾ -<Yb ⁽ⁿ⁾ >]} / (I × J-1)}
For example, if a = 1 and b = 1, (Equation 7) represents the calculation method of <V11 ⁽ⁿ⁾ >. In equation ^{_{(7), Ya 1 (n)}} , Ya 2 (n), ..., Ya L (n) is Ya ⁽ⁿ⁾ in the L good video, <Ya ^(n)> in these average is there.

D4 is a file in which the normalization constant <s ⁽ⁿ⁾ > in each image of the moving image at the time of non-defective welding is stored over all the constituent images of the moving image. Used. <s ⁽ⁿ⁾ > is, for example, an image corresponding to L non-defective moving images of Maharanobis' distance D ⁽ⁿ⁾ obtained from (Equation 3) from each image of moving images during non-defective welding Standard deviation for each can be used. Thereby, even when the Mahalanobis distance of the non-defective moving image changes with the movement of the welding torch, the soundness level determination based on the Mahalanobis distance can be performed. For example, in a non-defective movie, if D ⁽¹⁰⁰⁾ = 10 * D ⁽¹⁰⁰⁾ , even if D ⁽¹⁰⁰⁾ is 10 times larger than D ^{(10) in} the diagnosis target moving image, it is immediately determined that Can be avoided. As the standardization constant <s ⁽ⁿ⁾ >, in addition to the standard deviation described above, the calculated value of the function with the standard deviation as an argument, the median including distribution information of the Mahalanobis distance of moving images during non-defective welding, etc. This function can also be used, and can be appropriately configured depending on the welded part and the welding method. D4 may also include the health threshold value T ⁽ⁿ⁾ of ⁽ Equation 5). The soundness threshold value T ⁽ⁿ⁾ can be determined by comparing D ⁽ⁿ⁾ / <s ⁽ⁿ⁾ > calculated from the non-defective product video and the defective product video.

  The above data stored in the reference data storage device 710 can be updated every time a moving image is acquired when a good product is welded, or every time a certain amount of moving images are acquired. Typically, the diagnosis result storage device 720 is connected to the reference data storage device 710, and the reference data of the reference data storage device 710 is updated using the data stored in the diagnosis result storage device 720 when welding a good product.

FIG. 8 is a graph showing an example of the soundness index I ⁽ⁿ⁾ output from the front image diagnosis unit 703 to the control unit 701. Images (a), (b), and (c) on the graph are images used for diagnosis, and each corresponds to n indicated by a dashed arrow. In FIG. 8, the soundness index defined by (Equation 4) is used, and the smaller I ⁽ⁿ⁾ , the closer to a good weld image. In the example of FIG. 8, I ⁽ⁿ⁾ is high at the initial stage of welding and the possibility of failure is high. Comparing images (a) corresponding to high I ⁽ⁿ⁾ with images (b) and (c) corresponding to low I ⁽ⁿ⁾ , high I ⁽ⁿ⁾ occurred because the gap I102 was wider than that of the non-defective product. I understand. In such a case, since the prescribed welding method is appropriately set to weld a narrow gap, if the gap is wider than the prescribed, the probability of occurrence of welding defects increases. The gap width outside the specified value is, for example, a dimensional error in the cutting process of the metal member 101a or the metal member 101b, an arrangement error when the metal member 101a or the metal member 101b is arranged adjacently for welding, or for welding. This may occur due to a plurality of causes in the welding operation and the preceding process, such as an inclination error due to insufficient flatness of the lower surface when the metal member 101a or the metal member 101b is disposed adjacent to the lower surface. As shown in the example of FIG. 8, the soundness index I ⁽ⁿ⁾ output from the front image diagnostic unit 703 of the first embodiment is effective for evaluating the defect occurrence probability of the welded portion.

  Depending on the cause of the gap width variation described above, the gap position may change in addition to the gap width. When the position of the gap changes, the relative position between the welding torch and the gap also changes, so that the defect probability of the welded portion increases. As apparent from FIG. 8, the apparatus of Example 1 can detect the relative position between the gap and the welding torch, and is effective in evaluating the defect occurrence probability of the weld due to the change in the position of the gap. is there.

  Further, as apparent from FIG. 8, since the apparatus of Example 1 can photograph the surface state of the unwelded portion before welding, as described above, the surface state of the unwelded portion is compared with the non-defective image. It is possible to evaluate the defect occurrence probability of the weld based on the comparison. Here, the surface state can be evaluated from the reflected state (reflectance) of light on the surface taken and changes in surface roughness, and it can be used for processing oil, oxides or other impurities that affect the chemical composition of the weld pool. Includes adhesion status. Further, it includes an uneven state on the surface of the unwelded portion that can be evaluated from the change in shading on the image and affects the generation of cavities between the molten metal and the metal member 101a and the metal member 101b.

FIG. 9 is a graph showing an example of the soundness index I ⁽ⁿ⁾ output from the rear image diagnostic unit 704 to the control unit 701. The symbols of the graph and the image are the same as those in FIG. In the example of FIG. 9, the soundness index I ⁽ⁿ⁾ rises around the images (b) and (e). In these images, many black portions appear in the weld bead image I104 as compared with the non-defective weld image. These black parts are abnormal oxides (slug) generated in the high-temperature metal. When they are caught inside the molten pool and solidify, they become internal defects. Rise. The abnormal slag is, for example, a shortage of shielding gas ejected from the welding torch 201 to shield the molten pool 103 from the atmosphere or poor purity of the shielding gas, or processing oil remaining on the surface of the metal member 101a or the metal member 101b. This may be caused by impurities, or an oxide film on the surface of the metal member 101a or the metal member 101b. The example of FIG. 9 shows that the soundness index I ⁽ⁿ⁾ output from the rear image diagnostic unit 704 of one embodiment of the present invention is effective in evaluating the defect occurrence probability of the welded portion.

The user can know the occurrence of a part with a high probability of defect occurrence during welding by looking at the soundness index I ⁽ⁿ⁾ on the spot of the welding, and stop welding once for repair welding or welding. You can take measures such as repairing the part after completing the procedure.

The user can also know the presence or absence of a portion having a high defect occurrence probability in the welded portion by looking at the soundness index I ⁽ⁿ⁾ recorded in the diagnostic result storage device 720. When there is a part with a high probability of defect occurrence in the weld, the recorded welding speed is used to convert the time from the start of welding to the occurrence of the relevant part into the distance from the welding start point. It is also possible to confirm the position of the place where the occurrence probability is high. By confirming the position where the defect occurrence probability is high, it is possible to take measures such as repairing the relevant part.

  With the above configuration, the first embodiment can reduce the loss of the field of view due to arc light without increasing the size of the welding torch, and can photograph the state of the metal member in the front of the welding direction, and uses the in-situ welding image. An apparatus capable of diagnosing the soundness of an arc welded part is provided.

  FIG. 10: is a block diagram which shows the structure of the welding in-situ image diagnostic apparatus of the arc welding part which concerns on another embodiment of this invention with a welding part. Hereinafter, portions common to the first embodiment will be omitted, and the difference from the first embodiment will be mainly described.

  First, arc welding will be described. The metal member 101a and the metal member 101b before welding are provided with a groove 105 called a welding groove in the welded portion for the purpose of improving welding workability, and the metal member 101a and the metal member are provided in the groove 105. There is a gap 102 with 101b. An arc 203 is generated from the consumable electrode 205 by applying a voltage between the consumable electrode 205 provided at one end of the welding torch 201 and the metal member 101a or the metal member 101b. The metal member 101 a, the metal member 101 b, and the consumable electrode 205 are heated by the arc 203, and a part of the metal member 101 a, the metal member 101 b, and the consumable electrode 205 are melted into a liquid state. By melting a part of the consumable electrode 205 and adding the molten portion to the molten pool 103, the volume of the molten pool 103 and the volume of the weld bead 104 formed by solidification of the molten pool 103 are reduced. The metal member 101a and the metal member 101b are joined by increasing the volume substantially equal to and filling all or part of the groove 105.

  In such arc welding, as the molten metal moves from the consumable electrode 205 to the molten pool 103, spatter, which is high-temperature metal fine particles having a particle diameter of several μm to several mm, jumps out of the molten pool 103. There are many cases. When the spatter that has jumped out reaches the lens 311 of the front light receiving unit 310 or the lens 321 of the rear light receiving unit 320, the spatter continues to adhere to the lens surface during welding, which causes a problem that the quality of a photographed image deteriorates.

  As a countermeasure, for example, a protective glass is disposed between the lens 311 or 321 and the molten pool 103, and the protective glass can be exchanged when the amount of spatter adhesion to the protective glass surface exceeds a reference value. . However, since it is necessary to stop the welding operation in order to replace the protective glass, it is desirable to reduce the amount of spatter that reaches the protective glass per unit time in order to increase the efficiency of the welding operation.

In the second embodiment of the present invention, one or both of the front light receiving part 310 and the rear light receiving part 320 is removed from the molten pool 103 in order to reduce the amount of sputtering that reaches the front light receiving part 310 or the rear light receiving part 320 per unit time. It arrange | positions in the position out of the reach | attainment range of the spatter to disperse, for example, 500 mm or more from the molten pool 103.
The spatter scattering direction varies depending on the shape of the welding torch 201, the angle between the welding torch 201, the metal member 101a, and the metal member 101b. FIG. 10 shows that when the amount of spatter scattered forward in the welding direction 204 is large while the amount of spatter scattered backward is small, the front light receiving unit 310 is disposed away from the molten pool 103 and the rear light receiving unit. Reference numeral 320 denotes an example of being arranged at a location near the molten pool 103.
The light receiving unit disposed far from the molten pool 103, in the example of FIG. 10, the front light receiving unit 310 includes an optical zoom mechanism that is a well-known technique for photographing the molten pool 103 and the vicinity thereof. In the example of FIG. 10, the light receiving unit 310 disposed far away is connected to the second imaging device 610 having a digital zoom mechanism. By using the second imaging device 610 having a digital zoom mechanism, the configuration of the device becomes complicated by using a plurality of imaging devices, but the optical zoom mechanism of the light receiving unit 310 can be downsized. Since the small light receiving unit 310 is lightweight, it is advantageous for fine adjustment of the angle or position using a well-known technique for capturing light emitted from the molten pool 103 whose position is changed by the movement of the welding torch 201.
With the above configuration, an embodiment of the present invention can reduce the loss of the field of view due to arc light without increasing the size of the welding torch and can photograph the state of the metal member ahead in the welding direction. An apparatus capable of diagnosing the soundness of an arc welded portion using an image is provided.

  FIG. 11: is a block diagram which shows the structure of the welding spot image diagnostic apparatus of the arc welding part which concerns on another embodiment of this invention with a welding part. Hereinafter, portions common to the first and second embodiments will be omitted, and differences from the first and second embodiments will be mainly described.

  In the in-situ diagnostic imaging apparatus for arc welding according to the third embodiment, a user or a reference data storage device 710 or a diagnostic result storage device 720 is connected to the diagnostic imaging device 700 via a network 800. A plurality of diagnostic imaging apparatuses 700 ′ are connected to the network 800, and the diagnostic imaging apparatuses 700 ′ are connected to the respective apparatuses for taking welding moving images. The network 800 may be an in-factory network or a network outside the factory. For example, the network 800 is the Internet, and the plurality of diagnostic imaging apparatuses 700 may be arranged in a plurality of factories located in different countries. Compared with the diagnosis result storage device 720 of the first embodiment or the second embodiment, the diagnosis result storage device 720 having such a configuration stores more data in the same period. By updating the reference data of the reference data storage device 710 using the non-defective welding data stored in the diagnostic result storage device 720, it becomes easier to find the reference data of the conditions that match the welding conditions to be diagnosed. Diagnosis accuracy can be improved.

101a ... metal member, 101b ... metal member, 102 ... gap, 103 ... melt pool,
104 ... weld bead, 105 ... groove, 201 ... welding torch, 202 ... electrode,
203 ... arc, 204 ... welding direction, 205 ... consumable electrode, 310 ... front light receiving part,
311 ... Lens, 312 ... Mirror, 320 ... Back light receiving unit, 410 ... Front optical circuit,
420 ... Back light circuit, 500 ... Front / rear light circuit synthesizer, 600 ... Imaging device,
601: Image detection unit, 610: Second imaging device, 700: Image diagnostic device,
710 ... Reference data storage device, 720 ... Diagnostic result storage device, 800 ... Network

Claims (14)

A front light receiving unit for acquiring light from the front of the moving welding torch;
A rear light receiving unit for acquiring light from the rear of the welding torch;
An image generating unit that generates an image from the light acquired by the front light receiving unit and the light acquired by the rear light receiving unit;
A diagnostic apparatus comprising: a diagnostic unit that diagnoses a welding state of a welding target member from an image generated by the image generation unit. The diagnostic device according to claim 1,
The light acquired by the front light receiving unit is light from a member to be welded and a weld pool before welding. The diagnostic device according to claim 1 or 2,
An image generated by the image generation unit is an image of a surface of a member to be welded before welding. The diagnostic device according to claim 1 or 2,
An image generated by the image generation unit is an image of a gap between welding target members before welding. The diagnostic device according to any one of claims 1 to 4, wherein
The light acquired by the rear light receiving unit is light from a welding target member and a molten pool after welding. The diagnostic device according to any one of claims 1 to 5,
An image generated by the image generation unit is an image of a surface of a welding target member after welding. The diagnostic device according to claim 3,
The diagnostic device evaluates a probability of occurrence of a welding defect based on light reflectance, surface roughness, or unevenness on a surface of a member to be welded before welding. The diagnostic device according to claim 4,
The diagnostic device evaluates a probability of occurrence of a welding defect based on a width or position of a gap between welding target members before welding. The diagnostic device according to claim 6,
The diagnostic device evaluates the probability of occurrence of a welding defect based on the amount of oxide in a member to be welded after welding. The diagnostic device according to any one of claims 1 to 9,
The light acquired by the front light receiving unit and the rear light receiving unit is visible light having a wavelength of 600 to 1400 nm or light in a near infrared region. The diagnostic device according to any one of claims 1 to 10,
A diagnostic apparatus comprising: a front / rear optical circuit combiner that combines light acquired by the front light receiving unit and light acquired by the rear light receiving unit and outputs the resultant light to the image generation unit. The diagnostic device according to any one of claims 1 to 11,
The diagnostic apparatus, wherein the front light-receiving unit or the rear light-receiving unit is installed outside the reach of spatter scattered from the molten pool. The diagnostic device according to any one of claims 1 to 12,
The diagnostic unit compares the image generated by the image generation unit with an image indicating a proper welding state stored in advance, and diagnoses the welding state of the welding target member. The diagnostic device according to claim 13,
The diagnostic unit is characterized in that an image indicating an appropriate welding state stored in advance is acquired via a network.

Images