JP7628932B2 - Welding monitoring method and welding monitoring device, and additive manufacturing method and additive manufacturing device - Google Patents

Description

The present invention relates to a welding monitoring method and welding monitoring device, as well as an additive manufacturing method and additive manufacturing device.

In arc welding, various technologies have been proposed to monitor whether arc welding is being performed properly by grasping various welding conditions, such as the state of the arc generation and the state of the molten metal (molten pool) melted near the arc spot. For example, Patent Document 1 discloses a technology for monitoring the welding condition using captured images of the molten pool during welding.

JP 2018-192524 A

In general, it is difficult to directly confirm the occurrence of welding defects during arc welding. For example, ultrasonic testing is available as a means of identifying internal defects in welds, but this method has problems such as the need to polish the weld surface and the inclination of the surface making it difficult to detect echoes originating from defects.
For this reason, a method for monitoring the welding state using a captured image of the molten pool, such as that in Patent Document 1, has been proposed, but most of these methods are based on the premise of welding with a groove. In other words, in the conventional technology for monitoring the welding state, the molten pool is judged for welding defects according to the state of the molten pool while corresponding to the positional relationship between the molten pool and the groove or the bead adjacent to the molten pool, or the bead shape. However, the image of the groove and the adjacent bead is not clearly displayed due to a large brightness difference with the arc light, and the position and shape of the molten pool may be misidentified. As a result, there was a risk that it would be difficult to accurately monitor the welding state.
In addition, in additive manufacturing, in which a multi-layered bead is layered to produce an additively manufactured object, there is no reference object that can identify the position of the groove, as described above. Therefore, the state of the molten pool cannot be accurately grasped, and it is difficult to confirm the occurrence of welding defects from images taken during welding.

The present invention aims to provide a welding defect monitoring method and welding defect monitoring device, as well as an additive manufacturing method and additive manufacturing device, that can check for the occurrence of welding defects from captured images of a welded part without using existing beads or the like around the welded part as a reference.

The present invention comprises the following configurations.
(1) A welding monitoring method for monitoring a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising the steps of:
an image acquiring step of acquiring image information showing a molten pool generated in the welded portion and an arc light melting the filler metal;
a contour extraction step of extracting a contour of at least one of the molten pool and the arc light from the image information;
an index calculation step of determining a shape index corresponding to the extracted distortion of the contour;
a defect determination step of determining the occurrence status of a welding defect according to the shape index;
A welding monitoring method comprising:
(2) An additive manufacturing method for manufacturing a molded object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler metal, the method comprising the steps of:
When forming the weld bead, a welding state of the weld bead is monitored by the welding monitoring method described in (1).
Additive manufacturing methods.
(3) A welding monitoring device that monitors a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising:
an image acquisition unit that acquires image information showing a molten pool generated in the weld and an arc light that melts the filler metal;
a contour extraction unit that extracts a contour of at least one of the molten pool and the arc light from the image information;
an index calculation unit that calculates a shape index according to the extracted distortion of the contour;
a defect determination unit that determines the occurrence status of a welding defect according to the shape index;
A welding monitoring device comprising:
(4) An additive manufacturing apparatus for manufacturing a molded object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler metal, comprising:
When the weld bead is formed, the welding state of the weld bead is monitored by the welding monitoring device described in (3).
Additive manufacturing equipment.

According to the present invention, it is possible to check for the occurrence of welding defects from captured images of the welded area without using existing beads or other objects around the welded area as a reference point.

FIG. 1 is a schematic diagram of an additive manufacturing apparatus. FIG. 2 is a schematic explanatory diagram showing how an imaging unit captures an image of a welded portion where a weld bead is formed. FIG. 3 is a schematic diagram showing an image captured by the imaging section. FIG. 4 is a schematic block diagram of the control unit. FIG. 5 is a functional block diagram of the control unit shown in FIG. FIG. 6 is a flowchart showing the procedure of the first welding monitoring method. FIG. 7A is an explanatory diagram that illustrates a schematic shape of an arc light at the tip of a welding torch. FIG. 7B is a graph showing the distribution of the distance from the center point of the arc light shown in FIG. 7A to the contour of the arc light with respect to the azimuth angle around the center point. FIG. 8A is an explanatory diagram showing a schematic diagram of another shape of the arc light at the tip of the welding torch. FIG. 8B is a graph showing the distribution of the distance from the center point of the arc light shown in FIG. 8A to the contour of the arc light with respect to the azimuth angle around the center point. FIG. 9 is a schematic perspective view showing a process of forming a new weld bead adjacent to an existing weld bead. FIG. 10A is a captured image of a welded portion in which a new weld bead has been formed at a position adjacent to an existing weld bead. FIG. 10B is an explanatory diagram showing a main part of the captured image shown in FIG. 10A. FIG. 11A is a captured image of a welded portion in which a new weld bead has been formed at a position even closer to the existing weld bead than in the case shown in FIG. 10A. FIG. 11B is an explanatory diagram showing a main part of the captured image shown in FIG. 11A. FIG. 12 is a schematic diagram of a bead cross section when a new weld bead is formed at a position relatively distant from the existing weld bead shown in FIG. 10A. FIG. 13 is a schematic diagram of a bead cross section when a new weld bead is formed at a position relatively close to the existing weld bead shown in FIG. 11A. FIG. 14 is a graph showing a schematic diagram of the amount of change in the center point of the arc light over time. FIG. 15 is a flowchart showing the procedure of the second welding monitoring method. FIG. 16 is a schematic diagram showing the state of the arc light and the molten pool in a captured image. FIG. 17A is a captured image of a welded portion in which a new weld bead has been formed at a position adjacent to an existing weld bead. FIG. 17B is an explanatory diagram showing a main part of the captured image shown in FIG. 17A. FIG. 18A is a captured image of a welded portion in which a new weld bead has been formed at a position even closer to the existing weld bead than in the case shown in FIG. 17A. FIG. 18B is an explanatory diagram showing a main part of the captured image shown in FIG. 18A. FIG. 19 is a graph showing a schematic diagram of the change in the distance from the center of the arc light to the contour of the molten pool with respect to the azimuth angle. FIG. 20 is a graph showing a schematic diagram of the amount of change in the distance to the edge of the molten pool over time. FIG. 21 is a flowchart showing the procedure of the third welding monitoring method. FIG. 22 is an explanatory diagram that illustrates a schematic shape of the molten pool in a captured image. FIG. 23 is an explanatory diagram showing (A) to (E) how the contour of the molten pool changes over time. FIG. 24 is an explanatory diagram showing a specific example of extracting a shape index of the contour of the molten pool. FIG. 25 is an explanatory diagram showing an example of calculation of local curvature calculated from the contour of the molten pool in the captured image. FIG. 26A and FIG. 26B are explanatory diagrams showing examples of contour shapes and local curvatures.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
Here, as an example of a welding device for implementing the welding monitoring method according to the present invention, an additive manufacturing device that laminates weld beads formed by melting and solidifying a filler metal and is used to manufacture a shaped object composed of multiple layers of weld beads will be described. However, the type of welding and the configuration of the welding device are not limited to this, and the welding device may be a welding device that is applicable to various types of welding such as fillet welding and butt welding.

<Configuration of additive manufacturing device>
FIG. 1 is a schematic diagram of an additive manufacturing apparatus 100. As shown in FIG.
The additive manufacturing apparatus 100 includes a manufacturing unit 11 that manufactures a model by stacking weld beads B, and a control unit 13 that controls each part of the manufacturing unit 11 .

The modeling unit 11 includes a welding robot 17 having a welding torch 15 at its tip shaft, a robot drive unit 19 that drives the welding robot 17, a filler material supply unit 21 that supplies a filler material (welding wire) M to the welding torch 15, and a welding power supply unit 23 that supplies a welding current and welding voltage to the welding torch 15.

The modeling unit 11 also includes an imaging unit 25 having a camera that captures images of the state of the welded area. The imaging unit 25 is provided on the tip shaft side of the welding torch 15 or the welding robot 17, and captures images of the area near the tip of the welding torch 15.

The welding torch 15 has a shield nozzle (not shown) and is a torch for gas metal arc welding to which shielding gas is supplied from the shield nozzle. The arc welding method may be either a consumable electrode type such as shielded metal arc welding or carbon dioxide gas arc welding, or a non-consumable electrode type such as TIG welding or plasma arc welding, and is appropriately selected depending on the additive manufacturing object to be produced. For example, in the case of a consumable electrode type, a contact tip is placed inside the shield nozzle, and a filler material M to which a melting current is supplied is held by the contact tip. While holding the filler material M, the welding torch 15 generates an arc from the tip of the filler material M in a shielding gas atmosphere.

The welding robot 17 is an articulated robot. A continuously supplied filler material M is supported at the tip of a welding torch 15 attached to the tip shaft of the robot arm. The position and posture of the welding torch 15 can be set arbitrarily in three dimensions within the range of the degrees of freedom of the robot arm by commands from the robot drive unit 19. The welding robot 17 is not limited to an articulated robot, and may be in other forms such as a Cartesian coordinate robot or a parallel link robot.

The filler metal supply unit 21 is equipped with a reel 27 on which the filler metal M is wound. The filler metal M is sent from the filler metal supply unit 21 to a payout mechanism (not shown) attached to a robot arm or the like, and is fed to the welding torch 15 while being fed forward and backward by the payout mechanism as necessary.

Any commercially available welding wire can be used as the filler metal M. For example, welding wires specified in MAG and MIG welding solid wires for mild steel, high tensile steel and low temperature steel (JIS Z 3312), arc welding flux-cored wires for mild steel, high tensile steel and low temperature steel (JIS Z 3313), etc. can be used. Furthermore, filler metals M such as aluminum, aluminum alloys, nickel, nickel-based alloys, etc. can be used depending on the required characteristics.

The robot drive unit 19 drives the welding robot 17 to move the welding torch 15, and melts the continuously supplied filler metal M with an arc generated by the welding current and welding voltage from the welding power supply unit 23.

A modeling program based on the trajectory plan of the object to be created is sent from the control unit 13 to the robot driving unit 19. The modeling program is composed of a large number of command codes and is created based on an appropriate algorithm according to various conditions such as the shape data (CAD data, etc.) of the layered object, the material, and the amount of heat input.

More specifically, the shape model of the object corresponding to the input shape data is divided into layers for each predetermined weld bead height, and the bead formation procedure, such as the bead formation order and welding conditions, is determined so that each obtained layer is filled with a weld bead. This bead formation procedure is then made into a modeling program. The created modeling program is stored in the control unit 13 (memory unit 45, described later), and is output from the control unit 13 to the robot driving unit 19 when an output request is received from the robot driving unit 19.

The robot driving unit 19 executes the received modeling program to drive each unit, such as the welding robot 17, the filler material supply unit 21, and the welding power supply unit 23, to form the weld bead B based on the trajectory plan. In other words, the robot driving unit 19 drives the welding robot 17 to move the welding torch 15 along the trajectory (bead formation trajectory) of the welding torch 15 set in the trajectory plan. At the same time, the filler material supply unit 21 and the welding power supply unit 23 are driven according to the set welding conditions to melt and solidify the filler material M at the tip of the welding torch 15 by an arc. As a result, the weld bead B is formed on the base plate 29 along the trajectory of the welding torch 15. The weld bead B forms a weld bead layer by a group of adjacent weld beads, and the next weld bead layer is layered on top of this weld bead layer to form a desired three-dimensional object.

FIG. 2 is a schematic explanatory diagram showing how the imaging unit 25 images the welded portion 31 where the weld bead B is formed.
The imaging unit 25, while moving along the welding direction WD together with the welding torch 15, captures images of the arc light 33 of the arc generated at the welded portion 31 and the molten pool 35 made of molten metal formed by melting the filler material M by the arc. The imaging unit 25 is composed of a camera equipped with a camera body having a CCD or CMOS type imaging element and a lens using an appropriate filter such as an ND filter or a narrow band filter, and is capable of removing noise by filtering and acquiring images with a wide dynamic range from low brightness to high brightness. The imaging unit 25 may be equipped with multiple cameras and generate a captured image by combining imaging information from the multiple cameras.

FIG. 3 is a schematic diagram showing an image captured by the imaging section 25. As shown in FIG.
The image 37 captured by the imaging unit 25 shows the filler metal M protruding from the welding torch 15, the arc light 33 of the arc generated at the tip 39 of the filler metal M, and the molten pool 35 formed below the filler metal M.

In addition to detecting visible light, the imaging unit 25 may be an IR (infrared) camera (thermography) or may be configured to use an IR camera in combination.

FIG. 4 is a schematic block diagram of the control unit 13. As shown in FIG.
The control unit 13 is a computer device that includes a processor 41 such as a CPU or an MPU, a memory 43 such as a ROM or a RAM, a storage unit 45 such as an SSD (Solid State Drive) or a hard disk drive, an input unit 47, an input/output interface 49, an image processing unit 51, a display unit 53, a communication unit 55, etc.

The memory unit 45 stores the above-mentioned molding program and various reference values for determining the occurrence of welding defects, the details of which will be described later.

The input unit 47 is an input device such as a keyboard, mouse, or input operation panel, through which information is input by the operator as necessary.

The input/output interface 49 is connected to the imaging unit 25, filler metal supply unit 21, robot drive unit 19, and welding power supply unit 23, and transmits information to each unit in response to commands from the processor 41.

The image processing unit 51 processes the captured image 37 captured by the imaging unit 25 to extract the arc light 33, which will be described in detail later, and the contours of the molten pool 35, etc. The image processing unit 51 is preferably a dedicated processing circuit using a GPU (Graphics Processing Unit) or the like, but may also perform image processing using the processor 41.

The display unit 53 is composed of a display medium such as a liquid crystal display, and is capable of displaying various types of information, such as the captured image from the imaging unit 25, information on the welding status such as the welding pass and welding conditions, and information on the results of the welding defect judgment, which will be described in detail later.

The communication unit 55 communicates information between the control unit 13 and the outside. The control unit 13 may be configured to have the functions of each of the above-mentioned units provided in another computer device, such as a server, located away from the modeling unit 11, and to control the functions through that computer device. In that case, various control signals are input and output from the communication unit 55 to the other computer device in a remote location via a communication means such as a network.

FIG. 5 is a functional block diagram of the control unit 13 shown in FIG.
By using the configuration of each of the above-mentioned sections, the control section 13 realizes the functions of the image acquisition section 61, the contour extraction section 63, the index calculation section 65, and the defect determination section 67. The functions of these sections will be described in detail later.

<Procedure of the first welding monitoring method>
Next, a procedure of a welding monitoring method for monitoring the welding state using image information obtained by capturing an image of a welded portion during arc welding using the additive manufacturing apparatus 100 configured as described above will be described.
6 is a flow chart showing the steps of the first welding monitoring method. Each step will be described below with reference to FIGS.

First, the image acquisition unit 61 (see FIG. 5) captures an image of the welded portion 31 during welding as shown in FIG. 2, and acquires image information of the captured image 37 including the arc light 33 generated at the tip 39 of the filler metal M and the molten pool 35 in which the filler metal M is melted as shown in FIG. 3 (step S11, hereinafter referred to as S11).

Next, the contour extraction unit 63 (see FIG. 5) extracts the portion of the arc light 33 from the captured image 37 and determines the contour of the arc light 33 (S12). The contour of the arc light 33, that is, the coordinates of the contour of the arc light 33 in the captured image 37 (the coordinates of pixels at multiple points on the contour, but may also be an approximation of a line representing the contour) are determined by appropriate image processing by the image processing unit 51.

Then, the index calculation unit 65 (see FIG. 5) calculates a shape index corresponding to the extracted distortion of the contour (S13). The shape index here includes any of the following: the center point Oa of the arc light 33 obtained from the contour of the arc light 33 shown in FIG. 3, the distance r from the center point Oa to the contour of the arc light 33, the curvature R of the contour, and the degree of fit when the shape of the arc light 33 is fitted with a model curve (e.g., an ellipse, etc.).

The defect determination unit 67 determines the occurrence status of a welding defect, which indicates the presence or absence of a welding defect or its degree, according to the shape index obtained by the index calculation unit 65 (S14). The procedure for determining this welding defect is described below.

Figure 7A is an explanatory diagram showing a schematic of the shape of the arc light 33 at the tip of the welding torch, and Figure 7B is a graph showing the distribution of the distance r from the center point Oa of the arc light 33 shown in Figure 7A to the contour of the arc light 33 versus the azimuth angle θ around the center point Oa. Here, the azimuth angle θ is set clockwise in Figure 7A, with the upper side in the vertical direction in the captured image being 0° and the lower side being 180°. The center point Oa of the arc light 33 is determined by extracting the area of the arc light 33 by performing image processing such as binarization and edge processing on the captured image 37, and defining the center of gravity of that area (pixel area) as the center point Oa.

The center point Oa is not limited to the center of gravity position and may be defined by other methods. For example, for the coordinate values (X, Y) of the area of the arc light 33, the horizontal maximum value Xmax and minimum value Xmin, and the vertical maximum value Ymax and minimum value Ymin in the captured image are respectively calculated, and the midpoint between the maximum value and the minimum value in each direction (Xmax-Xmin)/2, (Ymax-Ymin)/2), that is, the geometric center point, may be set as the coordinates of the center point Oa.

The distance r from the center point Oa of the arc light 33 to the contour of the arc light 33 shown in FIG. 7B changes periodically according to the azimuth angle θ, indicating that the contour of the arc light 33 is substantially elliptical. In other words, the contour of the arc light 33 is linearly symmetrical about the welding line Lw that passes through the center point Oa of the arc light 33 and indicates the welding direction WD.

Depending on the welding conditions, the shape of the arc light 33 may be distorted from the approximately elliptical shape described above.
FIG. 8A is an explanatory diagram showing a schematic shape of another arc light 33 at the tip of the welding torch, and FIG. 8B is a graph showing the distribution of the distance from the center point Oa of the arc light 33 shown in FIG. 8A to the contour of the arc light 33 versus the azimuth angle θ around the center point Oa.

As shown in FIG. 8A, the contour of the arc light 33 is distorted from an axisymmetric shape centered on the welding line Lw that passes through the center point Oa of the arc light 33 and indicates the welding direction WD. This distortion of the arc light 33 occurs due to the influence of the adjacent weld beads.

FIG. 9 is a schematic perspective view showing a process of forming a new weld bead B adjacent to an existing weld bead B0.
When forming the weld bead B, if there is an existing weld bead B0 adjacent to it, the arc from the welding torch 15 will be attracted to the existing weld bead B0, causing the shape of the arc light 33 to become biased.

The change in the shape of the arc light 33 will be described below using captured images.
Fig. 10A is a captured image of a welded portion when a new weld bead B is formed at a position adjacent to an existing weld bead B0. Fig. 10B is an explanatory diagram showing a main part of the captured image shown in Fig. 10A. Fig. 11A is a captured image of a welded portion when a new weld bead B is formed at a position even closer to the existing weld bead B0 than the case shown in Fig. 10A. Fig. 11B is an explanatory diagram showing a main part of the captured image shown in Fig. 11A.

As shown in Figures 10A and 10B, when a new weld bead B is formed at a position relatively far from the existing weld bead B0, the arc is not attracted to the existing weld bead B0 and maintains the approximately elliptical shape as shown in Figure 7A.
On the other hand, as shown in FIGS. 11A and 11B, when a new weld bead B is formed at a position relatively close to an existing weld bead B0, the arc is attracted to the existing weld bead B0, and the shape of the arc light 33 is distorted.
Depending on the difference in the distance between the existing weld bead B0 and the new weld bead B, the occurrence conditions of welding defects differ.

FIG. 12 is a schematic diagram of a bead cross section when a new weld bead B is formed at a position relatively distant from the existing weld bead B0 shown in FIG. 10A.
In this case, a blowhole (air hole) DF occurs between the existing weld bead B0 and the new weld bead B. It is believed that this type of welding defect occurs because the molten metal does not flow sufficiently between adjacent beads during bead formation and solidifies, resulting in an unwelded portion.

On the other hand, FIG. 13 is a schematic diagram of a bead cross section when a new weld bead B is formed at a position relatively close to the existing weld bead B0 shown in FIG. 11A.
In the case shown in FIG. 13, the existing weld bead B0 and the new weld bead B come close to each other and are fused together, with no welding defects such as blowholes existing between them.

Therefore, as an index for judging whether the molten metal has been sufficiently poured between the weld beads, the shape index representing the shape of the arc light 33 is monitored, and abnormal behavior of the shape index (a state with little distortion) is extracted as a sign of a possible defect. For example, the area of the arc light 33 in the captured image 37 is extracted by performing image processing, and the contour and center point Oa of the area (pixel area) of the arc light 33 are obtained. The shape of the obtained contour is evaluated by obtaining the relationship between the azimuth angle θ and the distance r shown in Figures 7B and 8B. In other words, it is judged that the more the contour of the arc light 33 approaches a line-symmetric shape centered on the welding line Lw indicating the welding direction WD, passing through the center point Oa of the arc light 33, the more likely it is that a welding defect has occurred. For this judgment, appropriate parameters such as a specific contour position, the absolute value of the distance r in the area, and the amount of change with respect to the azimuth angle θ can be used, and each parameter can be combined to make the judgment.

In addition, when the shape of the arc light 33 changes, the welding state becomes transient, and a welding defect may occur. Therefore, the amount of variation over time of the center point Oa of the arc light 33 may be calculated, and if the amount of variation is equal to or greater than a predetermined threshold (reference variation amount), that is, if a disturbance occurs in the arc light 33, it may be determined that there is an indication of a welding defect. In this way, robustness of the determination result can be obtained, and the detection sensitivity of the welding defect can be easily adjusted by the threshold value.

FIG. 14 is a graph showing a schematic variation in the center point Oa of the arc light 33 over time.
When the position of center point Oa of arc light 33 is equal to or greater than a predetermined reference fluctuation amount TH1, it is determined that a welding defect has occurred in the region where the fluctuation amount is equal to or greater than the reference fluctuation amount _TH1 .
This determination is not limited to the center point Oa of the arc light 33, but may be based on the curvature of the contour of the arc light 33, or on the degree of conformity between the shape of the arc light 33 and a model curve (e.g., an ellipse, etc.). By determining a welding defect using a change in the curvature of the contour and the degree of conformity of the fitting, the change in the shape of the arc light 33 can be detected with high sensitivity, and the detection accuracy of the welding defect can be improved.

In general, when detecting abnormalities such as the occurrence of welding defects, there are fewer cases of abnormality compared to normal times. Therefore, it is possible to use normal times as a reference and capture the degree of deviation from that state as a sign of abnormality. For example, when welding is performed based on a welding plan in which multiple welding passes and welding conditions for the welding passes are previously determined, a specific welding pass among the welding passes in the welding plan in which welding was performed properly is set as a reference pass that serves as a criterion for determining welding defects. Then, the shape index of the reference pass is compared with the shape indexes of other welding passes that share the welding conditions of the reference pass, and the presence or absence or degree of a welding defect is determined according to the difference between the shape indexes. The welding conditions referred to here can include various conditions such as welding current, welding voltage, filler metal feed speed, welding speed, torch angle, pitch, and the presence or absence of weaving.

Comparison of the shape indices described above can include, for example, comparing the maximum amount of variation, comparing the frequency of exceeding a threshold, and so on. Furthermore, whether welding was performed properly can be determined by verifying the captured image, or by verifying other measurement data such as the welding current, welding voltage, and bead shape profile. Furthermore, when the trajectory plan is created, predicted values can be prepared in advance by simulating temperature prediction, deformation prediction, etc., and a judgment can be made by comparing with the predicted values.

Especially in additive manufacturing, passes with common welding conditions are often repeated multiple times. Therefore, by comparing the shape index of the arc light between passes with common welding conditions, it becomes easier to detect the degree of deviation from the healthy state of the reference pass, or sudden abnormal phenomena. In addition, by comparing weld beads within a welded workpiece, weld beads that show signs of defect occurrence can be identified without having to prepare a database of defect occurrence conditions in advance.

As described above, the characteristic shape index of the arc light 33 can be extracted from the contour of the arc light 33, and the occurrence of welding defects can be determined from the obtained shape index. Then, by continuously performing the calculation to obtain the shape index at regular time intervals, the transition of the shape index during the lamination of the weld bead can be grasped, and information for traceability in the manufacturing process can be obtained.

Therefore, this welding monitoring method can identify locations where welding defects are likely to exist non-destructively and determine the presence or absence and severity of welding defects across the entire range of the welded object. Furthermore, it can determine the occurrence of welding defects in real time during welding. This can be used to review the process, such as modifying welding conditions, and to ensure the quality of the welded object (model).

<Procedure of the second welding monitoring method>
Next, a procedure for determining the occurrence of welding defects from the shape of the molten pool will be described.
FIG. 15 is a flowchart showing the procedure of the second welding monitoring method.
First, the image acquisition unit 61 (see FIG. 5) images the welded portion 31 during welding as shown in FIG. 2, and acquires image information of the captured image 37 including the arc light 33 of the arc generated at the tip 39 of the filler metal M and the molten pool 35 in which the filler metal M is melted, as shown in FIG. 3 (S21).

Next, the contour extraction unit 63 (see FIG. 5) extracts the portion of the arc light 33 from the acquired captured image 37 to obtain the contour of the arc light 33, and calculates the center point Oa of the arc light 33 from the obtained contour. The contour extraction unit 63 also extracts the area of the molten pool 35 from the acquired captured image 37 to obtain the contour of the molten pool 35 (S22). The contour of the arc light 33, the center point Oa, and the contour of the molten pool 35 are obtained by the image processing by the image processing unit 51 (see FIG. 4) as described above.

Then, the index calculation unit 65 (see FIG. 5) calculates (S23) a shape index corresponding to the extracted distortion of the contour of the molten pool 35. The shape index here includes the distance from the center point Oa of the arc light 33, which is obtained from the contour of the arc light 33 shown in FIG. 3, to the contour of the molten pool 35.

Here, the distance from the center point Oa of the arc light 33 to the contour of the molten pool 35 as a shape index may be the distance between a representative position of the contour of the molten pool 35 (e.g., a position close to the existing weld bead) and the center point Oa of the arc light 33. Also, the distance between each of multiple points extracted as the contour of the molten pool 35 and the center point Oa of the arc light 33 may be calculated. This distance may be calculated for each azimuth angle θ around the center point of the arc light 33, or may be calculated continuously for a certain period of time to also calculate the variation in distance over time.

Next, the defect determination unit 67 (see FIG. 5) determines the occurrence state of a welding defect according to the distance between the center point Oa of the arc light 33 and the contour of the molten pool 35 (S24).
FIG. 16 is a schematic diagram showing the state of the arc light 33 and the molten pool 35 in a captured image 37.
As shown in Fig. 9, when a new weld bead B is formed adjacent to an existing weld bead B0, it is preferable that the contour of the molten pool 35 has a shape that is offset from the weld line Lw as shown in Fig. 11B, rather than being linearly symmetrical about the weld line Lw as the center as shown in Fig. 10B. In that case, as shown in Fig. 13, no unwelded welding defects occur, and high-quality welding can be performed.

Therefore, the index calculation unit 65 calculates the distribution of the distance ra from the center point Oa of the arc light 33 to the contour of the molten pool 35 shown in FIG. 16 with respect to the azimuth angle θ around the center point Oa. Then, the defect determination unit 67 determines that the more the contour of the molten pool 35 approaches a line-symmetric shape centered on the welding line Lw that passes through the center point Oa of the arc light 33 and indicates the welding direction WD, the more likely it is that a welding defect has occurred.

Figure 17A is a captured image of a welded portion when a new weld bead B is formed at a position adjacent to an existing weld bead B0. Figure 17B is an explanatory diagram showing a main portion of the captured image shown in Figure 17A. Also, Figure 18A is a captured image of a welded portion when a new weld bead B is formed at a position even closer to the existing weld bead B0 than in the case shown in Figure 17A. Figure 18B is an explanatory diagram showing a main portion of the captured image shown in Figure 18A.

As shown in Figures 17A and 17B, when a new weld bead B is formed at a position relatively far from the existing weld bead B0, the contour of the molten pool 35 is not affected by the existing weld bead B0 and maintains a roughly elliptical shape. On the other hand, as shown in Figures 18A and 18B, when a new weld bead B is formed at a position relatively close to the existing weld bead B0, the contour of the molten pool 35 is distorted in the portion close to the existing weld bead B0 because heat is removed due to the remelting of the existing weld bead B0.

The occurrence of welding defects varies depending on the difference in distance between the existing weld bead B0 and the new weld bead B. Therefore, the shape index of the molten pool 35 is monitored as an indicator for determining whether or not the molten metal has been sufficiently poured between the weld beads, and abnormal behavior of the shape index (a state with little distortion) is extracted as a possible indication of defect occurrence.

For example, the closer the contour of the molten pool 35 is to a shape that is linearly symmetrical about the welding line Lw that passes through the center point Oa of the arc light 33 and indicates the welding direction WD, the more likely it is that a welding defect has occurred. This determination can be made using appropriate parameters such as a specific contour position, the absolute value of the distance ra in the region, and the amount of change relative to the azimuth angle θ, or by combining each parameter.

FIG. 19 is a graph showing a schematic diagram of the change in the distance ra from the center point of the arc light 33 to the outline of the molten pool 35 with respect to the azimuth angle θ.
In the distribution of the distance ra illustrated in Fig. 19, when there is no sign of a welding defect, a change in the distance ra (local recess caused by a decrease in the distance ra) is observed in the area close to the adjacent existing weld bead B0, but when there is a sign of a welding defect, the distance ra is generally approximately constant regardless of the azimuth angle θ. Therefore, if the occurrence of a welding defect is judged from the distribution profile of the distance ra, the presence or absence or the degree of the welding defect can be distinguished. In this case, since the distance ra is calculated at multiple points on the contour, even if the local detection position accuracy is insufficient, the judgment can be made without reducing reliability by using the detection results at other points.

In addition, when the shape of the contour of the molten pool 35 changes, the welding state may become transient, and welding defects may occur. When observing the behavior of the contour of the molten pool 35, it has been confirmed that when welding defects occur, the shape of the end of the molten pool 35 on the side of the adjacent existing weld bead B0 is unstable, and irregularities occur repeatedly in the contour of the molten pool 35. In other words, when the shape of the molten pool 35 is unstable, it is estimated that the molten metal does not flow sufficiently to the toe of the adjacent bead, causing blowholes. Therefore, a threshold value can be set for the amount of change in the distance ra over time to extract signs of defect occurrence. In this case, since it is possible to make a judgment from trends over multiple times compared to when a simple distance threshold is used, the judgment result is more robust. In addition, the defect detection sensitivity can be easily adjusted by the threshold value.

FIG. 20 is a graph showing a schematic variation in the distance ra to the edge of the molten pool 35 over time.
When the amount of change in the distance ra is equal to or greater than a predetermined reference amount of change TH2, it is determined that a welding defect has occurred in the area where the amount of change is equal to or greater than the reference amount of change TH2.
This determination is not limited to the distance ra to the contour of the molten pool 35, but may also be based on the curvature of the contour of the molten pool 35, or on the degree of fit when fitting the shape of the molten pool 35 with a model curve (e.g., an ellipse, etc.). By making the determination based on the change in the curvature of the contour and the degree of fit, the change in the shape of the molten pool 35 can be detected with high sensitivity, and the accuracy of detecting welding defects can be improved.

Also, as in the first welding defect monitoring method, a deviation from a normal state can be regarded as a sign of an abnormality. In other words, when welding is performed based on a welding plan in which a plurality of welding passes and welding conditions for the welding passes are previously determined, a predetermined welding pass among the welding passes in the welding plan in which welding was performed properly is set as a reference pass that serves as a criterion for determining welding defects. Then, the shape index of the reference pass is compared with the shape indexes of other welding passes that share the welding conditions of the reference pass, and the presence or absence or degree of a welding defect is determined according to the difference between the shape indexes. The welding conditions referred to here include various conditions such as welding current, welding voltage, filler metal feed rate, welding speed, torch angle, pitch, and the presence or absence of weaving.

The comparison of the shape indices described above may be a comparison of the maximum amount of variation, or a comparison of the frequency of exceeding a threshold value. In this case, too, whether or not welding was performed properly may be determined by verifying the captured image, or by verifying other measurement data such as the welding current, welding voltage, and bead shape profile. Furthermore, when the trajectory plan is created, predicted values may be prepared in advance by simulating temperature prediction, deformation prediction, etc., and a judgment may be made by comparing with the predicted values.

In this way, by comparing the shape index of the molten pool 35 between passes with common welding conditions, it becomes easier to detect the degree of deviation from the healthy state of the reference pass or sudden abnormal phenomena. In addition, by comparing weld beads within a weldment, weld beads that show signs of defect occurrence can be identified without having to prepare a database of defect occurrence conditions in advance.

This welding monitoring method makes it possible to non-destructively identify locations where there is a possibility of unwelded defects, and to determine the presence or absence and extent of welding defects across the entire area of the welded product. Moreover, this can be determined in real time while welding is in progress. This can be used to review the process, such as modifying the building conditions, and to ensure the quality of the built product.

As described above, the characteristic shape index of the molten pool 35 can be extracted from the contour of the molten pool 35, and the occurrence of welding defects can be determined from the obtained shape index. Then, by continuously performing the calculation to obtain the shape index at regular time intervals, the progress of the shape index during the lamination of the weld bead can be grasped, and information for traceability in the manufacturing process can be obtained.

<Procedure of the third welding monitoring method>
Next, a procedure for determining the occurrence of welding defects from the amount of change in the shape index of the molten pool will be described.
FIG. 21 is a flowchart showing the procedure of the third welding monitoring method.
First, the image acquisition unit 61 (see FIG. 5 ) images the welded portion 31 during welding as shown in FIG. 2 , and acquires image information of the captured image 37 including the molten pool 35 in which the filler metal M is molten, as shown in FIG. 3 (S31).

Next, the contour extraction unit 63 (see FIG. 5) extracts the area of the molten pool 35 from the captured image 37 and obtains the contour of the molten pool 35 (S32). The contour of the molten pool 35 is obtained by the image processing performed by the image processing unit 51 (see FIG. 4) as described above.

Then, the index calculation unit 65 (see FIG. 5) calculates (S33) a shape index corresponding to the extracted distortion of the contour of the molten pool 35. The shape index here includes the amount of change over time in any of the following: the curvature of the contour of the molten pool 35, the inclination of the tangent to the contour, the normal direction to the contour, the number of irregularities appearing on the contour, and the size of the irregularities.

The defect determination unit 67 (see FIG. 5) determines the occurrence state of a welding defect (S34) based on the amount of change over time of the shape index obtained by the index calculation unit 65. The procedure for determining the welding defect will be described below.
FIG. 22 is an explanatory diagram that shows a schematic shape of the molten pool 35 in the captured image.
9, when an adjacent existing weld bead B0 exists, the contour of the molten pool 35 is affected by the distance to the adjacent weld bead B0. When an unwelded welding defect occurs, an uneven portion 71 tends to appear on the contour of the molten pool 35 near the bead toe, which is the boundary between the existing weld bead B0 and the new weld bead B. This uneven portion 71 appears and disappears discontinuously with the passage of time.

Figure 23 is an explanatory diagram showing (A) to (E) the change in the contour of the molten pool 35 over time. Figure 23 shows the tip of the molten pool 35 in the welding direction WD while welding is in progress.

When a weld bead is formed, the contour of the tip of the molten pool 35 shrinks from a smooth state as shown in FIG. 23(A) to a direction opposite the welding direction WD on the side of the existing weld bead (not shown) as shown in FIG. 23(B). Then, as shown in FIG. 23(C), it shrinks further in the opposite welding direction, and then as shown in FIG. 23(D), it again changes to a direction protruding in the welding direction WD, and returns to the smooth state as shown in FIG. 23(E). In this way, the contour of the molten pool 35 may repeatedly change between protruding and shrinking over time.

During the period when such a change in the contour of the molten pool 35 occurs, there is a high possibility that molten metal does not flow sufficiently between the adjacent weld beads B0, resulting in an unwelded weld defect as shown in Figure 12. Therefore, the contour of the molten pool 35 is monitored for the occurrence of unevenness 71, and if it does occur, this is extracted as a sign of a welding defect, and the occurrence status of the welding defect is judged.

FIG. 24 is an explanatory diagram showing a specific example of extracting a shape index of the contour of the molten pool 35.
For example, suppose that a depression is formed near point P2, which is part of the molten pool 35. In this case, the local curvature of the contour of the molten pool 35 changes more greatly near point P2 than near points P1 and P3. In this way, the part where the local curvature of the contour changes can be extracted as an indication of the occurrence of a welding defect. This indication is found as a three-dimensional extraction region for the contour of the molten pool 35 that is formed continuously along the welding direction.

The above-mentioned local curvature can be calculated, for example, as follows.
FIG. 25 is an explanatory diagram showing an example of calculation of a local curvature calculated from the contour of the molten pool 35 in a captured image.
First, the captured image is subjected to mask processing such as a first-order differential filter (e.g., a Prewitt operator) to extract edge components, and pixels with high edge strength are designated as edge pixels. In addition, a two-dimensional edge gradient is calculated for the extracted edge pixels. For example, the edge strength E(x, y) of edge pixel P(x, y) is calculated by equation (1), and the edge gradient φ(x, y) is calculated by equation (2). Note that x and y refer to the coordinates of the Cartesian coordinate system on the plane of the captured image.

Here, Δx(x,y) and Δy(x,y) are the first derivatives in the x and y directions due to masking.

A circular local area A(x, y) of radius rb is set with the above-mentioned edge pixel P(x, y) as its center. The standard deviation of the edge gradients at all edge pixels present within this local area is taken as the local curvature. In other words, if the edge gradient at each edge pixel Pi(xi, yi) (i = 0, 1, 2, ...) present within the local area A(x, y) is taken as φi, the local curvature ω(x, y) at the edge pixel P(x, y) can be calculated using equation (3).

where NB is the number of leading edge pixels in local region A, and φA is the average edge gradient of these edge pixels.

FIG. 26A and FIG. 26B are explanatory diagrams showing examples of contour shapes and local curvatures.
As shown in Fig. 26A, when the contour of the molten pool 35 composed of multiple edge pixels is close to a straight line, the edge gradient φA of each edge pixel is approximately constant, so the value of the local curvature ω(x, y) according to formula (3) is small. On the other hand, as shown in Fig. 26B, when the curvature of the contour of the molten pool 35 in the local region is large, the edge gradient φA of each edge pixel varies widely, so the value of the local curvature ω(x, y) according to formula (3) is large.

The local curvature ω(x, y) obtained as described above is a shape index of the contour of the molten pool 35. The shape index is not limited to the amount of change in the local curvature ω(x, y), but may also be the amount of change over time in the slope of the tangent to the contour, the normal direction to the contour, the number of irregularities appearing on the contour, or the size of the irregularities.

In this case, too, the degree of deviation from a normal state can be taken as a sign of an abnormality. When welding is performed based on a welding plan in which multiple welding passes and the welding conditions for the welding passes are set in advance, a specific welding pass in which welding was performed properly is set as a reference pass that serves as a criterion for determining welding defects, as in the case of the first and second welding defect monitoring methods described above. In addition, the shape index of the reference pass is compared with the shape indexes of other welding passes that share the welding conditions of the reference pass, and the presence or degree of a welding defect is determined based on the difference between the shape indexes. The comparison of shape indexes and other effects described above are the same as in the case of the first and second welding defect monitoring methods described above.

As described above, this welding monitoring method extracts the characteristic shape index of the molten pool 35 from the contour of the molten pool 35, and determines the amount of change in the obtained shape index over time, making it possible to determine the occurrence of a welding defect. Then, by continuously performing the calculation to obtain the shape index at regular time intervals, it is possible to grasp the change in the shape index during the build-up of the weld bead, and obtain information that can be used for traceability in the manufacturing process.

It is also possible to identify locations where there may be unwelded defects non-destructively, and determine the presence or absence and extent of welding defects across the entire area of the welded product. Moreover, it can make the determination in real time while welding. This can be used to review the process, such as modifying welding conditions, and to ensure the quality of the welded product (model).

<Extraction of shape indices>
In the procedures of the first to third welding defect monitoring methods described above, shape indices are extracted from captured images by image processing, but it is also possible to extract them using a machine learning model.

For example, supervised learning is performed using multiple training data including captured images including images of the molten pool and arc in arc welding, and status-related information regarding the state of the arc welding, and a machine learning model is constructed in which the image information of the captured images is input and the status-related information is output. The captured images obtained by the camera are then provided as input to the machine learning model, and the arc welding is controlled based on the status-related information output from the machine learning model.

By using such a machine learning model to extract shape indices from captured images, it is possible to capture subtle changes in the image information that would be missed by simple image processing, making it possible to extract shape indices accurately and reliably.

As described above, this welding monitoring method makes it possible to determine the occurrence of welding defects from the shape of the arc light, the shape of the molten pool, or the shapes of both, without having to correspond to the positional relationship between the groove or the molten pool and the adjacent bead, or the bead shape. This allows for high accuracy in determining welding defects, and makes it possible to constantly and stably monitor the welding condition.

As such, the present invention is not limited to the above-described embodiment, and the invention also contemplates the mutual combination of the various configurations of the embodiment, as well as modifications and applications by those skilled in the art based on the description in the specification and well-known technology, and is included in the scope of the protection sought.

As described above, the present specification discloses the following:
(1) A welding monitoring method for monitoring a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising the steps of:
an image acquiring step of acquiring image information showing a molten pool generated in the welded portion and an arc light melting the filler metal;
a contour extraction step of extracting a contour of at least one of the molten pool and the arc light from the image information;
an index calculation step of determining a shape index corresponding to the extracted distortion of the contour;
a defect determination step of determining the occurrence status of a welding defect according to the shape index;
A welding monitoring method comprising:
This welding monitoring method can identify locations where there is a possibility of unwelded defects in a non-destructive manner. It can also determine the occurrence of welding defects over the entire range of the welded object. It also enables real-time judgment during welding. Therefore, it can be used to review the process, such as modifying welding conditions, and to guarantee the quality of the welded object (molded object).

(2) in the contour extraction step, a contour of the arc light is extracted from the image information;
The welding monitoring method described in (1), wherein the shape index calculated in the index calculation step includes any one of a center point of the arc light calculated from the contour of the arc light, a distance from the center point to the contour of the arc light, a curvature of the contour of the arc light, and a goodness of fit when the shape of the arc light is fitted with a model curve.
According to this welding monitoring method, various characteristics of the arc light can be used as indications of the occurrence of a welding defect, thereby determining the occurrence of a welding defect.

(3) In the contour extraction step, a contour of the arc light is extracted from the image information,
The welding monitoring method described in (1), wherein in the defect determination process, it is determined that the more the contour of the arc light approaches an axisymmetric shape centered on a weld line that passes through a center point of the arc light and indicates the welding direction, the more likely it is that the welding defect has occurred.
According to this welding monitoring method, by correlating the distortion of the arc light contour with the welding defect, the welding defect can be easily determined.

(4) In the contour extraction step, a contour of the arc light and a contour of the molten pool are extracted from the image information,
The welding monitoring method according to (1), wherein the shape index calculated in the index calculation process includes a distance from a center point of the arc light to a contour of the molten pool, the distance being calculated from a contour of the arc light.
According to this welding monitoring method, welding defects are judged according to the distance from the center of the arc light to the contour of the molten pool, so that welding defects can be judged according to subtle changes in the shape of the molten pool.

(5) The welding monitoring method according to (4), wherein the defect determination step determines that the welding defect has occurred when the distance is equal to or greater than a predetermined reference distance.
According to this welding monitoring method, the occurrence of a welding defect can be easily determined by comparing the distance from the center of the arc light to the outline of the molten pool with a reference distance. In addition, the accuracy of the determination result can be further improved by fine-tuning the reference distance.

(6) In the index calculation step, the distance is calculated in a plurality of directions different from each other around a center point of the arc light,
The welding monitoring method described in (4), wherein in the defect determination process, it is determined that the welding defect has occurred if, in a distribution profile of the distance in multiple directions around the center point of the arc light, the rate of change of the distance is always smaller than a predetermined reference rate of change.
According to this welding monitoring method, the more constant the distance from the center of the arc light to the contour of the molten pool is around the center of the arc light, the more likely a welding defect is to occur. Therefore, the occurrence of a welding defect can be easily determined by using the fact that the above distance does not exceed a standard change rate as a judgment criterion.

(7) The welding monitoring method described in (4), wherein the defect determination step determines that the likelihood of the occurrence of the welding defect is high as the contour of the molten pool in the welding direction approaches an axisymmetric shape centered on a weld line that passes through a center point of the molten pool and indicates the welding direction.
According to this welding monitoring method, the probability of occurrence of a welding defect can be quantitatively evaluated.

(8) The welding monitoring method according to any one of (4) to (7), wherein the center point of the molten pool is a center of gravity position of the area of the molten pool in the image information.
According to this welding monitoring method, the center point of the molten pool can be found with high accuracy by simple calculation.

(9) In the index calculation step, a fluctuation amount of the shape index over time is calculated,
The welding monitoring method according to any one of (1) to (8), wherein in the defect determination step, it is determined that the welding defect has occurred when the fluctuation amount is equal to or greater than a predetermined reference fluctuation amount.
According to this welding monitoring method, the shape index fluctuates over time due to unstable behavior of the arc light or molten pool, and by determining the amount of this fluctuation, the occurrence of welding defects can be accurately evaluated.

(10) The welding monitoring method according to any one of (2) to (8), wherein the center point of the arc light is a center of gravity position of an area of the arc light in the image information.
According to this welding monitoring method, the center point of the arc light can be found with high accuracy by simple calculation.

(11) In the contour extraction step, a contour of the molten pool is extracted from the image information,
The welding monitoring method according to (1), wherein the shape index calculated in the index calculation step includes an amount of change over time in the contour of the molten pool.
According to this welding monitoring method, since welding defects are likely to occur when unstable behavior occurs in the molten pool, the occurrence of welding defects can be accurately evaluated by determining the amount of change in the contour of the molten pool over time.

(12) The welding monitoring method described in (11), wherein the shape index calculated in the index calculation process is the amount of change over time of any one of the following: the curvature of the contour of the molten pool, the slope of the tangent to the contour, the normal direction to the contour, the number of irregularities appearing on the contour, and the size of the irregularities.
According to this welding monitoring method, various features of the weld pool contour can be used as indications of the occurrence of welding defects to determine welding defects.

(13) The welding monitoring method according to (12), wherein in the defect determination step, it is determined that the welding defect has occurred when the amount of change over time of the shape index is equal to or greater than a predetermined reference amount of change over time.
According to this welding monitoring method, robustness of the judgment result of the occurrence of a welding defect can be obtained, and the accuracy of the judgment result can be further improved by fine-tuning the reference amount of change over time.

(14) When welding is performed based on a welding plan in which a plurality of welding passes and welding conditions for the welding passes are determined in advance,
In the defect determination step,
Among the welding passes of the welding plan, a reference pass that is a criterion for determining the welding defect is set;
A welding monitoring method described in any one of (1) to (13), which compares the shape index in the reference path with shape indexes of other welding paths that have common welding conditions with the reference path, and judges the welding defect based on the difference between the shape indexes.
According to this welding monitoring method, it is possible to identify weld beads that have signs of defect occurrence by comparing the welding passes in the welded workpiece without preparing a database of conditions for occurrence of welding defects in advance. Therefore, even if cases in which welding defects occur are not fully understood, it is possible to determine the presence of welding defects simply by comparing with the normal state.

(15) An additive manufacturing method for manufacturing a shaped object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler material, comprising the steps of:
An additive manufacturing method, comprising: monitoring a welding state of the weld bead by the welding monitoring method according to any one of (1) to (14) when the weld bead is formed.
According to this additive manufacturing method, even in additive manufacturing where there is no reference object to identify the position of a weld groove, etc., the occurrence and extent of welding defects can be accurately evaluated simply by understanding the state of the arc light or molten pool.

(16) A welding monitoring device that monitors a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising:
an image acquisition unit that acquires image information showing a molten pool generated in the weld and an arc light that melts the filler metal;
a contour extraction unit that extracts a contour of at least one of the molten pool and the arc light from the image information;
an index calculation unit that calculates a shape index according to the extracted distortion of the contour;
a defect determination unit that determines the occurrence status of a welding defect according to the shape index;
A welding monitoring device comprising:
This welding monitoring device can identify locations where there is a possibility of unwelded defects in a non-destructive manner. It can also determine the occurrence of welding defects over the entire range of the welded object. It also enables real-time judgment during welding. Therefore, it can be used to review the process, such as modifying welding conditions, and to guarantee the quality of the welded object (molded object).

(17) The contour extraction unit extracts a contour of the arc light from the image information,
The welding monitoring device according to (16), wherein the shape index calculated by the index calculation unit includes any one of a center point of the arc light calculated from a contour of the arc light, a distance from the center point to the contour of the arc light, and a curvature of the contour of the arc light.
According to this welding monitoring device,

(18) The contour extraction unit extracts a contour of the arc light and a contour of the molten pool from the image information,
The welding monitoring device according to claim 16, wherein the shape index calculated by the index calculation unit includes a distance from a center point of the arc light to a contour of the molten pool, the distance being calculated from a contour of the arc light.
This welding monitoring device can determine welding defects by using various characteristics of the arc light as indications of the occurrence of welding defects.

(19) The contour extraction unit extracts a contour of the molten pool from the image information,
The welding monitoring device according to claim 16, wherein the shape index calculated by the index calculation unit is the amount of change over time in the contour of the molten pool.
According to this welding monitoring device, since welding defects are likely to occur when unstable behavior occurs in the molten pool, the occurrence of welding defects can be accurately evaluated by determining the amount of change in the contour of the molten pool over time.

(20) An additive manufacturing apparatus for manufacturing a molded object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler metal, comprising:
When the weld bead is formed, a welding state of the weld bead is monitored by a welding monitoring device according to any one of (16) to (19).
Additive manufacturing equipment.
With this welding monitoring device, even in additive manufacturing where there is no reference object to identify the position of the weld groove, etc., it is possible to accurately evaluate the occurrence and extent of welding defects simply by grasping the state of the arc light or molten pool.

REFERENCE SIGNS LIST 11 Modeling section 13 Control section 15 Welding torch 17 Welding robot 19 Robot driving section 21 Filler metal supply section 23 Welding power supply section 25 Imaging section 27 Reel 29 Base plate 31 Welded section 33 Arc light 35 Molten pool 37 Captured image 39 Tip 41 Processor 43 Memory 45 Storage section 49 Input/output interface 51 Image processing section 53 Display section 55 Communication section 61 Image acquisition section 63 Contour extraction section 65 Index calculation section 67 Defect determination section 71 Concave/convex section 100 Additive manufacturing device B Weld bead M Filler metal

Claims (15)

A welding monitoring method for monitoring a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising:
an image acquiring step of acquiring image information showing a molten pool generated in the welded portion and an arc light melting the filler metal;
a contour extraction step of extracting a contour of at least one of the molten pool and the arc light from the image information;
an index calculation step of determining a shape index corresponding to the extracted distortion of the contour;
a defect determination step of determining the occurrence status of a welding defect according to the shape index;
Equipped with
In the contour extraction step, a contour of the arc light is extracted from the image information,
In the defect determination step, it is determined that the more the contour of the arc light approaches an axisymmetric shape centered on a weld line that passes through a center point of the arc light and indicates a welding direction, the more likely it is that the welding defect has occurred.
Welding monitoring methods. A welding monitoring method for monitoring a welding state using image information obtained by capturing an image of a welded portion during arc welding, comprising:
an image acquiring step of acquiring image information showing a molten pool generated in the welded portion and an arc light melting the filler metal;
a contour extraction step of extracting a contour of at least one of the molten pool and the arc light from the image information;
an index calculation step of determining a shape index corresponding to the extracted distortion of the contour;
a defect determination step of determining the occurrence status of a welding defect according to the shape index;
Equipped with
In the contour extraction step, a contour of the arc light and a contour of the molten pool are extracted from the image information,
the shape index calculated in the index calculation step includes a distance from a center point of the arc light to a contour of the molten pool, the distance being calculated from a contour of the arc light.
Welding monitoring methods. The defect determination step determines that the welding defect has occurred when the distance is equal to or greater than a predetermined reference distance.
The method of claim 2 . The index calculation step calculates the distance in a plurality of different directions around a center point of the arc light,
3. The welding monitoring method according to claim 2, wherein in the defect determination process, it is determined that the welding defect has occurred when a rate of change in the distance in a distribution profile of the distance in multiple directions around a center point of the arc light is always smaller than a predetermined reference rate of change. the defect determination step determines that the likelihood of the occurrence of the welding defect is high as the contour of the molten pool in the welding direction approaches an axisymmetric shape centered on a weld line that passes through a center point of the molten pool and indicates the welding direction;
The method of claim 2 . The center point of the molten pool is a center of gravity of the area of the molten pool in the image information.
The welding monitoring method according to any one of claims 2 to 5 . The index calculation step calculates a change amount of the shape index over time,
The welding monitoring method according to claim 1 , wherein in the defect determination step, it is determined that the welding defect has occurred when the amount of fluctuation is equal to or greater than a predetermined reference amount of fluctuation. In the contour extraction step, a contour of the molten pool is extracted from the image information,
The welding monitoring method according to claim 1 or 2 , wherein the shape index calculated in the index calculation step includes an amount of change over time of a contour of the molten pool. The shape index calculated in the index calculation step is a change over time in any one of the curvature of the contour of the molten pool, the inclination of a tangent line to the contour, the normal direction of the contour, the number of irregularities appearing on the contour, and the size of the irregularities.
The method of monitoring a weld as recited in claim 8 . In the defect determination step, it is determined that the welding defect has occurred when the amount of change over time of the shape index is equal to or greater than a predetermined reference amount of change over time.
The method of monitoring a weld as recited in claim 9 . When welding is performed based on a welding plan in which a plurality of welding passes and welding conditions for the welding passes are determined in advance,
In the defect determination step,
Among the welding passes of the welding plan, a reference pass that is a criterion for determining the welding defect is set;
comparing the shape index in the reference pass with shape indexes of other welding passes that have the same welding conditions as the reference pass, and determining the welding defect based on a difference between the shape indexes;
The welding monitoring method according to any one of claims 1 to 10 . An additive manufacturing method for manufacturing a molded object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler material, comprising:
When the weld bead is formed, a welding state of the weld bead is monitored by the welding monitoring method according to any one of claims 1 to 11 .
Additive manufacturing methods. A welding monitoring device that monitors a welding state using image information obtained by capturing an image of a welded portion during arc welding,
an image acquisition unit that acquires image information showing a molten pool generated in the weld and an arc light that melts the filler metal;
a contour extraction unit that extracts a contour of at least one of the molten pool and the arc light from the image information;
an index calculation unit that calculates a shape index according to the extracted distortion of the contour;
a defect determination unit that determines the occurrence status of a welding defect according to the shape index;
Equipped with
The contour extraction unit extracts a contour of the arc light from the image information,
The defect determination unit determines that the likelihood of the occurrence of the welding defect is high as the contour of the arc light approaches an axisymmetric shape centered on a weld line that passes through a center point of the arc light and indicates a welding direction.
Welding monitoring equipment. A welding monitoring device that monitors a welding state using image information obtained by capturing an image of a welded portion during arc welding,
an image acquisition unit that acquires image information showing a molten pool generated in the weld and an arc light that melts the filler metal;
a contour extraction unit that extracts a contour of at least one of the molten pool and the arc light from the image information;
an index calculation unit that calculates a shape index according to the extracted distortion of the contour;
a defect determination unit that determines the occurrence status of a welding defect according to the shape index;
Equipped with
the contour extraction unit extracts a contour of the arc light and a contour of the molten pool from the image information;
the shape index calculated by the index calculation unit includes a distance from a center point of the arc light to a contour of the molten pool, the distance being calculated from a contour of the arc light.
Welding monitoring equipment. An additive manufacturing apparatus for manufacturing a molded object composed of multiple layers of weld beads by stacking weld beads formed by melting and solidifying a filler material, comprising:
When the weld bead is formed, a welding state of the weld bead is monitored by the welding monitoring device according to claim 13 or 14 .
Additive manufacturing equipment.

Images