Additive manufacturing apparatus, additive manufacturing method, additive manufacturing system, and program

Abstract

The additive manufacturing apparatus according to the present disclosure includes a wire supply unit that supplies a wire (6), an irradiation unit that irradiates the wire (6) with a laser beam (5) to melt the wire (6), and an imaging unit that captures an image including a connection portion where the melted wire (6) is connected to a workpiece to which the wire (6) is added and generates image data. The apparatus further includes a determination unit (183) that determines the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated by the imaging unit.

Description

Technical Field

[0001] The present disclosure relates to an additive manufacturing apparatus and an additive manufacturing method for manufacturing a three-dimensional object.

Background Art

[0002] As one of the technologies for manufacturing three-dimensional objects, the technology of additive manufacturing (AM) is known. Among these, in an additive manufacturing apparatus of the directed energy deposition (DED) method, a wire as a welding material is fed to a workpiece, and a bead is formed by locally melting the tip of the wire with a laser beam, and a three-dimensional object is manufactured by laminating them.

[0003] In such an additive manufacturing apparatus, when the wire melts at a position away from the workpiece, the melt may remain on the wire, and while the melt is not added to the workpiece, a drop phenomenon may occur in which a drop, which is a lump of the melted welding material after melting, remains on the wire. Further, in an additive manufacturing apparatus that melts the wire fed to the workpiece with a laser beam, a stub phenomenon may occur in which the wire before melting collides with the workpiece. When the drop phenomenon or the stub phenomenon occurs, it becomes difficult to continue stable processing. For this reason, it is required to predict the drop phenomenon or the stub phenomenon in advance and control so that these phenomena do not occur.

[0004] Regarding avoidance of the drop phenomenon or the stub phenomenon, various techniques have been proposed. For example, in Patent Document 1, paying attention to the fact that the drop phenomenon or the stub phenomenon occurs when the positional relationship between the tip of the wire and the workpiece during processing is not appropriate, the melting position of the tip of the wire is estimated from the wire feeding speed, beam output, etc., and the distance between the tip of the wire and the three-dimensional object is adjusted to be constant, and a technique for avoiding stubs and drops is disclosed.

Prior Art Documents

Patent Documents

[0005] [Patent Document 1] International Publication No. 2022 / 107196 [Summary of the Invention] [Problems to be Solved by the Invention]

[0006] However, in the above Patent Document 1, changes in the heat storage state of the workpiece, changes in the gas shielding effect due to the influence of the shape of the three-dimensional workpiece, changes in the absorption rate and reflectivity of the laser beam due to oxidation of the workpiece accompanying the change in the gas shielding effect, changes in the size of the molten pool resulting from the uneven shape on the workpiece side, etc. are not considered, and it has not been possible to accurately capture the signs of the stub phenomenon and the drop phenomenon. Here, if the processing state can be accurately grasped and the signs of the stub phenomenon and the drop phenomenon can be captured, it is considered that by correcting various commands related to processing, the stub phenomenon and the drop phenomenon can be prevented.

[0007] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide an additive manufacturing apparatus capable of accurately grasping the processing state. [Means for Solving the Problems]

[0008] In order to achieve the above object, the additive manufacturing apparatus according to the present disclosure includes a wire supply unit that supplies a wire, an irradiation unit that irradiates the wire with a laser beam to melt the wire, an imaging unit that captures an image including a connection portion where the melted wire and the workpiece to which the wire is added are connected and generates image data, and a determination unit that determines the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated by the imaging unit. The connection portion includes a solid-liquid phase molten pool formed by melting the wire, and a first portion that connects the molten pool and the wire, has a pixel value lower than the region indicating the molten pool and higher than the region indicating the wire. The determination unit determines the processing state based on the state changes of the size, shape, and position of the first portion and a change in the positional relationship between the wire and the workpiece and determines that the processing state isa dropping tendency or a stubbing tendency It is configured to determine whether or not it is.

Advantages of the Invention

[0009] According to the present disclosure, an additive manufacturing apparatus capable of accurately grasping the processing state can be provided. is It can be done.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

Figure 22

Figure 23

Figure 24

Figure 25

Figure 26

Figure 27

Figure 28

Figure 29

Embodiments for Carrying Out the Invention

[0011] Hereinafter, the additive manufacturing apparatus according to Embodiment 1 will be described in detail with reference to the drawings. Note that the present invention is not limited by these embodiments.

[0012] Embodiment 1. FIG. 1 is a diagram showing the configuration of an additive manufacturing apparatus 100 according to Embodiment 1. The additive manufacturing apparatus 100 is a machine tool that manufactures a three-dimensional shaped object by adding a melted filler material to a workpiece. The additive manufacturing apparatus 100 melts the filler material by irradiating a beam. In the present embodiment, the beam is a laser beam 5, and the filler material is a metal wire 6.

[0013] The additive manufacturing apparatus 100 forms a bead 11 by moving the processing point 10 while supplying the wire 6 to the processing point 10, which is the irradiation position of the laser beam 5. The irradiation position of the laser beam 5 and the wire 6 are arranged to intersect, and the wire 6 melts when irradiated with the laser beam 5. The processing point 10 is the contact point between the tip of the melted wire 6 and the workpiece. The workpiece is, for example, an object to which the melted wire 6 is added, and is the base material 13 or the bead 11 on the base material 13. During processing, the tip of the wire 6 continues to melt at the processing point 10, but as the processing head 3 moves, the melted wire 6 cools and solidifies to form the bead 11. The additive manufacturing apparatus 100 manufactures a workpiece, that is, a three-dimensional shaped object, by stacking beads 11 on the base material 13. The base material 13 shown in FIG. 1 is a plate material. The base material 13 may be other than a plate material. In Embodiment 1, the X-axis, Y-axis, and Z-axis are three mutually perpendicular axes. The X-axis and Y-axis are horizontal axes. The Z-axis is a vertical axis. In each of the X-axis direction, Y-axis direction, and Z-axis direction, the direction indicated by the arrow may be referred to as the positive direction, and the direction opposite to the arrow may be referred to as the negative direction.

[0014] The laser oscillator 1, which is a beam source, oscillates a laser beam 5. The laser beam 5 output by the laser oscillator 1 propagates through the fiber cable 2, which is an optical transmission path, to the processing head 3. The laser oscillator 1, the fiber cable 2, and the processing head 3 constitute an irradiation unit that irradiates a workpiece with the laser beam 5 that melts the supplied filler material.

[0015] The processing head 3 has a beam nozzle 4 that emits the laser beam 5 toward the workpiece. Inside the processing head 3, a collimating optical system that collimates the laser beam 5 and a condenser lens that focuses the laser beam 5 are provided. The collimating optical system and the condenser lens are means for adjusting the beam diameter. The illustration of the collimating optical system and the condenser lens is omitted. The direction of the center line of the laser beam 5 irradiated onto the workpiece is the Z-axis direction. The additive manufacturing apparatus 100 has an actuator, which is a drive system that moves the processing head 3 in each of the X-axis direction, the Y-axis direction, and the Z-axis direction. The illustration of the actuator is omitted. Note that the laser beam 5 output from the beam nozzle 4 is, for example, a laser beam having a top-hat type cross-sectional intensity distribution.

[0016] The processing head 3 injects a shielding gas from the beam nozzle 4 toward the workpiece. As the shielding gas, for example, argon gas, which is an inert gas, is used. The additive manufacturing apparatus 100 suppresses the oxidation of the bead and cools the formed bead by injecting the shielding gas. The shielding gas is supplied, for example, from a gas cylinder (not shown), which is a shielding gas supply source. Further, the additive manufacturing apparatus 100 is provided with a cooler (not shown). From the cooler, pipes extend toward the laser oscillator 1 and the processing head 3, and the laser oscillator 1 and the processing head 3 are cooled by circulating cooling water through the pipes. The cooler has a function of detecting the water temperature of the circulating water and keeping the temperature of the circulating water constant, for example.

[0017] The additive manufacturing apparatus 100 is attached with a wire spool 7 which is a supply source of the wire 6. The wire 6 is wound around the wire spool 7. The wire feeder 8 is fixed to the processing head 3 via a support bracket 16. The wire feeder 8 is a supply unit that supplies a welding material to the workpiece. The wire feeder 8 is an example of a wire supply unit. The wire feeder 8 feeds out the wire 6 from the wire spool 7 toward the workpiece. Thereby, the wire 6 is fed out from the wire nozzle 9. Also, the wire feeder 8 pulls back the fed-out wire 6 toward the wire spool 7. Since the beam nozzle 4 and the wire feeder 8 are fixed to the processing head 3, the relative positions of the beam nozzle 4 and the wire feeder 8 are in a fixed relationship. Note that the wire spool 7, the wire feeder 8, and the wire nozzle 9 constitute a wire supply unit that supplies a wire.

[0018] The base material 13 is fixed to the rotary stage 14. The rotary stage 14 rotates around the Z axis. The rotary stage 15 changes the inclination of the rotary stage 14, for example, by rotating around the Y axis. The XYZ stage 17 moves the processing head 3 in each direction of the X-axis direction, the Y-axis direction, and the Z-axis direction. The additive manufacturing apparatus 100 changes the posture of the base material 13 by the operations of the rotary stages 14 and 15. The additive manufacturing apparatus 100 moves the irradiation position of the laser beam 5 on the workpiece by changing the posture of the base material 13 and moving the processing head 3. The additive manufacturing apparatus 100 forms a bead 11 by feeding the wire 6 to the molten pool formed at the processing point 10 and melting the wire 6. Note that the rotary stage 15 may change the inclination of the rotary stage 14 by rotating around the X axis.

[0019] The drive controller 22 includes a head drive unit 221 that moves the processing head 3 by driving the XYZ stage 17, a wire feed drive unit 222 that drives the wire feeder 8, and a stage drive unit 223 that drives the rotary stages 14 and 15.

[0020] In addition, the additive manufacturing apparatus 100 includes an internal sensor unit 23 for monitoring the processing status and an external sensor unit 24. The internal sensor unit 23 is provided near the processing head 3, and the observation target area thereof is indirectly restricted by the size of the inner diameter of the beam nozzle 4. The internal sensor unit 23 includes sensors such as a camera, a thermometer, or a shape measuring device. The camera is, for example, a visible light camera, an infrared camera, a high-speed measurement camera, etc. The camera measures the shape of the workpiece, the melting state, the shape of the molten pool, the temperature, etc. The additive manufacturing apparatus 100 can observe the melting state of the workpiece, the melting state of the wire 6, the fumes or spatter generated during processing, the position of the wire 6, the temperature of the workpiece, the temperature of the wire 6, the temperature of the molten pool, etc. by providing a camera in the internal sensor unit 23. The thermometer detects the light radiated from the workpiece. The thermometer is a non-contact type thermometer such as a radiation thermometer or a thermo camera. The shape measuring device is a measuring device that measures the height in the Z-axis direction and the shape in the X-axis and Y-axis directions, and includes a laser displacement meter, an optical coherence tomography (OCT) device that performs optical coherence tomography, etc. The internal sensor unit 23 may have a spectroscope, an acoustic measuring device, etc.

[0021] The internal sensor unit 23 measures the state of the workpiece or the state of the formed bead 11. The internal sensor unit 23 measures the shape of the workpiece, the temperature of the workpiece, the shape of the molten pool, etc., and generates image data 60. The internal sensor unit 23 transmits the generated image data 60 to the arithmetic unit 18. Note that the internal sensor unit 23 may be provided at a position coaxial with the beam nozzle 4 or at a position slightly separated from the coaxial position of the beam nozzle 4. The internal sensor unit 23 may be provided on the side surface of the processing head 3 or at any position near the processing head 3.

[0022] The external sensor unit 24 is provided at an arbitrary position away from the processing head 3, for example, so as to widely observe the periphery of the processing point 10. The external sensor unit 24 is not restricted by the size of the inner diameter of the beam nozzle 4 and can observe a wider range. The external sensor unit 24 includes sensors such as a camera, a thermometer, or a shape measuring device. The external sensor unit 24 may have a spectroscope, an acoustic measuring device, etc. The external sensor unit 24 measures the state of the workpiece or the state of the formed bead 11. The external sensor unit 24 measures the shape of the workpiece, the temperature of the workpiece, the shape of the molten pool, etc., and generates image data 70. The external sensor unit 24 transmits the generated image data 70 to the arithmetic unit 18.

[0023] The internal sensor unit 23 and the external sensor unit 24 are an example of an imaging unit.

[0024] The additive manufacturing apparatus 100 has an arithmetic unit 18 that performs calculations for additive manufacturing and a numerical control (NC) unit 19 that controls the entire additive manufacturing apparatus 100.

[0025] The arithmetic unit 18, for example, acquires predetermined design data, executes various calculations based on the acquired design data, and generates a machining program 27 and machining condition information 28. The arithmetic unit 18 transmits the generated machining program 27, machining condition information 28, etc. to the NC unit 19. Further, the arithmetic unit 18 determines the machining state of the workpiece based on the image data received from the internal sensor unit 23 or the external sensor unit 24. Note that the arithmetic unit 18 is not limited to being provided separately from the NC unit 19 within the additive manufacturing apparatus 100. The arithmetic unit 18 may be provided, for example, inside the NC unit 19.

[0026] The NC device 19 receives the machining program 27 and the machining condition information 28 from the arithmetic unit 18. The machining program 27 specifies a movement path. The movement path is a path for moving the machining point 10 on the workpiece. The NC device 19 analyzes the movement path based on the content described in the machining program 27. The NC device 19 generates various commands according to the movement path and various machining conditions included in the machining condition information 28. The NC device 19 controls the control target by sending the generated various commands to the control target.

[0027] The NC device 19 controls the drive controller 22 by sending the generated various commands to the drive controller 22. Specifically, the NC device 19 generates a position command, which is an interpolation point group for each unit time on the movement path, and controls the XYZ stage 17 by sending the generated position command to the head drive unit 221. Also, the NC device 19 generates, for example, an axis command according to the description of the machining program 27, and controls the rotary stages 14 and 15 by sending the generated axis command to the stage drive unit 223. Further, the NC device 19 generates, for example, a feed command according to the machining program 27 and the machining conditions, and controls the wire feeder 8 by sending the generated feed command to the wire feed drive unit 222.

[0028] The NC device 19 generates a gas supply command according to the machining program 27. The NC device 19 controls the gas flow regulator 21 by sending the gas supply command to the gas flow regulator 21.

[0029] The NC device 19 generates a laser output command according to the machining program 27 and the machining conditions. The NC device 19 controls the laser oscillator 1 by sending the laser output command to the laser oscillator 1.

[0030] Next, the details of the functions of the arithmetic unit 18 according to the first embodiment will be described. The arithmetic unit 18 includes an acquisition unit 181, a creation unit 182, a determination unit 183, and an adjustment unit 184.

[0031] The acquisition unit 181 acquires image data from each sensor unit. For example, the acquisition unit 181 acquires the image data 60 from the internal sensor unit 23. Also, the acquisition unit 181 acquires the image data 70 from the external sensor unit 24. At this time, the acquisition unit 181 may control, for example, the internal sensor unit 23 and / or the external sensor unit 24 to capture an image including a connection portion where the wire 6 melted by the irradiation of the laser beam 5 and the workpiece to which the wire 6 is added are connected, and generate image data. That is, the acquisition unit 181 may have an imaging function for controlling the imaging unit.

[0032] The creation unit 182 creates processing state information capable of determining the connection state between the wire 6 and the workpiece using the image data acquired from each sensor unit. The creation unit 182 performs, for example, predetermined image processing on the image data 60 acquired from the internal sensor unit 23 to create the processing state information. At this time, the processing state information is, for example, luminance distribution data which is image data indicating a luminance distribution subjected to predetermined contrast processing. Also, the processing state information may include, for example, time-series image data and be information capable of recognizing the transition of the processing state. Further, the creation unit 182 creates processing state information capable of determining the connection state between the wire 6 and the workpiece using the image data 70 acquired from the external sensor unit 24.

[0033] The determination unit 183 determines the connection state of the connection portion between the wire 6 and the workpiece using the processing state information created by the creation unit 182. The connection portion is, for example, a region that connects the wire 6 and the workpiece, and is a portion having a state in which the liquid phase and the solid phase are mixed from after the wire 6 is melted until it solidifies as the workpiece. The determination unit 183 determines the processing state of the workpiece, for example, by referring to the processing state information and using information indicating the distribution of pixel values in the region including the connection portion shown by the image data included in the processing state information. The pixel value indicates, for example, the intensity of the radiant light emitted from a red-hot wire or a molten pool. Specifically, the determination unit 183 determines the processing state of the workpiece based on, for example, at least one state change among the size, shape, and position of the connection portion. Note that the determination unit 183 may determine the processing state of the workpiece using information indicating the distribution of pixel values in the region including the connection portion between the wire 6 and the workpiece in the image based on the image data 60 acquired from the internal sensor unit 23 and / or the image based on the image data 70 acquired from the external sensor unit 24.

[0034] The adjustment unit 184 adjusts various processing conditions based on the connection state of the connection part determined by the determination unit 183. The various processing conditions to be adjusted include, for example, the laser output of the laser beam 5 and the feeding speed of the wire 6. Further, the various processing conditions to be adjusted include, for example, information for controlling the position of the processing head 3. The information for controlling the position of the processing head 3 is, for example, information such as "move the position on the Z-axis by +2 mm" and "move the position on the X-axis by -3 mm" in a preset coordinate space. The information for controlling the position of the processing head 3 may include information indicating an absolute position rather than information indicating a relative position. Also, the various processing conditions to be adjusted may include the axis movement speed of the processing head 3. The axis movement speed of the processing head 3 is a processing condition adjusted in combination with at least one of, for example, the laser output of the laser beam 5, the feeding speed of the wire 6, and the information for controlling the position of the processing head 3. The adjustment unit 184 generates the processed condition information 28 after adjustment based on the adjusted various processing conditions. The adjustment unit 184 transmits the processed condition information 28 after adjustment to the NC device 19.

[0035] Next, the basic processing principle of the additive manufacturing apparatus 100 will be described with reference to FIG. 2. FIG. 2 is a side view, a top view, and a cross-sectional view for explaining the basic processing principle of the additive manufacturing apparatus 100. In FIG. 2, a state in which beads are laminated on a base material and a three-dimensional object is being manufactured is shown. According to FIG. 2, a beam nozzle 4 and a wire feeder 8 are fixed to the processing head 3, and these move integrally on the three axes of the X-axis, Y-axis, and Z-axis. Note that in FIG. 2, a part of the configuration is omitted for the sake of explanation.

[0036] Figure 2 shows the state of producing beads under good processing conditions. Hereinafter, the state where production can be carried out under good processing conditions is referred to as a smooth state. In Figure 2, the fed wire 6 is melted by the laser beam 5, and a molten pool 110 is formed. The molten pool 110 is a part where the wire 6 is red-hot and partially melted, and the solid-liquid phase state where the liquid phase and the solid phase are mixed. The molten pool 110 is included in the connection part between the wire 6 and the workpiece. The part 61 shown in the side view and top view of Figure 2 indicates the part where the wire 6 is red-hot by the laser beam 5 but remains in the solid phase state. The part 61 is a part of the wire 6. The part 111 indicates the part where the wire 6 is red-hot and partially melted, and the solid-liquid phase state where the solid phase and the liquid phase are mixed, that is, the part where the molten pool 110 changes back to the solid phase and becomes a bead. The part 111 is a part of the bead 11. The wire 6 shown in the side view and top view of Figure 2 is connected to the workpiece through the molten pool 110. The joining region 200 is a region in an alloy state where a part of the surface of the base material 13 is melted and mixed with the wire 6. The joining region 200 is included in the workpiece.

[0037] Also, in the side view and top view of Figure 2, it can be seen that as the processing head 3 moves axially in the positive direction of the X-axis, that is, to the right side of the processing head 3, the laser irradiation region R1 moves out of the molten pool 110, the temperature gradually decreases, and it becomes the state of the part 111, and further becomes the state of the part 112 which is a completely cooled and solidified bead. The part 112 is a part of the bead 11. In Figure 2, at the red-hot part 61, the molten pool 110, the part 111, and the part 112 which is a completely cooled and solidified bead, the boundary between the solid phase and the liquid phase and the boundary between red-hot and non-red-hot are not actually clearly shown, but are clearly distinguished and shown for the purpose of explanation. In the cross-sectional view of Figure 3, the cross-sectional shape of the part 112 which is a completely cooled and solidified bead is shown. The bead cross-sectional shape varies in width and height depending on the processing conditions and the wettability of the base material, but is approximately semi-circular.

[0038] Next, the drop state, smooth state, and stub state according to Embodiment 1 will be described. FIG. 3 is a diagram for explaining the drop state, smooth state, and stub state according to Embodiment 1. The drop state means that, for example, when the wire melts at a position away from the workpiece, the melt may remain on the wire, and the melt is not added to the workpiece, that is, the connection between the wire 6 and the workpiece is cut, while the drop, which is a lump of the filler material after melting, remains on the wire. If this state continues, even if the lump of the filler material remaining on the wire is remelted and the processing is continued, a part of the lump may not be completely melted and may be added to the workpiece, which may adversely affect the quality of the finished product. The stub state means, for example, a state in which the wire fed to the workpiece collides with the workpiece in an additive manufacturing apparatus that melts the wire by a laser beam.

[0039] The smooth state shown in the center of FIG. 3 is a state in which various processing conditions such as information for controlling the laser output of the laser beam 5, the feeding speed of the wire 6, and the position of the processing head 3 are appropriately adjusted. However, during the progress of the processing, changes in the heat storage state of the workpiece, changes in the gas shielding effect due to the influence of the shape of the three-dimensional workpiece, changes in the absorption rate and reflectivity of the laser light due to the oxidation of the workpiece accompanying the change in the gas shielding effect, changes in the size of the molten pool 110 due to the uneven shape on the workpiece side, etc. occur, and it shifts to the drop state shown on the left side of FIG. 3 or the stub state shown on the right side of FIG. 3. Therefore, for example, when it is desired to avoid the drop state or the stub state, it is necessary to observe changes in the heat storage state of the shaped object, changes in the gas shielding effect due to the influence of the shape of the shaped object, changes in the absorption rate and reflectivity of the laser light due to the oxidation of the workpiece accompanying the change in the gas shielding effect, changes in the size of the molten pool 110 due to the uneven shape on the workpiece side, etc., and adjust various processing conditions so that the heat storage state of the smooth state is maintained.

[0040] Next, the factors that inhibit the smooth state according to Embodiment 1 will be described. FIG. 4 is a diagram for explaining the factors that inhibit the smooth state according to Embodiment 1. FIG. 4 shows a cross-sectional view in the XY plane of the base material 13 and the beads laminated on the base material 13.

[0041] In the upper part of FIG. 4, a cross-sectional view is shown when forming another bead B adjacent to the linearly formed bead A. In the processing of three-dimensional shaped objects, usually, an arbitrary shape is realized by laminating a plurality of beads on top of each other in multiple layers. In the upper part of FIG. 4, a state is shown where bead B is to be laminated adjacent to bead A at a distance d. As shown in the left diagram in the upper part of FIG. 4, bead B should be laminated and shaped in the assumed area surrounded by the dotted line if bead A is not laminated. However, due to the lamination of bead A, as shown in the right diagram in the upper part of FIG. 4, it is shaped outside the assumed area. At this time, the height of bead B is shaped in a region at a position higher than the height of the assumed area due to the presence of bead A. Thus, it can be seen that when laminating beads adjacent to each other, the height of the bead is affected by the shape of the surrounding beads. Here, when shaping adjacent beads, by performing the shaping multiple times in advance at a predetermined interval and considering the optimal processing conditions, processing in a smooth state can be realized.

[0042] In the middle part of FIG. 4, a cross-sectional view is shown to explain that even for beads laminated on the base material 13 having the same plane, the shape of the bead changes due to the change in wettability depending on the surface state of the base material 13, the material used for the base material 13, the temperature of the base material 13, etc. This cross-sectional view shows that depending on the change in the wettability of the base material 13, even if other processing conditions are the same, the shape may be like that of bead C or may be like that of bead D. Also, in the lower part of FIG. 4, when laminated on a horizontal plane like bead E and bead F, it is shown that the shape is the one indicated by the area surrounded by the dotted line, whereas depending on the shape of the base material 13 or the workpiece during processing, the shape is the one indicated by the area surrounded by the solid line.

[0043] As described above, the bead shape takes various forms in response to various conditions from the start to the end of shaping. This can change the appropriate distance between the wire 6 and the workpiece, leading to a transition to the drop state or the stub state, which may result in processing failure. Also, even if the actual drop state or stub state is not reached, if processing continues in a smooth state close to the stub state or a smooth state close to the drop state, it may lead to deterioration of the processing quality.

[0044] Next, an overview of the processing executed by the additive manufacturing apparatus 100 according to Embodiment 1 will be described. FIG. 5 is a diagram for explaining an example of an overview of the control of the processing state by the additive manufacturing apparatus 100 according to Embodiment 1. FIG. 5 shows how the additive manufacturing apparatus 100 according to Embodiment 1 determines and controls the processing state from the start to the end of processing. In FIG. 5, the horizontal axis represents the progress of processing. In FIG. 5, the vertical axis represents the processing state. On the vertical axis, the smooth state indicates that the processing state is normal, and the drop state and the stub state indicate that the processing state is abnormal. According to FIG. 5, it can be seen that the risk of processing failure increases when the processing state changes from the smooth state to the drop state or the stub state. Note that the letters A, B, and C shown in FIG. 5 represent a predetermined state A, state B, and state C, respectively.

[0045] According to FIG. 5, the additive manufacturing apparatus 100 first sets, as a target processing state, the average value of the values indicating the processing state with a predetermined period from the start of processing as an evaluation section. The target processing state is, for example, a processing state in which it is considered that processing can be performed particularly stably among smooth states. The value indicating the processing state is calculated based on, for example, the position information and luminance information included in the image data 60 or the image data 70. The value indicating the processing state may be calculated based on information indicating the distribution of pixel values. Although the processing state gradually changes due to various factors during processing, the additive manufacturing apparatus 100 monitors the processing state and determines, for example, based on the amount of change in the position information and luminance information included in the image data 60 or the image data 70, whether the change is approaching the drop state or the stub state. At this time, as shown in FIG. 5, the additive manufacturing apparatus 100 preset, for the value indicating the processing state, a threshold value corresponding to the drop boundary, which is the boundary between the smooth state and the drop state, and a threshold value corresponding to the stub boundary, which is the boundary between the smooth state and the stub state, and uses them for the determination of the processing state. Note that the additive manufacturing apparatus 100 may change the preset threshold value according to the processing situation after the setting.

[0046] In FIG. 5, when it is determined that the additive manufacturing apparatus 100 has a dropping tendency (state A in FIG. 5), the additive manufacturing apparatus 100 adjusts various processing conditions so as to approach the stub state. The adjusted processing conditions are set in, for example, the NC apparatus 19, and then processing is executed so as to approach the stub state. After further processing has progressed, when the additive manufacturing apparatus 100 determines that it has a stub tendency (state B in FIG. 5), the additive manufacturing apparatus 100 adjusts various processing conditions so as to approach the dropping state. After further processing has progressed, when the additive manufacturing apparatus 100 determines that it has a dropping tendency (state C in FIG. 5), the additive manufacturing apparatus 100 adjusts various processing conditions so as to approach the stub state. At this time, the additive manufacturing apparatus 100 compares state A, state B, or state C with a preset target processing state, and adjusts various processing conditions so as to approach the target processing state. As a result, it is possible to perform control to always maintain the target processing state while avoiding the dropping state and the stub state. Note that the target processing state is not limited to being set to the average value of the values indicating the processing state with a predetermined period from the start time of processing as the evaluation section, and the average value of the values indicating the processing state with a predetermined period from an arbitrary point in the middle of processing as the evaluation section may be set as the target processing state. Further, the additive manufacturing apparatus 100 may set the target processing state by extracting an arbitrary processing state from information on the processing state collected in the past.

[0047] Next, the processing executed by the arithmetic unit 18 according to the first embodiment will be described. FIG. 6 is a diagram showing an example of the processing executed by the arithmetic unit 18 according to the first embodiment. In this embodiment, it is assumed that when the shaping process is started, the internal sensor unit 23 and the external sensor unit 24 start monitoring the processing status. At this time, the internal sensor unit 23 and the external sensor unit 24 are assumed to capture an image of, for example, the connection portion where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and generate image data.

[0048] According to FIG. 6, when the shaping process is started, the acquisition unit 181 included in the arithmetic unit 18 acquires image data (step S1). The acquisition unit 181 acquires, for example, luminance distribution data from the internal sensor unit 23 or the external sensor unit 24.

[0049] The creation unit 182 included in the arithmetic unit 18 creates machining state information capable of determining the connection state between the wire 6 and the workpiece using the image data acquired in step S1 (step S2). For example, the creation unit 182 creates machining state information capable of determining the connection state between the wire 6 and the workpiece using the luminance distribution data acquired from the internal sensor unit 23 or the external sensor unit 24. Further, for example, the creation unit 182 creates machining state information capable of determining the connection state between the wire 6 and the workpiece using the time-series luminance distribution data acquired from the internal sensor unit 23 or the external sensor unit 24.

[0050] Next, the determination unit 183 included in the arithmetic unit 18 determines the machining state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion between the wire 6 and the workpiece among the images based on the image data included in the machining state information created in step S2 (step S3). For example, the determination unit 183 refers to the connection portion between the wire 6 and the workpiece using the time-series luminance distribution data included in the machining state information and determines the machining state of the workpiece. Specifically, for example, the determination unit 183 compares a reference image preset as a reference for determining the machining state of the workpiece with the captured image captured by the imaging unit, and when the similarity of the information indicating the distribution of pixel values in the region including the connection portion exceeds the threshold value, determines that it is the machining state corresponding to the reference image. The reference images include a reference image corresponding to a smooth state, a reference image corresponding to a stub tendency, a reference image corresponding to a drop tendency, and the like. Hereinafter, a method for determining the connection state of the connection portion between the wire 6 and the workpiece will be described.

[0051] First, an explanation will be given on how the smooth state of the present embodiment is determined. When creating a three-dimensional object using the additive manufacturing apparatus 100, depending on the installation position of the sensor unit, a portion may be formed that creates a blind spot where the processing point cannot be seen from the sensor unit on the three-dimensional object, making it difficult to constantly observe the connection state of the connection portion between the wire 6 and the workpiece with a single sensor unit. Also, although it is conceivable to use a plurality of sensor units to address this, the cost increases as the number of units increases. Therefore, in the present embodiment, as the internal sensor unit 23 which is the imaging unit, a head camera capable of observing the processing position coaxially with the laser beam 5 shall be used. The head camera is, for example, a visible light camera. By using the head camera, there is no need to consider the optimal installation position, and it is not subject to restrictions such as the processing position becoming invisible due to the shadow of the object. For this reason, the head camera is suitable for three-dimensional freeform shaping. Although the external sensor unit 24 which is the imaging unit can acquire a wider range of image information compared to the internal sensor unit 23, it is easily affected by the external environment between the external sensor unit 24 and the workpiece, making it difficult to obtain accurate information. For this reason, in the present embodiment, the image data 60 obtained from the internal sensor unit 23 shall be used for determining the connection state of the connection portion between the wire 6 and the workpiece. That is, the image data 70 obtained from the external sensor unit 24 shall not be directly used for determining the connection state of the connection portion between the wire 6 and the workpiece. However, the image data 70 obtained from the external sensor unit 24 may be used subsidiarily for determining the connection state of the connection portion between the wire 6 and the workpiece.

[0052] FIG. 7 is a side view and a top view for explaining the smooth state according to the first embodiment. The side view shown in FIG. 7 shows an image directly observing the connection shape from the side using a predetermined camera. On the other hand, the top view shown in FIG. 7 corresponds to the side view shown in FIG. 7 and shows an image captured from the head camera. According to FIG. 7, when machining is progressing in a smooth state, metal derived from the wire 6 contained in the wire 6 is supplied from the tip of the wire 6 toward the molten pool 110, and the portion 80 is formed. The portion 80 has fluidity and connects the wire 6 and the molten pool 110. The portion 80 has a higher temperature than the portion 61 but a lower temperature than the molten pool 110. Therefore, among the images observed from the head camera and represented by the generated image data, in an image in which the luminance distribution can be recognized, the portion 80 can be identified as a portion having a higher luminance than the portion 61 and a lower luminance than the molten pool 110. The portion 80 is included in the connection portion between the wire 6 and the workpiece. The determination unit 183 determines that the machining state is a smooth state, for example, when it can be recognized that a region such as the portion 80 is formed in the image captured from the head camera.

[0053] FIG. 8 is an image diagram for explaining the smooth state according to the first embodiment. FIG. 8 represents an image in which the luminance distribution of a region including the connection part can be recognized among the images captured by the head camera and represented by the generated image data. As shown in FIG. 8, the higher the luminance of a part, the lighter the color indicating it, and the lower the luminance of a part, the darker the color indicating it. FIG. 8 shows an image represented in, for example, grayscale. Hereinafter, the same applies to the image diagrams captured by the head camera and for explaining various processing states. The molten pool 110 shown in FIG. 8 is formed within the range of the laser irradiation region R1 or in the vicinity of the laser irradiation region R1. It can be seen that the molten pool 110 shown in FIG. 8 has a relatively high luminance and a high temperature. Further, the part 111 shown in FIG. 8 is formed outside the range of the laser irradiation region R1. It can be seen that the part 111 shown in FIG. 8 has a relatively lower luminance compared to the molten pool 110, and after the molten pool 110 is formed on the base material 13, it is out of the laser irradiation region R1 and has cooled down a little and is in a low temperature state. According to the luminance of the part 61 shown in FIG. 8, it can be seen that the temperature is higher than that of the wire 6 and lower than that of the part 80. The determination unit 183 determines that the processing state is a smooth state, for example, when a region as shown in the part 80 is formed with a predetermined size using the image data indicating the luminance distribution shown in FIG. 8. Note that the image diagram for determining the processing state is not limited to grayscale. The image diagram for determining the processing state may be in any format as long as the shape and temperature, etc. of each part of the connection part where the melted wire 6 is connected to the workpiece to which the wire 6 is added and the vicinity of the connection part can be recognized. For example, it may be shown in full color or with set values of black and white binary values.

[0054] Next, an explanation will be given of how the drop state of the present embodiment is determined. FIG. 9 is a diagram for explaining a method for determining the processing state by the additive manufacturing apparatus 100 regarding the drop phenomenon according to the first embodiment.

[0055] In FIG. 9, state 1 indicates a smooth state, which is set as the initial state. At this time, in state 1 shown in FIG. 9, it is assumed that the distance from the boundary L between the tip of wire 6 and the melting pool 110 to the right end of the irradiation range of the laser beam 5 is d1. In state 1 shown in FIG. 9, when the melting of wire 6 becomes excessive, as in state 2, a lump part 1101 is formed in a part of the melting pool 110 as a lump. When the processing is continued from the state of state 2 shown in FIG. 9, it may shift to state 3 or state 4. State 3 shows, for example, a situation where the lump part 1101 at the tip of wire 6 increased from state 2 did not fall to the main body side of the melting pool 110 and did not become a drop state. At this time, in state 3 shown in FIG. 9, the distance from the boundary L between the tip of wire 6 and the melting pool 110 to the right end of the irradiation range of the laser beam 5 is, for example, d2 which is larger than d1. When a certain period of time elapses from state 3, for example, when wire 6 moves in the positive direction of the X-axis with respect to the workpiece, the boundary L between wire 6 and melting pool 110 also moves in the positive direction of the X-axis with respect to the workpiece and returns to state 1.

[0056] State 4 shown in FIG. 9 shows a situation where it further progresses from the drop tendency of state 2 toward the drop state, and after a certain period of time has elapsed, it finally becomes the drop state. From the above, when transitioning from state 1 to state 2, there is a possibility of transitioning to the drop state of state 4, and if this state transition can be recognized, it is considered that the omen of the drop phenomenon can be captured.

[0057] Therefore, the determination unit 183 provided in the arithmetic unit 18 according to the first embodiment recognizes the transition from state 1 to state 2 shown in FIG. 9, for example, using an image captured by the head camera. And when the processing state of the workpiece transitions from state 1 to state 2, the determination unit 183 determines that the processing state has a drop tendency.

[0058] FIG. 10 is an image transition diagram for explaining the method for determining the drop tendency according to Embodiment 1. The diagrams showing State 1 and State 2 in FIG. 10 respectively correspond to the diagrams showing State 1 and State 2 in FIG. 9. FIG. 10 represents a time-series image capable of recognizing the transition of the luminance distribution of the region including the connection part among the images captured by the head camera and represented by the generated image data. According to FIG. 10, it can be seen that the region of part 80 in State 1 expands and the luminance increases, and finally changes to damming part 1101 in a part of molten pool 110 in State 2. Damming part 1101 is located above the main body of molten pool 110, that is, on the front side when viewed from the irradiation part or the head camera. For this reason, when damming part 110 receives the irradiation of laser beam 5 on the front side and partially blocks laser beam 5, the thermal energy reaching the main body of molten pool 110 decreases, and the temperature of the main body of molten pool 110 becomes lower than that of damming part 1101. On the other hand, there is no place for heat to escape from damming part 1101, and the temperature of damming part 1101 is higher than, for example, the temperature of part 80 in State 1 shown in FIG. 10. At this time, in the image shown in FIG. 10, the luminance of the region indicated by damming part 1101 is higher than the luminance of the region indicated by the main body of molten pool 110. That is, damming part 1101 looks brighter than the main body of molten pool 110. When the determination unit 183 can recognize such a state change, it determines that the processing state of the workpiece has a drop tendency.

[0059] Hereinafter, another example of the method for determining the drop tendency using the image captured by the head camera will be described.

[0060] FIG. 11 is an image transition diagram for explaining the method for determining the drop tendency according to Embodiment 1. FIG. 11 shows a time-series image of an image captured by the head camera and capable of recognizing changes in the luminance distribution of the region including the connection part among the images represented by the generated image data. According to FIG. 11, as time elapses from the left figure to the right figure, the region indicating part 80 gradually becomes smaller, and the luminance distribution changes from state 1 showing a smooth state to state 4 where the region of part 80 finally seems to disappear. As a result, the size of part 80 connecting wire 6 and melting pool 110 becomes so small that it cannot be recognized in the image captured by the head camera, and it can be determined that the connected region is small. And if processing is continued without adjusting various processing conditions as it is, it can be recognized that there is a high possibility of reaching the drop state. At this time, determination unit 183 determines that the processing state of the workpiece is in a drop tendency.

[0061] FIG. 12 is an image transition diagram for explaining the method for determining the drop tendency according to Embodiment 1. FIG. 12 shows a time-series image of an image captured by the head camera and capable of recognizing the transition of the luminance distribution of the region including the connection part among the images represented by the generated image data. According to FIG. 12, as time elapses from the left figure to the right figure, the luminance of the region indicating part 80 gradually becomes higher, and the luminance distribution changes from state 1 showing a smooth state to state 4 where it finally reaches the same luminance as melting pool 110 and the region of part 80 seems to disappear. As a result, the luminance of part 80 connecting wire 6 and melting pool 110 becomes so high that it cannot be distinguished from the melting pool in the image captured by the head camera, and it can be determined that the viscosity of part 80 is low. And if processing is continued without adjusting various processing conditions as it is, it can be recognized that there is a high possibility of reaching the drop state. At this time, determination unit 183 determines that the processing state of the workpiece is in a drop tendency.

[0062] FIG. 13 is an image transition diagram for explaining the method for determining the drop tendency according to Embodiment 1. FIG. 13 shows a time-series image capable of recognizing the transition of the luminance distribution in the region including the connection part among the images captured by the head camera and represented by the generated image data. According to FIG. 13, it can be seen that the region of the molten pool 110 in the solid-liquid phase state moves greatly in the positive direction of the Y axis from state 1 to state 2. As this factor, for example, when machining the corner portion of the workpiece, it is considered that gravity or surface tension is applied to the molten pool 110 in the solid-liquid phase state due to the uneven shape formed on the workpiece. At this time, the connection between the part 80 and the molten pool 110 is likely to reach the drop state because some of the connections are cut due to the movement of the molten pool 110. When all of the part 80 is cut, a part of the part 80 will fall into the molten pool 110, so it is necessary to avoid this from the viewpoint of machining quality. At this time, the determination unit 183 determines that the machining state of the workpiece has a drop tendency.

[0063] Next, an explanation will be given of how the stub state of the present embodiment is determined. FIG. 14 is a diagram for explaining a method for determining the machining state by the additive manufacturing apparatus 100 regarding the stub phenomenon according to Embodiment 1.

[0064] In FIG. 14, state 1 shows a smooth state, and this is set as the initial state. At this time, in state 1 shown in FIG. 14, it is assumed that the distance from the boundary L between the tip of wire 6 and the melting pool 110 to the right end of the irradiation range of laser beam 5 is d1. In state 1 shown in FIG. 14, for example, when the feeding speed of wire 6 is faster than the speed at which wire 6 is melted, or when the laser output of laser beam 5 is weak and wire 6 is fed out in a state where it is difficult to melt, it can be seen that the tip of wire 6 approaches the workpiece as in state 2. Specifically, the tip of wire 6 approaches a part 111 of the workpiece that is in a solid state. At this time, in state 2 shown in FIG. 14, the distance from the boundary L between the tip of wire 6 and the melting pool 110 to the right end of the irradiation range of laser beam 5 is, for example, d2 which is smaller than d1. If processing is continued from the state of state 2 shown in FIG. 14, it may transition to state 3 or state 4. State 3 shows, for example, a state where the melting of the tip of wire 6 progresses from state 2 and does not become a stub state. At this time, in state 3 shown in FIG. 14, the distance from the boundary L between the tip of wire 6 and the melting pool 110 to the right end of the irradiation range of laser beam 5 is, for example, d3 which is larger than d2 and smaller than d1. When a certain period of time elapses from state 3, for example, the melting of the tip of wire 6 further progresses and returns to state 1.

[0065] State 4 shown in FIG. 14 shows a state where it progresses further toward the stub state from the stub tendency of state 2, and after a certain period of time has elapsed, it finally becomes a stub state. From the above, when transitioning from state 1 to state 2, there is a possibility of transitioning to the stub state of state 4, and if this state transition can be recognized, it is considered that the omen of the stub phenomenon can be captured.

[0066] Therefore, the determination unit 183 provided in the arithmetic unit 18 according to the first embodiment recognizes the transition from state 1 to state 2 shown in FIG. 14 using, for example, an image captured by a head camera. Then, when the processing state of the workpiece transitions from state 1 to state 2, the determination unit 183 determines that the processing state has a stub tendency.

[0067] FIG. 15 is an image transition diagram for explaining the stub tendency determination method according to Embodiment 1. The diagrams showing State 1 and State 2 in FIG. 15 respectively correspond to the diagrams showing State 1 and State 2 in FIG. 14. FIG. 15 represents a time-series image capable of recognizing the transition of the luminance distribution of the region including the connection part among the images represented by the image data captured from the head camera and generated. According to FIG. 15, from State 1 which is a smooth state, the position of the tip of wire 6 moves to the right side, that is, the positive direction of the X axis, and in State 2, it can be seen that the tip of wire 6 is approaching part 111 which is a part of the workpiece on the XY plane. At this time, the existence of part 80 connecting the tip of wire 6 and the molten pool 110 can hardly be confirmed, and the tip of wire 6 is located near the right end of the molten pool 110. Here, although the position of the tip of wire 6 in the Z-axis direction is not clear only from the image diagram captured by the head camera, the feeding angle of wire 6 in the additive manufacturing apparatus 100 of this embodiment is set to the angle as shown in FIG. 14, and the position of wire 6 is usually arranged such that when the tip of wire 6 approaches the workpiece on the XY plane when continuously fed without being melted, it also approaches the workpiece on the Z axis. From this, when the tip of wire 6 approaches part 111 which is a part of the workpiece on the XY plane in the image diagram captured by the head camera, it can be determined that the tip of wire 6 also approaches part 111 which is a part of the workpiece in the Z-axis direction. And when there is such a change in state, if the machining is continued without adjusting various machining conditions as it is, the possibility of reaching the stub state becomes high. As shown in FIG. 15, the determination unit 183 determines that the machining state of the workpiece has a stub tendency when at least one state change can be recognized among, for example, the size, shape, and position of the connection part including part 80 based on the state change of the luminance distribution of the connection part. Note that the determination unit 183 may use, in addition to the image diagram captured by the head camera which is the internal sensor unit 23, an image diagram based on the image data 70 obtained from an external sensor unit 24 including, for example, a visible light camera to determine the position of the tip of wire 6 in the Z-axis direction.

[0068] Next, another example of a method for determining the stub tendency using an image captured by the head camera will be described.

[0069] FIG. 16 is an image transition diagram for explaining the method for determining the stub tendency according to the first embodiment. FIG. 16 shows a time-series image in which the transition of the luminance distribution of the region including the connection portion can be recognized among the images represented by the image data captured by the head camera and generated. According to FIG. 16, it can be seen that the size, shape, and position of the molten pool 110 change moment by moment from state 1 to state 6 due to the change in the luminance distribution of the molten pool 110 caused by the vibration of the light radiated from the tip of the wire 6, the connection portion, and the workpiece. As this factor, for example, the feeding speed of the wire 6 is faster than the melting speed of the tip of the wire 6 by the irradiation of the laser beam 5. At this time, since the melting of the wire 6 reaches the molten pool 110 in a state where the melting is not sufficient, the temperature of the molten pool 110 is lower than in the smooth state, the viscosity of the molten pool 110 is higher, and at the connection portion between the molten pool 110 and the wire 6, the vibration accompanying the movement of the wire 6 is transmitted. When processing is continued in such a state, for example, the tip of the wire 6 continues to be fed without being completely melted, and there is a high possibility of directly colliding with the portion 111 which is a part of the workpiece. As shown in FIG. 16, when the determination unit 183 can recognize at least one state change among the size, shape, and position of the connection portion including the molten pool 110 based on the state change of the luminance distribution of the connection portion, it determines that the processing state of the workpiece has a stub tendency.

[0070] When it is determined that the adjustment unit 184 provided in the arithmetic unit 18 has an abnormal tendency in the processing state (Yes in step S3), the adjustment unit 184 generates adjusted processing condition information 28 corresponding to the content of the abnormal tendency (step S4). For example, when it is determined in step S3 that there is a drop tendency, the adjustment unit 184 generates processing condition information 28 including various processing conditions after adjustment so as to approach a preset target processing state, that is, so as to approach the stub state. At this time, the various processing conditions after adjustment include, for example, at least one of a processing condition for increasing the feeding speed of the wire 6, a processing condition for decreasing the laser output of the laser beam 5, and a processing condition for changing the position of the processing head 3 in the XYZ space so that the tip of the wire 6 approaches the molten pool 110. All of them are processing conditions that promote the melting of the damming part generated during the drop tendency, and thus it is possible to make an adjustment to approach the stub state from the drop tendency. Among the processing conditions to be adjusted, when the feeding speed of the wire 6 and the laser output of the laser beam 5 are also related to the control of the shaping shape of the three-dimensional shaped object to be produced, it is preferable to preferentially adjust the change of the information for controlling the position of the processing head 3. At this time, the adjustment unit 184 may, for example, adjust the axial movement speed of the processing head 3 in combination with at least one of the laser output of the laser beam 5, the feeding speed of the wire 6, and the information for controlling the position of the processing head 3. Specifically, for example, when it is determined that there is a stub tendency, the adjustment unit 184 adjusts so that the axial movement speed of the processing head 3 decreases. For example, when it is determined that there is a drop tendency, the adjustment unit 184 adjusts so that the axial movement speed of the processing head 3 increases.

[0071] Further, when it is determined in step S3 that there is a stub tendency, for example, the adjustment unit 184 generates processing condition information 28 including various processing conditions after adjustment so as to approach a preset target processing state, that is, so as to approach a drop state. At this time, the various processing conditions after adjustment include, for example, at least one of a processing condition of reducing the feeding speed of the wire 6, a processing condition of increasing the laser output of the laser beam 5, and a processing condition of changing the position of the processing head 3 in the XYZ space so that the tip of the wire 6 moves away from the molten pool 110. All of them are processing conditions for avoiding collisions by changing the state of the tip of the wire 6 approaching the workpiece from the state of approaching to the state of not approaching during the stub tendency, and thus it is possible to make an adjustment to approach from the stub tendency to the drop state.

[0072] When the adjustment unit 184 generates the processed condition information 28 after adjustment in step S4, the adjustment unit 184 transmits the processed condition information 28 after adjustment to, for example, the NC device 19. As a result, the processed condition information 28 after adjustment is set in the NC device 19. Thereafter, when it is determined in step S3 that there is a drop tendency, processing is executed to approach from the drop tendency to the stub state according to the processed condition information 28 after adjustment. When it is determined in step S3 that there is a stub tendency, processing is executed to approach from the stub tendency to the drop state according to the processed condition information 28 after adjustment.

[0073] As described above, according to the adjustment unit 184, it is possible to return to the smooth state before reaching the drop state or the stub state, and it is possible to continuously perform processing in a stable smooth state at all times, thereby improving the quality of the workpiece. In addition, it is possible to automatically perform the pre-processing condition adjustment work and the trial and error of the processing program that were conventionally performed by the operator, thereby reducing the burden on the operator. Note that when the adjustment unit 184 gives priority to avoiding processing failures, information for controlling the position of the processing head 3 and other processing conditions may be adjusted simultaneously. Further, when the adjustment unit 184 gives priority to shape accuracy, information for controlling the position of the processing head 3 may be preferentially adjusted compared to other processing conditions.

[0074] When the adjusted processing condition information is set in step S5, the arithmetic unit 18 determines whether or not the shaping process has ended (step S6).

[0075] When the arithmetic unit 18 determines that the shaping process has not ended (No in step S6), it acquires the image data again in step S1 and executes steps S2 to S6. In this way, by repeating steps S1 to S6 during the shaping process, it becomes possible to always execute the processing in a smooth state, more specifically, in a state close to a preset target processing state.

[0076] When the arithmetic unit 18 determines that the shaping process has ended (Yes in step S6), it ends the monitoring of the processing status. Note that when the arithmetic unit 18 determines in step S3 that there is no abnormal tendency in the processing state, that is, it is in a smooth state (No in step S3), it executes the determination process of whether or not the shaping process in step S6 has ended.

[0077] According to the above embodiment, the additive manufacturing apparatus 100 includes a wire feeder 8 that supplies the wire 6, an irradiation unit that irradiates the wire 6 with a laser beam 5 to melt the wire 6, and an imaging unit that captures an image including a connection portion where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and generates the image data 60 or / and the image data 70, and a determination unit 183 that determines the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data 60 or / and the image data 70.

[0078] Thereby, it becomes possible to recognize the connection state between the wire 6 and the workpiece, and to grasp whether it is in a smooth state, a drop tendency, or a stub tendency. In addition, the processing state can be monitored in real time using the image captured by the imaging unit and generated based on the image data 60 or / and the image data 70. For example, it becomes possible to recognize the drop tendency and the stub tendency at an early stage.

[0079] Therefore, according to the present embodiment, it is possible to provide an additive manufacturing apparatus capable of accurately grasping the processing state.

[0080] Embodiment 2. In the above Embodiment 1, the arithmetic unit 18 has been described as an example provided inside the additive manufacturing apparatus 100, but the present invention is not limited thereto. The arithmetic unit 18 may be provided, for example, inside a personal computer (PC) existing in the same network or a different network outside the additive manufacturing apparatus 100. FIG. 17 is a diagram showing the configuration of an additive manufacturing system 300 according to Embodiment 2. The additive manufacturing system 300 shown in FIG. 17 includes an additive manufacturing apparatus 100A and an arithmetic unit 18. Except for the configurations of the additive manufacturing apparatus 100A and the arithmetic unit 18, the configurations of the additive manufacturing apparatus 100 in the above embodiment are the same.

[0081] Embodiment 3. In Embodiment 3, an example of learning various parameter sets in order to improve the accuracy of determining the processing state of a workpiece by a machine learning method will be described. FIG. 18 is a block diagram showing the configuration of an arithmetic unit included in the additive manufacturing apparatus according to Embodiment 3. The arithmetic unit 18A shown in FIG. 18 performs calculations for additive manufacturing. FIG. 18 shows the functional configuration of the arithmetic unit 18A.

[0082] The arithmetic unit 18A has, for example, an acquisition unit 181, a creation unit 182, a determination unit 183, and an adjustment unit 184, similar to the arithmetic unit 18 shown in FIG. 1. The arithmetic unit 18A has a learning device 185A that performs machine learning, an inference device 186A that determines the processing state of the workpiece, and a model storage unit 187A that stores a learned model.

[0083] The learning device 185A learns various parameter sets and generates a learned model for determination to improve the accuracy of determining the workpiece.

[0084] The inference device 186A infers the processing state of the workpiece using the learned model for determination generated by the learning device 185A.

[0085] FIG. 19 is a block diagram showing the functional configuration of the learning device according to Embodiment 3. The learning device 185A shown in FIG. 19 includes a data acquisition unit 1851A and a model generation unit 1852A. The data acquisition unit 1851A acquires, as a learning data set, image data including a connection part where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and a label associated with the image data. The data acquisition unit 1851A outputs the acquired learning data set to the model generation unit 1852A. Note that the image data may be time-series image data including image data at a plurality of time points.

[0086] The label in the present embodiment is information indicating the connection state of the connection part included in the image data, that is, the processing state of the workpiece. The label includes, for example, a smooth state close to the initial state, a drop tendency, a drop state, a stub tendency, and a stub state. Note that the types of labels are not limited to these and may be further subdivided. Also, the label is set by the manufacturer or the user, for example, for each piece of image data or time-series image data. However, the method of setting the label may be arbitrary.

[0087] The model generation unit 1852A learns the correspondence between the image data and the processing state based on the data set created by the data acquisition unit 1851A. As the learning algorithm used by the model generation unit 1852A, known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be used. As an example, the case where a neural network is applied will be described.

[0088] The model generation unit 1852A learns, for example, a processing state associated with image data by so-called supervised learning according to a neural network model. Here, supervised learning refers to a method of giving a learning device a set of input and result (label) data, learning a feature in the learning data, and inferring a result from the input.

[0089] A neural network is composed of an input layer consisting of a plurality of neurons, an intermediate layer (hidden layer) consisting of a plurality of neurons, and an output layer consisting of a plurality of neurons. The intermediate layer may be one layer or two or more layers. FIG. 20 is a diagram showing an example of a neural network used by the model generation unit 1852A shown in FIG. 19. In the three-layer neural network shown in FIG. 20, when a plurality of inputs are input to the input layer (X1-X3), the values are multiplied by weights W1 (w11-w16) and input to the intermediate layer (Y1-Y2), and the result is further multiplied by weights W2 (w21-w26) and output from the output layer (Z1-Z3). This output result varies depending on the values of weights W1 and W2.

[0090] In Embodiment 3, the neural network learns a processing state associated with image data by so-called supervised learning according to learning data created based on a combination of the image data acquired by the data acquisition unit 1851A and a label associated with the image data. That is, the neural network learns by adjusting weights W1 and W2 so that the result output from the output layer when the image data is input to the input layer approaches the processing state indicated by the label associated with the image data.

[0091] The model generation unit 1852A generates and outputs a learned model for determination by executing the above learning. The model storage unit 187A stores the learned model for determination output from the model generation unit 1852A.

[0092] Next, the processing executed by the learning device 185A will be described. FIG. 21 is a diagram showing an example of the processing executed by the learning device in Embodiment 3.

[0093] According to FIG. 21, first, the data acquisition unit 1851A acquires, as a learning data set, image data including a connection portion where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and a label associated with the image data (step S11). Although the image data and the label are acquired simultaneously, it is only necessary that the image data and the label can be input in association with each other, and the image data and the label may be acquired at different timings.

[0094] Based on the image data acquired by the data acquisition unit 1851A and the learning data created based on the combination of the label associated with the image data, the model generation unit 1852A learns the processing state associated with the image data by so-called supervised learning and generates a learned model for determination (step S12).

[0095] The model storage unit 187A stores the learning model generated by the model generation unit 1852A (step S13).

[0096] FIG. 22 is a block diagram showing the functional configuration of the inference device according to Embodiment 3. The inference device 186A shown in FIG. 22 includes a data acquisition unit 1861A and an inference unit 1862A. The data acquisition unit 1861A acquires image data including a connection portion where the melted wire 6 and the workpiece to which the wire 6 is added are connected. This image data is, for example, the image data captured and generated by the internal sensor unit 23 or the external sensor unit 24 shown in FIG. 1, that is, the imaging unit.

[0097] The inference unit 1862A infers the processing state obtained by using the learned model for determination. That is, by inputting the image data acquired by the data acquisition unit 1861A into this learned model for determination, the processing state inferred from the image data can be output.

[0098] Next, the processing executed by the inference device 186A will be described. FIG. 23 is a diagram showing an example of the processing executed by the inference device in Embodiment 3.

[0099] According to FIG. 23, the data acquisition unit 1861A acquires image data including a connection part where the melted wire 6 and the workpiece to which the wire 6 is added are connected (step S21).

[0100] The inference unit 1862A acquires the learned model for determination from the model storage unit 187A, inputs the image data acquired in step S21 to the acquired learned model for determination, and obtains the processing state inferred from the image data (step S22).

[0101] The inference unit 1862A outputs the processing state obtained from the learned model for determination to the determination unit 183 as an inference result (step S23). Thereby, the determination unit 183 can determine the processing state of the workpiece using the processing state obtained from the learned model for determination. That is, the determination unit 183 learns about the relationship between the image data including the connection part where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and the processing state associated with the image data, and based on the learned model thus obtained, can determine the processing state of the workpiece.

[0102] According to Embodiment 3, the additive manufacturing apparatus 100 determines the processing state of the workpiece using a learned model for determination for improving the accuracy of determining the processing state of the workpiece. Specifically, the additive manufacturing apparatus 100 determines the processing state of the workpiece based on a learned model obtained by learning about the relationship between the image data including the connection part where the melted wire 6 and the workpiece to which the wire 6 is added are connected, and the processing state associated with the image data. Thereby, it becomes possible to improve the accuracy of determining the processing state. Also, since the accuracy of determining the processing state can be improved, it becomes possible to recognize a more refined processing state.

[0103] In addition, in the third embodiment, the learning device 185A has been described as being built into the additive manufacturing device 100, but the present invention is not limited to this. The learning device 185A is not limited to the devices included in the additive manufacturing device 100, and may be a device external to the additive manufacturing device 100. The learning device 185A may be a device that can be connected to the additive manufacturing device 100 via a network. The learning device 185A may be a device existing on a cloud server.

[0104] The learning device 185 may learn parameters according to a data set created for a plurality of additive manufacturing devices 100. The learning device 71 may acquire a data set from a plurality of additive manufacturing devices 100 used at the same site, or may acquire a data set from a plurality of additive manufacturing devices 100 used at different sites. The data set may be collected from a plurality of additive manufacturing devices 100 operating independently of each other at a plurality of sites. After starting the collection of the data set from the plurality of additive manufacturing devices 100, a new additive manufacturing device 100 may be added to the target from which the data set is collected. Further, after starting the collection of the data set from the plurality of additive manufacturing devices 100, a part of the plurality of additive manufacturing devices 100 may be excluded from the target from which the data set is collected.

[0105] The learning device 185 that has learned about one additive manufacturing device 100 may learn about other additive manufacturing devices 100 other than the additive manufacturing device 100. The learning device 71 that learns about the other additive manufacturing device 100 can update the output prediction model by relearning in the other additive manufacturing device 100.

[0106] Embodiment 4. In Embodiment 4, an example of learning various parameter sets will be described in order to improve the accuracy of adjusting processing conditions by a machine learning method. FIG. 24 is a block diagram showing the configuration of an arithmetic unit included in the additive manufacturing apparatus according to Embodiment 4. The arithmetic unit 18B shown in FIG. 24 performs arithmetic operations for additive manufacturing. FIG. 24 shows the functional configuration of the arithmetic unit 18B.

[0107] The arithmetic unit 18B has, for example, an acquisition unit 181, a creation unit 182, a determination unit 183, and an adjustment unit 184, similar to the arithmetic unit 18 shown in FIG. 1. The arithmetic unit 18B includes a learning device 185B that performs machine learning, an inference device 186B that adjusts processing conditions, and a model storage unit 187B that stores a learned model.

[0108] The learning device 185B learns various parameter sets and generates an adjusted learned model for improving the accuracy of adjusting processing conditions.

[0109] The inference device 186B infers the adjusted processing conditions using the adjusted learned model generated by the learning device 185B.

[0110] FIG. 25 is a block diagram showing the functional configuration of the learning device according to Embodiment 4. The learning device 185B includes a data acquisition unit 1851B and a model generation unit 1852B. The data acquisition unit 1851B acquires at least one of the laser output of the laser beam 5, the feeding speed of the wire 6, and the information for controlling the position of the processing head, the processing conditions including the above, the image data including the wire 6 and the workpiece corresponding to a preset target processing state, and the image data including the wire 6 and the workpiece acquired after the processing is started according to the processing conditions. The image data including the wire 6 and the workpiece acquired after the processing is started according to the processing conditions may be image data acquired at any timing as long as it is image data acquired after the processing is started. For example, it may be time-series image data acquired after a predetermined time has elapsed immediately after the processing is started or until the processing is completed. Further, the image data including the wire 6 and the workpiece acquired after the processing is started according to the processing conditions may be image data acquired until a further predetermined time has elapsed after a predetermined time has elapsed after the processing is started. Further, the image data including the wire 6 and the workpiece acquired after the processing is started according to the processing conditions may be image data acquired at the end of the processing. Further, as the timing when the processing is completed, for example, when a finished product is formed by a plurality of layers, it may be the timing when any one layer of the plurality of layers is formed or the timing when the finished product is formed.

[0111] The data acquisition unit 1851B creates a data set including the acquired processing conditions, image data, and other various measurement data. The data acquisition unit 1851B outputs the created data set to the model generation unit 1852B.

[0112] The model generation unit 1852B generates an adjusted pre-trained model using the data set output from the data acquisition unit 1851B. As the learning algorithm used by the model generation unit 1852B, known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be used. As an example, the case of applying reinforcement learning will be described. Reinforcement learning is a method in which an agent, which is an acting entity in a certain environment, observes the current state and determines the action to be taken. By selecting an action, the agent obtains a reward from the environment and learns a policy that maximizes the reward through a series of actions. As typical methods of reinforcement learning, Q-learning and TD-learning are known. For example, in the case of Q-learning, the action value table, which is a general update formula for the action value function Q(s,a), is represented by the following formula (1). The action value function Q(s,a) represents the action value Q, which is the value of the action of selecting the action "a" under the environment "s".

[0113]

Number

[0114] In the above formula (1), "st" represents the state of the environment at time "t". "at" represents the action at time "t". Due to the action "at", the state changes from "st" to "st+1". "rt+1" represents the reward obtained by the change of the state from "st" to "st+1". "γ" represents the discount rate and satisfies 0 < γ ≤ 1. "α" represents the learning coefficient and satisfies 0 < α ≤ 1. In Embodiment 4, the action "at" is a processing condition including at least one of the information for controlling the laser output of the laser beam 5, the feeding speed of the wire 6, and the position of the processing head 3. Note that the axial movement speed of the processing head 3 may be added to the processing condition. The state "st" is, for example, image data including the wire 6 and the workpiece obtained after processing is started according to the processing condition. The state "st" is not limited in the acquisition period and the number of acquisitions as long as it is image data obtained from after processing is started to when processing is completed according to the processing conditions included in the dataset. That is, the acquisition period and the number of acquisitions can be arbitrarily set according to the purpose of learning. The model generation unit 1852 learns the best action "at" in the state "st" at time "t".

[0115] In the update formula represented by the above formula (1), if the action value of the best action "a" at time "t+1" is greater than the action value Q of the action "a" executed at time "t", the action value Q is increased, and in the reverse case, the action value Q is decreased. In other words, the action value function Q(s,a) is updated so that the action value Q of the action "a" at time "t" approaches the best action value at time "t+1". Thereby, the best action value in a certain environment is sequentially propagated to the action values in the previous environments.

[0116] The model generation unit 1852B includes a reward calculation unit 1853B and a function update unit 1854B. The reward calculation unit 1853B calculates a reward based on the difference between the image data included in the dataset and the image data corresponding to a preset target processing state. The image data included in the dataset is, for example, image data acquired after processing is started under the processing conditions included in the dataset, and represents an image including a connection portion where the melted wire and the workpiece to which the wire is added are connected. The function update unit 1854B updates a function for determining processing conditions according to the reward calculated by the reward calculation unit 1853B. The function update unit 1854B outputs the learned model created by the update of the function to the model storage unit 187B.

[0117] The reward calculation unit 1853B calculates a reward "r" based on the difference between the image data included in the dataset and the image data corresponding to a preset target processing state. When the difference becomes smaller, the reward calculation unit 1853B increases the reward "r". The reward calculation unit 1853B increases the reward "r" by giving the value "1" of the reward. Note that the value of the reward is not limited to "1". Also, when the difference becomes larger, the reward calculation unit 1853B reduces the reward "r". The reward calculation unit 1853B reduces the reward "r" by giving the value "-1" of the reward. Note that the value of the reward is not limited to "-1". When there are constraints on the image data that can be captured or generated by the internal sensor unit 23 or the external sensor unit 24, various variables that can be measured by the internal sensor unit 23 or the external sensor unit 24, or the processing conditions, the reward calculation unit 1853B may reduce the reward when the constraints are not satisfied.

[0118] FIG. 26 is a diagram showing an example of the processing executed by the learning device in Embodiment 4. Hereinafter, with reference to FIG. 26, a reinforcement learning method for updating the action value function Q(s, a) will be described.

[0119] According to FIG. 26, the learning device 185B acquires learning data (step S31). Next, the learning device 185B calculates a reward (step S32). Next, the learning device 185B updates the action value function Q(s,a) based on the reward (step S33). Then, the learning device 185B determines whether the action value function Q(s,a) has converged (step S34). The learning device 185B determines that the action value function Q(s,a) has converged when the update of the action value function Q(s,a) in step S23 no longer occurs.

[0120] When the learning device 185B determines that the action value function Q(s,a) has not converged (No in step S34), it returns the operation procedure to step S11. Note that the learning device 185 may continue learning by returning the operation procedure from step S33 to step S31 without performing the determination in step S14.

[0121] When the learning device 185B determines that the action value function Q(s,a) has converged (Yes in step S34), it ends the learning. Then, the model storage unit 187B stores the adjusted learned model, which is the generated action value function Q(s,a) (step S35).

[0122] FIG. 27 is a block diagram showing the functional configuration of the inference device according to Embodiment 4. The inference device 186B shown in FIG. 27 infers processing conditions that approach preset target processing conditions based on the learned model. The inference device 186B includes a data acquisition unit 1861B and an inference unit 1862B.

[0123] Before adjustment, the data acquisition unit 1861B acquires, as inference data, image data corresponding to a preset target processing state. The data acquisition unit 1861B outputs the acquired image data to the inference unit 1862B. By inputting the image data corresponding to the preset target processing state to the learned model read from the model storage unit 187B, the inference unit 1862B infers processing conditions approaching the preset target processing conditions. Note that the data acquisition unit 1861B may temporarily stop the process to acquire image data before adjustment, or may acquire image data in real time while performing the process. Further, the data acquisition unit 1861B may acquire image data for each material used for the workpiece, for each processing condition, or for each processing shape. Further, the preset target processing state is set based on, for example, the image data captured before adjustment. The target processing state can be arbitrarily set according to the processing situation, and may be set by the user, for example.

[0124] Figure 28 is a diagram showing an example of the process executed by the inference device according to Embodiment 4. According to Figure 28, in the inference device 186B, the data acquisition unit 1861B acquires inference data (step S41). Next, in the inference unit 1862B of the inference device 186B, the acquired inference data is input to the adjusted learned model to obtain processing conditions approaching the preset target processing conditions (step S42). Then, the inference device 186B outputs the processing conditions obtained from the adjusted learned model to the adjustment unit 184 as an inference result (step S43). As a result, the adjustment unit 184 can generate the adjusted processing condition information 28 using the processing conditions obtained from the adjusted learned model.

[0125] According to Embodiment 4, the additive manufacturing apparatus 100 generates adjusted processing condition information 28 using a learned model for adjustment to improve the accuracy of adjustment of processing conditions. Specifically, the additive manufacturing apparatus 100 includes, among the information for controlling the laser output of the laser beam 5, the feeding speed of the wire 6, and the position of the processing head 3, at least one processing condition, image data including the wire 6 and the workpiece corresponding to a preset target processing state, and image data including the wire 6 and the workpiece acquired after processing is started under the processing condition. Based on the learned model obtained by learning the relationship therebetween, the adjusted processing condition information 28 is generated. Thereby, it becomes possible to improve the accuracy of adjustment of processing conditions.

[0126] In Embodiment 4, the case where reinforcement learning is applied to the learning algorithm used by the learning apparatus 185B has been described. However, learning other than reinforcement learning may be applied to the learning algorithm. The learning apparatus 185B may perform machine learning using a known learning algorithm other than reinforcement learning, for example, a learning algorithm such as deep learning, neural network, genetic programming, functional logic programming, or support vector machine.

[0127] In Embodiment 4, the learning apparatus 185B has been described as an example built into the additive manufacturing apparatus 100, but it is not limited thereto. The learning apparatus 185B is not limited to the apparatus included in the additive manufacturing apparatus 100, and may be an apparatus external to the additive manufacturing apparatus 100. The learning apparatus 185B may be an apparatus connectable to the additive manufacturing apparatus 100 via a network. The learning apparatus 185 may be an apparatus existing on a cloud server.

[0128] The learning device 185B may learn parameters according to a data set created for a plurality of additive manufacturing devices 100. The learning device 71 may acquire a data set from a plurality of additive manufacturing devices 100 used at the same site, or may acquire a data set from a plurality of additive manufacturing devices 100 used at different sites. The data set may be collected from a plurality of additive manufacturing devices 100 operating independently of each other at a plurality of sites. After starting the collection of the data set from the plurality of additive manufacturing devices 100, a new additive manufacturing device 100 may be added to the target from which the data set is collected. Also, after starting the collection of the data set from the plurality of additive manufacturing devices 100, a part of the plurality of additive manufacturing devices 100 may be excluded from the target from which the data set is collected.

[0129] The learning device 185B that has learned about one additive manufacturing device 100 may learn about other additive manufacturing devices other than the additive manufacturing device 100. The learning device 185B that learns about the other additive manufacturing device can update the output prediction model by relearning in the other additive manufacturing device.

[0130] Next, the hardware configuration of the arithmetic units 18, 18A, or 18B according to Embodiments 1 to 4 will be described. FIG. 29 is a diagram showing an example of the hardware configuration of the arithmetic unit included in the additive manufacturing device according to Embodiment 1. FIG. 29 shows the hardware configuration in the case where the functions of the arithmetic units 18, 18A, or 18B are realized by using the hardware that executes the program.

[0131] The arithmetic units 18, 18A, or 18B include a processor 901 that executes various processes, a memory 902 that is a built-in memory, a storage device 903 that stores information, and an interface circuit 904 for inputting information to and outputting information from the arithmetic unit 18.

[0132] The processor 901 is a CPU (Central Processing Unit). The processor 901 may also be a processing device, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor). The memory 902 is a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory), or an EEPROM (registered trademark) (Electrically Erasable Programmable Read Only Memory).

[0133] The storage device 903 is an HDD (Hard Disk Drive) or an SSD (Solid State Drive). The program that causes the computer to function as the arithmetic unit 18 is stored in the storage device 903. The processor 901 reads out the program stored in the storage device 903 to the memory 902 and executes it.

[0134] The program may be stored in a storage medium readable by a computer system. The arithmetic unit 18 may store the program recorded in the storage medium into the memory 902. The storage medium may be a portable storage medium such as a flexible disk or a flash memory which is a semiconductor memory. The program may be installed in the computer system from another computer or a server device via a communication network.

[0135] The functions of the acquisition unit 181, creation unit 182, determination unit 183, adjustment unit 184, learning device 185A, learning device 185B, inference device 186A, and inference device 186B are realized by a combination of the processor 81 and software. Each of these functions may be realized by a combination of the processor 901 and firmware, or may be realized by a combination of the processor 901, software, and firmware. The software or firmware is described as a program and stored in the storage device 903. The function of the model storage unit 187 is realized by using the storage device 903.

[0136] The interface circuit 904 receives signals from external devices connected to the hardware. The external devices are the NC device 19, the internal sensor unit 23, and the external sensor unit 24.

[0137] In the above embodiment, the adjustment unit 184 adjusts, for example, the laser output of the laser beam 5, the feeding speed of the wire 6, the information for controlling the position of the processing head 3, and the axial movement speed of the processing head 3, but is not limited thereto. For example, the adjustment unit 184 may adjust the intensity distribution of the laser beam 5 irradiated on the workpiece.

[0138] The present disclosure can be implemented in various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Also, the above-described embodiments are for explaining the present disclosure and do not limit the scope of the present disclosure. That is, the scope of the present disclosure is indicated by the claims rather than the embodiments. And various modifications made within the scope of the claims and within the scope of the meaning of the disclosure equivalent thereto are considered to be within the scope of the present disclosure.

Industrial Applicability

[0139] According to the present disclosure, it is possible to provide an additive manufacturing apparatus capable of accurately grasping the processing state.

Explanation of Signs

[0140] 1 Laser oscillator, 2 Fiber cable, 3 Processing head, 4 Beam nozzle, 5 Laser beam, 6 Wire, 7 Wire spool, 8 Wire feeder, 9 Wire nozzle, 10 Processing point, 11 Bead, 13 Base material, 14, 15 Rotary stage, 16 Support bracket, 17 XYZ stage, 18, 18A, 18B Arithmetic unit, 19 NC device, 21 Gas flow regulator, 22 Drive controller, 23 Internal sensor unit, 24 External sensor unit, 27 Processing program, 28 Processing condition information, 60, 70 Image data, 181 Acquisition unit, 182 Creation unit, 183 Judgment unit, 184 Adjustment unit, 185A, 185B Learning device, 186A, 186B Inference device, 187A, 187B Model storage unit, 221 Head drive unit, 222 Wire feeding drive unit, 223 Stage drive unit, 100, 100A Additive manufacturing device, 300 Additive manufacturing system, 901 Processor, 902 Memory, 903 Storage device, 904 Interface circuit.

Claims

1. A wire supply unit that supplies a wire, An irradiation unit that irradiates the wire with a laser beam to melt the wire, An imaging unit that captures an image including a connection portion where the melted wire and a workpiece to which the wire is added are connected, and generates image data, A determination unit that determines the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated by the imaging unit, Comprising: The connection portion includes a solid-liquid phase molten pool formed by melting the wire, and a first portion that connects the molten pool and the wire, has a lower pixel value than the region indicating the molten pool, and has a higher pixel value than the region indicating the wire, The determination unit determines whether the processing state has a drop tendency or a stub tendency based on changes in the size, shape, and position state of the first portion, and changes in the positional relationship between the wire and the workpiece. An additive manufacturing apparatus.

2. The determination unit determines the processing state of the workpiece using information indicating changes in the distribution of pixel values in a region including the connection portion among the images based on the time-series image data generated by the imaging unit. The additive manufacturing apparatus according to Claim 1.

3. The image data generated by the imaging unit includes information indicating the distribution of the intensity of light radiated from the connection portion. The additive manufacturing apparatus according to Claim 1.

4. When the determination unit determines that there is an abnormal tendency in the processing state, the additive manufacturing apparatus according to any one of Claims 1 to 3 further includes an adjustment unit that generates adjusted processing condition information so as to approach a preset target processing state corresponding to the content of the abnormal tendency.

5. The determination unit determines the processing state of the workpiece based on a learned model obtained by learning the relationship between the image data including the connection portion and the processing state associated with the image data. The additive manufacturing apparatus according to any one of Claims 1 to 3.

6. The adjustment unit generates the adjusted processing condition information based on a learned model obtained by learning about the relationship between at least one processing condition including information for controlling the laser output of the laser beam, the feeding speed of the wire, and the position of the processing head, image data including the wire and the workpiece corresponding to a preset target processing state, and image data including the wire and the workpiece obtained after processing is started according to the processing condition. The additive manufacturing apparatus according to claim 4.

7. A wire supply step of supplying a wire; An irradiation step of irradiating the wire with a laser beam to melt the wire; An imaging step of imaging an image including a connection portion where the melted wire and the workpiece to which the wire is added are connected and generating image data; A determination step of determining the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated in the imaging step; comprising The connection portion includes a solid-liquid phase molten pool formed by melting the wire, and a first portion that connects the molten pool and the wire, has a lower pixel value than the region indicating the molten pool, and has a higher pixel value than the region indicating the wire. The determination step determines whether the processing state has a dropping tendency or a stub tendency based on the state changes in the size, shape, and position of the first portion, and the change in the positional relationship between the wire and the workpiece. Additive manufacturing method.

8. A wire supply unit for supplying a wire; An irradiation unit for irradiating the wire with a laser beam to melt the wire; An imaging unit for imaging an image including a connection portion where the melted wire and the workpiece to which the wire is added are connected and generating image data; A determination unit for determining the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated by the imaging unit; comprising The connection portion includes a solid-liquid phase molten pool formed by melting the wire, and a first portion that connects the molten pool and the wire, has a lower pixel value than the region indicating the molten pool, and has a higher pixel value than the region indicating the wire. The determination unit determines whether the processing state has a dropping tendency or a stubbing tendency based on changes in the size, shape, and position state of the first part, and changes in the positional relationship between the wire and the workpiece. Additive manufacturing system.

9. To a processor, an imaging function that causes an imaging unit to image an image including a connection portion where a wire melted by irradiation with a laser beam is connected to a workpiece to which the wire is added and generates image data; a determination function that determines the processing state of the workpiece using information indicating the distribution of pixel values in a region including the connection portion among the images based on the image data generated by the imaging function; to be realized, The connection portion includes a molten pool in a solid-liquid phase formed by melting the wire, and a first part that connects the molten pool and the wire, has a lower pixel value than the region indicating the molten pool, and has a higher pixel value than the region indicating the wire. The determination function determines whether the processing state has a dropping tendency or a stubbing tendency based on changes in the size, shape, and position state of the first part, and changes in the positional relationship between the wire and the workpiece. Program.

Images