Metal additive manufacturing system, metal additive manufacturing device, and method for manufacturing metal additive manufactured article

Abstract

A metal additive manufacturing system (1) comprises: an article information input unit (5) to which article information indicating a material and a shape of an article to be manufactured is input; a manufacturing condition generation unit (6) that generates a manufacturing condition of the article on the basis of the article information; a calculation unit (10) that predicts a physical property value indicating a characteristic of the article at each of a plurality of positions of the article or a temperature history of each of the plurality of positions when the article indicated by the article information is manufactured according to the manufacturing condition; a processing unit (11) that determines whether a prediction result of the physical property value or the temperature history falls within a set allowable range; and a manufacturing unit (7) that manufactures the article according to the manufacturing condition adjusted on the basis of a result of the determination by the processing unit (11).

Description

Metal Additive Manufacturing System, Metal Additive Manufacturing Apparatus, and Method for Manufacturing Metal Additive Manufactured Object

[0001] The present disclosure relates to a metal additive manufacturing system, a metal additive manufacturing apparatus, and a method for manufacturing a metal additive manufactured object.

[0002] As one of the technologies for manufacturing three-dimensional objects, the technology of additive manufacturing (AM) is known. In ensuring the quality of an object obtained by additive manufacturing using a metal material, it is important to maintain a certain accuracy in the shape of the object and to maintain the homogeneity of the physical properties of the object for each position. Among these, regarding homogeneity, conventionally, it has been confirmed by measuring physical property values using a pre-manufactured standard shape sample. However, since the process of additive manufacturing is complex, the thermal history often differs greatly between the standard shape sample and the actual manufactured object. Therefore, in the case of a method of confirming the physical property values of a manufactured object using data obtained from a standard shape sample, it has been difficult to accurately grasp the homogeneity of the manufactured object. Hereinafter, the thermal history is also referred to as the "temperature history".

[0003] Patent Document 1 discloses a database that represents the correspondence relationship between input information including a material, welding conditions, and a welding track, and the characteristics of a manufactured object obtained when manufacturing is performed based on the input information, and based on the database, a manufacturing plan support apparatus that predicts the characteristics corresponding to each condition of the material, welding conditions, and welding track is disclosed. Patent Document 1 also discloses obtaining each condition of the material, welding conditions, and welding track corresponding to the target characteristics based on the database.

[0004] Japanese Patent Application Laid-Open No. 2022-20393

[0005] Because the way in which the temperature changes varies depending on the location in the fabricated object, the temperature history of the fabricated object changes depending on its location. For example, the temperature history differs significantly between the start point and the end point of fabrication. This difference in temperature history at each location in the fabricated object results in variations in the metallic structure of the fabricated object. According to the technology described in Patent Document 1, the variation in metallic structure at each location in the fabricated object is not taken into consideration, and only one material property is presented as the predicted property for each fabricated object. With the technology described in Patent Document 1, it is not possible to accurately grasp the homogeneity of the fabricated object, making it difficult to determine the conditions for making the properties at each location in the fabricated object the desired properties. For this reason, the conventional technology shown in Patent Document 1 has the problem that it is difficult to manufacture a homogeneous fabricated object.

[0006] This disclosure has been made in view of the above, and aims to provide a metal additive manufacturing system that can manufacture homogeneous molded objects.

[0007] To solve the above-mentioned problems and achieve the objectives, the metal additive manufacturing system according to this disclosure comprises: an object information input unit into which object information indicating the material and shape of the object is input; a manufacturing condition generation unit that generates manufacturing conditions for the object based on the object information; a calculation unit that predicts physical property values ​​indicating the characteristics of the object at each of a plurality of positions or the temperature history at each of a plurality of positions when the object indicated in the object information is manufactured according to the manufacturing conditions; a processing unit that determines whether the predicted results of the physical property values ​​or temperature history fall within a set tolerance range; and a manufacturing unit that manufactures the object according to the manufacturing conditions adjusted based on the result of the determination by the processing unit.

[0008] The metal additive manufacturing system described herein has the effect of being able to manufacture homogeneous objects.

[0009] Figures showing an example configuration of the metal additive manufacturing system according to Embodiment 1 Figures showing an example configuration of the molding section of the metal additive manufacturing system according to Embodiment 1 Flowchart showing an example of the operation procedure of the metal additive manufacturing system according to Embodiment 1 Figures explaining the relationship between molding conditions and the physical properties of the molded object in Embodiment 1 Figure 1 shows an example of a molded object manufactured by the metal additive manufacturing system according to Embodiment 1 Figure 2 shows an example of a molded object manufactured by the metal additive manufacturing system according to Embodiment 1 Figure 3 shows an example of a molded object manufactured by the metal additive manufacturing system according to Embodiment 1 Figures showing an example of a database held in the storage unit of the metal additive manufacturing system according to Embodiment 1 Figure showing an example configuration of a metal additive manufacturing system according to Embodiment 3; a flowchart showing an example of the operation procedure of the metal additive manufacturing system according to Embodiment 3; a first figure for explaining the heat transfer characteristics in an object manufactured by the metal additive manufacturing system according to Embodiment 4; a second figure for explaining the heat transfer characteristics in an object manufactured by the metal additive manufacturing system according to Embodiment 4; a first figure for explaining a method for calculating the sum of heat transfer indices in the metal additive manufacturing system according to Embodiment 4; a second figure for explaining a method for calculating the sum of heat transfer indices in the metal additive manufacturing system according to Embodiment 4; a figure showing an example configuration of a control circuit according to Embodiments 1 to 6.

[0010] The following describes in detail, with reference to the drawings, a metal additive manufacturing system, a metal additive manufacturing apparatus, and a method for manufacturing metal additive manufactured products according to the embodiment.

[0011] Embodiment 1. Figure 1 shows an example of the configuration of a metal additive manufacturing system 1 according to Embodiment 1. The metal additive manufacturing system 1 comprises a metal additive manufacturing apparatus 2 and an information processing apparatus 3.

[0012] The metal additive manufacturing apparatus 2 manufactures metal additive manufactured objects, which are three-dimensional objects made of metal material. In the following description, "manufactured object" refers to a metal additive manufactured object. The information processing device 3 is communicatively connected to the metal additive manufacturing apparatus 2. The information processing device 3 processes information related to additive manufacturing in the metal additive manufacturing apparatus 2.

[0013] The metal additive manufacturing apparatus 2 comprises a molded object information input unit 5, a molding condition generation unit 6, and a molding unit 7. The molded object information input unit 5 receives molded object information indicating the material and shape of the molded object. The molded object information input unit 5 outputs the molded object information to the molding condition generation unit 6.

[0014] The molding condition generation unit 6 is implemented, for example, by using CAM (Computer Aided Manufacturing). The molding condition generation unit 6 generates molding conditions for the object based on the object information. The molding condition generation unit 6 generates various conditions for molding an object of the material and shape indicated in the object information as molding conditions. The molding conditions generated by the molding condition generation unit 6 include a molding path, which is the movement path that moves the processing point. Details of the molding conditions will be described later. The molding condition generation unit 6 outputs a numerical control (NC) program including the generated molding conditions to the molding unit 7. The molding unit 7 performs molding according to the NC program.

[0015] The information processing device 3 comprises a data input unit 8, a storage unit 9, an arithmetic unit 10, and a processing unit 11. The data input unit 8 receives data related to additive manufacturing. The data input unit 8 stores the input data in the storage unit 9.

[0016] The memory unit 9 stores various data related to additive manufacturing. The data stored in the memory unit 9 includes manufacturing performance data and thermal simulation data. Manufacturing performance data is data that associates the manufacturing conditions and manufactured object information from past manufacturing processes with the physical properties or temperature history at each of multiple locations on the manufactured object. Manufacturing performance data shows the results of past additive manufacturing. Thermal simulation data is the result of simulating the temperature history at each of multiple locations on the manufactured object. The data stored in the memory unit 9 is read out as appropriate by the calculation unit 10.

[0017] The calculation unit 10 comprises a prediction unit 12 and a learning unit 13. The prediction unit 12 predicts physical properties or temperature history for each of multiple locations of a printed object, when the printed object shown in the printed object information is printed according to the printing conditions. The prediction unit 12 predicts the physical properties or temperature history based on the printing performance data or thermal simulation data, which are data read from the storage unit 9. The following describes a case in which the prediction unit 12 predicts physical properties or temperature history based on the printing performance data.

[0018] The learning unit 13 reads the molding performance data stored in the memory unit 9. The learning unit 13 generates a trained model by learning from the molding performance data. The learning unit 13 outputs the generated trained model to the prediction unit 12.

[0019] The prediction unit 12 receives the molding conditions generated by the molding condition generation unit 6. The prediction unit 12 inputs the molding conditions into a trained model, which is the result of learning from the molding performance data by the learning unit 13, and predicts physical properties or temperature history based on the molding performance data. The prediction unit 12 outputs the predicted physical properties or predicted temperature history to the processing unit 11.

[0020] The processing unit 11 determines whether the predicted results, such as physical properties or temperature history, input to the processing unit 11 fall within a set tolerance range. Details of the tolerance range will be described later. The processing unit 11 outputs the determination result to the prediction unit 12. The prediction unit 12 adjusts the molding conditions based on the determination result from the processing unit 11. The prediction unit 12 outputs the adjusted molding conditions to the molding condition generation unit 6. The molding condition generation unit 6 outputs the molding conditions input to the molding condition generation unit 6. As a result, the molding conditions adjusted based on the determination result from the processing unit 11 are output by the molding condition generation unit 6.

[0021] The molding unit 7 molds the object according to the molding conditions output by the molding condition generation unit 6. If the molding conditions are adjusted based on the determination result by the processing unit 11, the molding unit 7 molds the object according to the adjusted molding conditions.

[0022] Figure 2 shows an example of the configuration of the molding unit 7 of the metal additive manufacturing system 1 according to Embodiment 1. In Embodiment 1, the metal additive manufacturing apparatus 2 manufactures a molded object using a so-called Directed Energy Deposition (DED) method. The molding unit 7 of the metal additive manufacturing apparatus 2 supplies material to a commanded position and forms a bead with the material melted using a beam. The bead is a solidified product obtained when the molten material solidifies. The metal additive manufacturing apparatus 2 manufactures a molded object by sequentially stacking layers of beads using the molding unit 7.

[0023] The beam, which is the heat source for melting the material, is a laser beam, an electron beam, or an arc. In Embodiment 1, the heat source is assumed to be a laser beam. Also in Embodiment 1, the material is a metal wire 21. The material is not limited to a wire 21, and may be a metal powder.

[0024] The X, Y, and Z axes are three axes perpendicular to each other. The X and Y axes are two horizontal axes. The Z axis is a vertical axis. In each of the X, Y, and Z axes, the direction indicated by the arrow is considered positive, and the direction opposite to the arrow is considered negative. The positive Z direction is considered to be the vertically upward direction. The bead layers are stacked in the positive Z direction.

[0025] The molding unit 7 includes a wire nozzle 22 for supplying wire 21, a laser oscillator 23 for outputting a laser beam, a beam nozzle 24 through which the laser beam emitted toward the processing point passes, a fiber cable 25 which is an optical transmission path, a processing head 26, a gas nozzle 27 for injecting shielding gas toward the processing point, a gas supply device 28, a radiation thermometer 32, and a head drive device 33. The beam nozzle 24, gas nozzle 27, and radiation thermometer 32 are attached to the processing head 26. The molding unit 7 also includes a stage 31 on which the base material is placed, and a rotation mechanism 30 for rotating the stage 31. The molding unit 7 creates an object by layering beads on the base material. The molding unit 7 also includes a control device 35 which is an NC device.

[0026] The laser beam output by the laser oscillator 23 propagates through the fiber cable 25 and enters the processing head 26. Inside the processing head 26, an optical system such as a collimating optical system or a focusing optical system is arranged. The diagram of the optical system is omitted. The central axis of the beam nozzle 24 coincides with the optical axis of the optical system. The central axis of the beam nozzle 24 also coincides with the Z-axis. The center line of the laser beam emitted from the beam nozzle 24 coincides with the Z-axis. The laser beam passes through the optical system inside the processing head 26, passes through the beam nozzle 24, and is emitted from the processing head 26. The processing point is the irradiation position of the laser beam and the position where the molten material is added. The position of the processing point is the position where the heat source and material are supplied, and is on the central axis of the beam nozzle 24.

[0027] The gas supply device 28 supplies shielding gas from a gas supply source to the gas nozzle 27. An example of a gas supply source is a gas cylinder. The gas supply source is not shown in the diagram. The gas supply device 28 supplies shielding gas to the gas nozzle 27 through piping 29. The molding unit 7 reduces oxidation of the molded object and cools the object by injecting shielding gas toward the processing point. The shielding gas is an inert gas such as argon gas, for example.

[0028] The wire nozzle 22 supplies the wire 21, which is fed out by the material supply mechanism 34, to the processing point. Figure 2 shows an example of a side supply method in which the wire 21 is supplied from a wire nozzle 22 positioned diagonally above the processing point. The molding unit 7 may also employ a center supply method in which the wire 21 is supplied from a wire nozzle 22 positioned directly above the processing point, rather than a side supply method.

[0029] The head drive unit 33 moves the machining head 26. For example, the head drive unit 33 has a servo motor that moves the machining head 26 in the X-axis direction, a servo motor that moves the machining head 26 in the Y-axis direction, and a servo motor that moves the machining head 26 in the Z-axis direction. The illustration of each servo motor is omitted. The positional relationship between the machining head 26 and the wire nozzle 22 is fixed. The molding unit 7 moves the machining head 26 and the wire nozzle 22 within the stroke range by driving the head drive unit 33. By moving the machining head 26 and the wire nozzle 22, the molding unit 7 can move the machining point to any position within the stroke range.

[0030] The rotation mechanism 30 is an operating mechanism that enables the stage 31 to rotate around a first axis and around a second axis perpendicular to the first axis. In the rotation mechanism 30 shown in Figure 2, the first axis is parallel to the X axis, and the second axis is parallel to the Z axis. The rotation mechanism 30 includes a servo motor that rotates the stage 31 around the first axis and a servo motor that rotates the stage 31 around the second axis. The rotation mechanism 30 causes the stage 31 to perform rotational motion around each of the two axes by driving each servo motor. The individual servo motors are not shown.

[0031] The molding unit 7 rotates the stage 31 using the rotation mechanism 30, thereby changing the orientation of the base material placed on the stage 31. The molding unit 7 can change the orientation of the base material to an orientation suitable for processing.

[0032] The molding unit 7 moves the machining point by moving the machining head 26 relative to the stage 31. The molding unit 7 may also move the machining point by moving the stage 31 relative to the machining head 26. In this case, the molding unit 7 moves the stage 31 relative to the machining head 26 by moving the stage 31 in at least one of the three axes.

[0033] In Embodiment 1, the molding unit 7 is a 5-axis machining center capable of translational movement in the X, Y, and Z axes and rotational movement centered on two axes. Alternatively, the molding unit 7 may be a 3-axis machining center capable of translational movement in the X, Y, and Z axes, or a 4-axis machining center capable of translational movement in the X, Y, and Z axes and rotational movement centered on one axis.

[0034] The radiation thermometer 32 is a device for measuring the temperature of the fabricated object. The radiation thermometer 32 is a non-contact type thermometer. The location on the fabricated object to be measured can be selected as appropriate. In this case, the radiation thermometer 32 will be used to measure the temperature of the processing point. The radiation thermometer 32 is mounted coaxially with the laser beam emitted from the beam nozzle 24. Therefore, the radiation thermometer 32 can measure the temperature of the processing point while the laser beam is being emitted.

[0035] Furthermore, the radiation thermometer 32 can also measure the temperature at the processing point, i.e., the position where the bead was formed, after the laser beam emission has been stopped. Hereinafter, the position where the bead was formed will be referred to as the build point. The metal additive manufacturing system 1 can observe the temperature history of the build point after the laser beam emission has been stopped using the radiation thermometer 32. In addition, the radiation thermometer 32 can also measure the interlayer temperature, i.e., the interpass temperature, by measuring the temperature of the substrate immediately before the laser beam emission is started.

[0036] As described above, the metal additive manufacturing system 1 can acquire temperature information under various conditions by mounting a radiation thermometer 32 coaxially with the laser beam. Note that the radiation thermometer 32 mounted on the manufacturing unit 7 is not limited to one; there may be multiple. The manufacturing unit 7 may also be equipped with radiation thermometers 32 that measure the temperature at locations other than the manufacturing point.

[0037] The control device 35 controls the entire molding unit 7 according to the NC program. The NC program includes information necessary for molding, such as the laser output, which is the output of the laser beam from the laser oscillator 23; the movement speed of the processing head 26; the rotation speed of the stage 31; the material supply speed, which is the speed at which the wire 21 is supplied by the wire nozzle 22; and the gas flow rate, which is the flow rate of the shielding gas.

[0038] The control device 35 controls the head drive device 33 by outputting a position command to the head drive device 33. The head drive device 33 moves the machining head 26 along a preset molding path according to the position command. The control device 35 controls the rotary mechanism 30 by outputting a rotation command to the rotary mechanism 30. The rotary mechanism 30 rotates the stage 31 according to the rotation command.

[0039] The control device 35 controls the laser oscillator 23 by outputting a laser output command to the laser oscillator 23. The laser oscillator 23 outputs a laser beam according to the laser output command. The control device 35 controls the material supply mechanism 34 by outputting a material supply command to the material supply mechanism 34. The material supply mechanism 34 supplies the wire 21 at a material supply rate according to the material supply command. The control device 35 controls the gas supply device 28 by outputting a gas supply command to the gas supply device 28. The gas supply device 28 supplies shielding gas at a gas flow rate according to the gas supply command.

[0040] The molding conditions generated by the molding condition generation unit 6 shown in Figure 1 include various conditions for controlling each part of the molding unit 7 by the control device 35. The various conditions included in the molding conditions are conditions for the laser beam output, laser beam irradiation time, laser beam on and off timings, material supply speed, gas flow rate, processing head 26 movement speed, or stage 31 rotation speed.

[0041] In Embodiment 1, the temperature history is the value of the temperature measured by the radiation thermometer 32 or the value calculated from that value. The maximum temperature is the highest temperature among the measured temperatures for each position of the shaped object. The cooling rate is obtained by dividing the temperature difference from the maximum temperature to the freezing point of the material by the time required for the temperature to drop from the maximum temperature to the freezing point. The interlayer temperature is the temperature of the base immediately before the emission of the laser beam is started. The holding time is the time during which the temperature is held in a constant temperature range for each position of the shaped object.

[0042] Here, the constant temperature range is determined by the behavior of precipitates of each steel type which is the material. For example, in the case of an aluminum alloy in which age precipitation occurs when the temperature is within the range of 200°C to 300°C, the constant temperature range is 200°C to 300°C. In the case of a nickel-based alloy in which age precipitation occurs when the temperature is within the range of 620°C to 750°C, the constant temperature range is 620°C to 750°C. Thus, the constant temperature range can be appropriately selected according to the material.

[0043] Next, the operation of the metal laminated manufacturing system 1 will be described. FIG. 3 is a flowchart showing an example of the operation procedure by the metal laminated manufacturing system 1 according to Embodiment 1.

[0044] In step S1, the shaped object information input unit 5 receives the shaped object information. The shaped object information includes information indicating the material of the shaped object and information indicating the shape of the shaped object. Thereby, the material of the shaped object and the shaped shape which is the shape of the shaped object are set in the metal laminated manufacturing system 1.

[0045] In step S2, the molding condition generation unit 6 generates molding conditions based on the molded object information received in step S1. The molding condition generation unit 6 generates various conditions for molding an object of the material and shape indicated in the molded object information. The molding condition generation unit 6 also generates a molding path for molding an object of the material and shape indicated in the molded object information. The molding path is a path calculated by CAM. The molding path is set to avoid interference with the components of the molding unit 7. Furthermore, it is designed so that the shortest route among the paths that can build the object is generated as the molding path. The molding condition generation unit 6 outputs the various conditions and the molding path as molding conditions.

[0046] In step S3, the prediction unit 12 predicts physical properties or temperature history based on the manufacturing performance data or thermal simulation data. As an example, the prediction unit 12 predicts physical properties or temperature history based on the manufacturing performance data by inputting the manufacturing conditions into a trained model which is the result of learning the manufacturing performance data.

[0047] In step S4, the processing unit 11 determines whether the physical property values ​​predicted in step S3, or the temperature history predicted in step S3, exceed the acceptable range. The acceptable range is the range of values ​​that are acceptable as physical property values, or the range of values ​​that are acceptable as included in the temperature history. The acceptable range is set, for example, by the user of the metal additive manufacturing system 1. For example, the acceptable range may be the range of variation in values. For example, the processing unit 11 determines that the physical property values ​​exceed the acceptable range if the variation in physical property values ​​exceeds a predetermined range. Alternatively, the acceptable range may be a range defined by a threshold. For example, the processing unit 11 determines that the physical property values ​​exceed the acceptable range if the physical property values ​​exceed a predetermined threshold. Regarding the temperature history, for example, the processing unit 11 determines that the temperature history exceeds the acceptable range if the variation in the measured temperature, maximum temperature, cooling rate, interlayer temperature, or holding time exceeds a predetermined range. Alternatively, for example, the processing unit 11 determines that the temperature history exceeds an acceptable range if the measured temperature, maximum temperature, cooling rate, interlayer temperature, or holding time exceeds a predetermined threshold.

[0048] When it is determined that the predicted physical property value or the predicted temperature history does not exceed the allowable range (step S4, No), the prediction unit 12 outputs the determination result by the processing unit 11 to the shaping condition generation unit 6. Based on the determination result indicating that the physical property value or the temperature history does not exceed the allowable range, the shaping condition generation unit 6 outputs the shaping conditions generated in step S2 to the shaping unit 7. As a result, in step S8, the shaping unit 7 shapes the shaped object according to the shaping conditions output by the shaping condition generation unit 6.

[0049] On the other hand, when it is determined that the predicted physical property value or the predicted temperature history exceeds the allowable range (step S4, Yes), in step S5, the prediction unit 12 adjusts the shaping conditions. As an example, the prediction unit 12 expands the shaping result data or the thermal simulation data referred to when predicting the physical property value or the temperature history, and recalculates the shaping conditions. Alternatively, the prediction unit 12 resets the allowable range. Here, the adjustment of the shaping conditions is assumed to include expanding the shaping result data or the thermal simulation data to recalculate the shaping conditions and resetting the allowable range.

[0050] Next, in step S6, the processing unit 11 determines whether the physical property value predicted in step S3 or the temperature history predicted in step S3 is within the allowable range. When it is determined that the predicted physical property value or the predicted temperature history is not within the allowable range (step S6, No), the metal lamination shaping system 1 returns the procedure to step S5.

[0051] On the other hand, if it is determined that the predicted physical properties or predicted temperature history are within an acceptable range (step S6, Yes), in step S7, the prediction unit 12 outputs the molding conditions adjusted in step S5 to the molding condition generation unit 6. The molding condition generation unit 6 outputs the molding conditions input to the molding condition generation unit 6, i.e., the molding conditions adjusted in step S5, to the molding unit 7. As a result, in step S8, the molding unit 7 performs molding according to the molding conditions output by the molding condition generation unit 6. The metal additive manufacturing system 1 completes the operation according to the procedure shown in Figure 3 by completing step S8.

[0052] Next, we will explain the prediction of physical properties or temperature history by the prediction unit 12. Figure 4 is a diagram illustrating the relationship between the molding conditions and the physical properties of the molded object in Embodiment 1.

[0053] The physical properties of a fabricated object after fabrication are determined by its microstructure. The physical properties of a fabricated object can be predicted from its microstructure, for example, by the Hall-Petch law. For the physical properties of a fabricated object to be uniform regardless of location, the microstructure must be homogeneous. The microstructure is determined by the temperature history during fabrication, including the maximum temperature, cooling rate, and holding time. Therefore, the microstructure of a fabricated object can be predicted from its temperature history. Based on this, whether a fabricated object is homogeneous can be estimated from its temperature history. Furthermore, if the temperature history can be predicted, the physical properties of the fabricated object can be predicted. Moreover, the temperature history is determined by the composition of the fabricated object (the types and proportions of constituent elements), its shape, and the fabrication conditions. The temperature history can be predicted from the composition, shape, and fabrication conditions of the fabricated object, which are set before fabrication.

[0054] The molding performance data is data that associates the molding conditions and information of the molded object from past molding processes with the physical properties or temperature history at each of multiple locations on the molded object. Examples of physical properties included in the molding performance data include Vickers hardness.

[0055] To improve the accuracy of predictions made by the prediction unit 12, it is desirable to accumulate a large amount of past data as fabrication performance data. In order to accumulate a large amount of past data, it is desirable that the data to be accumulated be easily acquired. Metal structure data obtained by cross-sectional observation takes a long time to acquire, making it unsuitable for large-scale accumulation. In contrast, Vickers hardness data can be acquired from minute areas and is easier to acquire than metal structure data. Furthermore, with data obtained by tensile testing, the test piece is large, making it unclear which part of the test piece the data pertains to. In contrast, Vickers hardness data has the advantage that the measurement area is small, making it easy to correlate the fabrication point with the material properties.

[0056] Based on the above, it can be said that the Vickers hardness value is preferable as the material property value to be included in the molding performance data. However, the material property value to be included in the molding performance data is not limited to the Vickers hardness value; values ​​representing other material properties are also acceptable.

[0057] Furthermore, the printing performance data associates the printing conditions and printed object information from past printing processes with the monitoring data from the printing process. Temperature history is included in this monitoring data. In addition, the monitoring data may include data other than temperature history.

[0058] For example, the molding unit 7 is equipped with various sensors to monitor the molding process. The infrared thermometer 32 is one of these sensors. In Figure 2, the illustration of sensors other than the infrared thermometer 32 is omitted. The molding performance data includes data acquired by the various sensors. The data acquired by the various sensors includes, for example, data indicating the load during material supply at each molding point, data on the height of the molded object, or data on the width of the molten pool. This data is included in the molding performance data as time-series data. This data may also be associated with various command values ​​output by the control device 35.

[0059] Next, a specific example of the prediction of material properties by the prediction unit 12 will be explained. For example, the prediction unit 12 predicts material properties based on the manufacturing data by inputting the manufacturing conditions into a trained model, which is the result of learning the manufacturing data.

[0060] The learning unit 13 generates a trained model by machine learning, using the physical properties of each of the multiple locations in the molded object as the target variable. The prediction unit 12 uses the trained model to predict the physical properties of each of the multiple locations.

[0061] The memory unit 9 constructs a database of molding performance data from the data stored in the memory unit 9. The database stores data on material, molding shape, various molding conditions and molding paths, temperature history, and physical properties. The learning unit 13 reads the database from the memory unit 9. Using the read database, the learning unit 13 learns the relationship between material, molding shape, molding conditions, temperature history, and physical properties. The machine learning performed by the learning unit 13 can employ general-purpose algorithms such as neural networks, random forests, linear regression, or support vectors. Note that the algorithms used in machine learning are not limited to those exemplified here.

[0062] Next, a specific example of temperature history prediction by the prediction unit 12 will be explained. When physical properties are used as the target variable, it is difficult to obtain a large amount of data on physical properties because work is required to measure the physical properties. In contrast, since temperature history data can be obtained from the output of the radiation thermometer 32, it is possible to obtain a large amount of data more easily than in the case of physical property data. Furthermore, temperature history data can be obtained in real time while the fabrication of the object is taking place. As explained with reference to Figure 4, the physical properties of the fabricated object can be estimated from the temperature history, so it is considered that in order to make the physical properties of the fabricated object uniform, the temperature history should be kept constant.

[0063] For example, the prediction unit 12 predicts the temperature history based on the molding history data by inputting the molding conditions into a trained model, which is the result of learning from the molding history data. The learning unit 13 generates a trained model by machine learning with the temperature history for each of the multiple locations in the molded object as the target variable. The prediction unit 12 uses this trained model to predict the temperature history for each of the multiple locations.

[0064] The learning unit 13 reads the database from the storage unit 9, similar to the case where physical properties are the target variable. The learning unit 13 uses the read database to learn the relationship between the material, molded shape, and molding conditions and the temperature history. When temperature history is the target variable, a general-purpose algorithm can be used in machine learning, similar to the case where physical properties are the target variable. However, when predicting temperature history using a trained model, the accuracy of the prediction of the physical properties of the molded object is considered to be lower compared to when predicting physical properties using a trained model, since physical properties are not directly predicted. However, as mentioned above, when predicting temperature history using a trained model, a large amount of data can be easily obtained, so an improvement in accuracy can be expected by using a large amount of data.

[0065] Next, the tolerance range used for determination in the processing unit 11 will be explained. The above tolerance range can be set arbitrarily by the experimenter. For example, if characteristics conforming to a specific standard such as Japanese Industrial Standards (JIS) are desired, the tolerance range may be the range of variation or value that satisfies that standard. The stricter the tolerance range, the more possible it is to obtain molding conditions that enable the production of more homogeneous molded objects, but on the other hand, a smaller tolerance may result in more cases where molding is impossible. Also, a strict tolerance range may lead to the proposal of extreme molding conditions. For this reason, it is desirable that the tolerance range be set appropriately by the experimenter.

[0066] Next, the operation of each part of the metal additive manufacturing system 1 will be explained by referring to a specific example of a manufactured object. Here, we will explain the case in which the manufacturing unit 7 manufactures an object by point manufacturing. Point manufacturing is a manufacturing method in which material is supplied to a processing point while the processing point to which the heat source is supplied is stopped. The metal additive manufacturing apparatus 2 manufactures an object by point manufacturing by depositing layers consisting of point-shaped beads. Furthermore, below, we will use as an example the case in which a trained model is generated with material properties as the target variable based on manufacturing performance data, which is training data.

[0067] Figure 5 is the first diagram showing an example of an object manufactured by the metal additive manufacturing system 1 according to Embodiment 1. Figure 6 is the second diagram showing an example of an object manufactured by the metal additive manufacturing system 1 according to Embodiment 1. Figure 7 is the third diagram showing an example of an object manufactured by the metal additive manufacturing system 1 according to Embodiment 1.

[0068] Here, the object manufactured by the metal additive manufacturing system 1 is assumed to be the block shape shown in Figure 5. The material is a titanium alloy. In Figures 5 to 7, each of the multiple circles with halftone dots represents a dot bead. The object shown in Figure 5 is formed by arranging multiple dot beads, with 4 in the X direction, 4 in the Y direction, and 50 in the Z direction. x, y, and z represent the X, Y, and Z coordinates, respectively. Each dot bead is represented by the coordinates (x, y, z). Figure 6 shows the arrangement of dot beads in the X and Y directions. Figure 7 shows the arrangement of dot beads in the X and Z directions. The object information input unit 5 receives information indicating that it is a titanium alloy and information indicating that the shape of the object is as shown in Figures 5 to 7 as object information. It is known that differences in cooling rate easily affect the hardness of the solidified titanium alloy. Therefore, when the material is a titanium alloy, it can be said that changes in the metallic structure or physical properties are likely to occur due to differences in the temperature history inside the fabricated object.

[0069] In point-based printing, the order in which each point bead is formed can be arbitrarily selected in the CAM software. Furthermore, in point-based printing, various printing conditions can be adjusted for each printing point. Here, it is assumed that conditions such as laser power, laser irradiation time, and material supply rate can be appropriately selected for each printing point.

[0070] Here, we will explain an example of the procedure for fabricating the objects shown in Figures 5 to 7. First, a dot bead at (1,1,1) is fabricated, and by moving the processing point in the Y direction, fabrication is carried out up to (1,4,1), thereby fabricating the x=1 column. Next, the processing point is moved from (1,4,1) to (2,1,1), and the x=2 column is fabricated. After that, the x=3 and x=4 columns are also fabricated, and the z=1 layer is fabricated. From (4,4,1), where the last layer of z=1 was fabricated, the processing point is moved to (1,1,2), and the first dot bead of the z=2 layer is fabricated on top of (1,1,1). After that, the z=2 layer is fabricated in the same way as the z=1 layer. This fabrication process is repeated up to the z=50 layer, thereby fabricating the objects shown in Figures 5 to 7. The molding condition generation unit 6 generates molding conditions that include such molding paths.

[0071] The metal additive manufacturing system 1 uses manufacturing performance data from point-forming of the objects shown in Figures 5 to 7 to predict material properties and adjust manufacturing conditions. The manufacturing performance data includes data on the laser output during manufacturing and data on the pause time for each layer. The pause time for each layer refers to the time from when the manufacturing of one layer, z = N, is completed until the manufacturing of the next layer, z = N+1, begins. N is any integer.

[0072] The temperature history data stored in the memory unit 9 as manufacturing performance data is based on the measurement results from the radiation thermometer 32 shown in Figure 2. Since the radiation thermometer 32 is mounted coaxially with the laser beam in the metal additive manufacturing system 1, the temperature history of each manufacturing point can be acquired accurately and in real time. The temperature history includes information on several temperature-related items, such as the temperature of the processing point and the manufacturing point, the maximum temperature, the cooling rate, the interlayer temperature, and the holding time.

[0073] For example, when creating (1,1,N), the radiation thermometer 32 measures the temperature of (1,1,N-1), which corresponds to the base layer of (1,1,N). This temperature is the interlayer temperature of (1,1,N). Also, while creating (1,1,N), the radiation thermometer 32 measures the temperature of the processing point, (1,1,N). Since the material is molten when creating, the temperature of the processing point will be above its melting point. When the laser beam is irradiated, the processing point becomes high temperature. At this time, the radiation thermometer 32 measures the highest temperature. Alternatively, the time data for when the temperature is above the melting point may be acquired as one item of data in the temperature history.

[0074] When the laser beam irradiation is stopped, the temperature of the build point, which was the processing point, decreases over time. In the case of point buildup, the processing head 26 remains on the build point from the time the laser beam irradiation is stopped until the processing head 26 starts moving, so the radiation thermometer 32 can accurately measure the temperature change as the temperature of the build point decreases. The cooling rate is calculated by dividing the temperature difference from the maximum temperature to a predetermined temperature by the time required for the temperature to decrease from the maximum temperature to the predetermined temperature. The predetermined temperature can be arbitrarily selected depending on the material. For example, in order to understand the temperature change until the material solidifies, the solidification point can be selected as the predetermined temperature. In the case of titanium alloys, since phase transformation is observed, the temperature at which the phase transformation begins or the temperature at which the phase transformation ends may be selected as the predetermined temperature.

[0075] Once the cooling rate is calculated, the processing head 26 is moved to the next build point, and the temperature history is acquired for that next build point in the same manner as described above. By repeating this operation, the metal additive manufacturing system 1 can acquire the temperature history of each build point accurately and in real time.

[0076] In the above description, the metal additive manufacturing system 1 acquires temperature history using a radiation thermometer 32 mounted coaxially with the laser beam. However, the temperature history may also be acquired using a radiation thermometer 32 mounted at a position other than coaxial with the laser beam. The metal additive manufacturing system 1 may also use a thermometer mounted at a position other than coaxial with the laser beam to measure the temperature at a position other than the processing point while the manufacturing is being performed.

[0077] For example, the metal additive manufacturing system 1 may acquire images showing the temperature history of multiple build points by placing a thermoviewer. This allows the metal additive manufacturing system 1 to acquire the temperature history of any build point. Furthermore, the metal additive manufacturing system 1 can acquire the temperature of build points other than the processing point while the build process is underway. For example, in the case of materials that undergo age precipitation, the holding time in the temperature range where age precipitation occurs after build is important in determining the physical properties of the build point. In such cases, the metal additive manufacturing system 1 may acquire the temperature of build points other than the processing point in order to obtain data on the holding time in the temperature range where age precipitation occurs.

[0078] The metal additive manufacturing system 1 acquires temperature history by performing manufacturing under each of multiple manufacturing conditions. In addition, physical properties are measured for the manufactured object. For example, the manufactured object is cut, and the Vickers hardness at each manufacturing point on the cut surface is measured. In this way, the metal additive manufacturing system 1 acquires data on physical properties at each of the multiple manufacturing points and data on the temperature history at each of the multiple manufacturing points.

[0079] The data for material, molded shape, molding conditions, temperature history, and physical properties are compiled into a database and stored in the storage unit 9. The molding conditions include various conditions during molding and the molding path. The molding path data is stored as numerical data related to the molding path. One example of numerical data related to the molding path is the time data from the previous molding point. The time from the previous molding point is defined as the sum of the movement time of the processing head 26 from the previous molding point (where the molding point immediately preceding the current molding point is defined as the previous molding point) to the current molding point, and the rest time at the previous molding point.

[0080] Figure 8 shows an example of a database held in the storage unit 9 in the metal additive manufacturing system 1 according to Embodiment 1. In Figure 8, "No." represents the number assigned to each build point. "x", "y", and "z" represent the x, y, and z coordinates of the build point, respectively. "Output" represents the laser output. "Hardness" represents the Vickers hardness. In the example shown in Figure 8, the values ​​for each item—build point number, x, y, and z coordinates, material, laser output, time since the previous build point, rest time, maximum temperature, interlayer temperature, cooling rate, and Vickers hardness—are associated with each other for each build point.

[0081] Note that the items that can be included in the database are not limited to those exemplified in Figure 8. The database may exclude one or more of the items shown in Figure 8. Alternatively, the database may include one or more items other than those shown in Figure 8.

[0082] The learning unit 13 reads the above-mentioned fabrication performance data, which is stored in a database. Based on the fabrication performance data, which is the learning data, the learning unit 13 generates a trained model with Vickers hardness, a material property value, as the target variable. The explanatory variables of the trained model are the values ​​of x-coordinate, y-coordinate, z-coordinate, material, laser output, time from the previous fabrication point, rest time, maximum temperature, interlayer temperature, and cooling rate. The items to be used as explanatory variables are not limited to those exemplified here. One or more of the items exemplified here may be excluded from the explanatory variables. Alternatively, one or more items other than those exemplified here may be included in the explanatory variables. For example, the explanatory variables may include the melting point of the material, the material supply rate, or the laser beam irradiation time.

[0083] The machine learning performed by the learning unit 13 can employ general-purpose algorithms such as neural networks, random forests, linear regression, or support vectors. Here, the learning unit 13 will perform machine learning using a neural network.

[0084] When the prediction unit 12 predicts material properties, various conditions and a build path, which are build conditions for the build for which material properties are to be predicted, are entered into the database in the storage unit 9. The build conditions entered into the database here are the initial build conditions generated by the build condition generation unit 6. Initial build conditions refer to build conditions that have not been adjusted by the calculation unit 10. The prediction unit 12 reads the build conditions entered into the database and inputs these build conditions into the trained model to obtain material properties for each build point.

[0085] Suppose that by determining the physical properties for each build point, it is found that, for example, in each layer, the Vickers hardness is lower in the part built towards the end (4,1,N) to (4,4,N) than in the part built initially (1,1,N) to (1,4,N). Furthermore, the variation in Vickers hardness across layers is assumed to be 30 HV. Such differences in physical properties across layers are thought to be due to heat accumulation during build, and as build progresses, the interlayer temperature rises, the cooling rate decreases, or the maximum temperature rises significantly.

[0086] The processing unit 11 determines whether the physical property value predicted by the prediction unit 12 exceeds the allowable range. Here, for example, suppose the allowable range is set to "10 HV or less," which is the range of variation in Vickers hardness. In the above case, the variation in Vickers hardness is 30 HV, which exceeds the set allowable range, so the processing unit 11 determines that the physical property value exceeds the allowable range.

[0087] The prediction unit 12 determines, based on the determination result by the processing unit 11, that a homogeneous object cannot be fabricated under the initial fabrication conditions described above. Alternatively, the prediction unit 12 may reset the tolerance range. By resetting the tolerance range, the prediction unit 12 changes, for example, the tolerance range of Vickers hardness variation being 10 HV or less to a tolerance range of Vickers hardness of 350 HV or more at all fabrication points. In this case, the prediction unit 12 outputs the information of the reset tolerance range to the processing unit 11. The processing unit 11 determines, based on the reset tolerance range, whether the physical properties satisfy the tolerance range.

[0088] The prediction unit 12 adjusts the molding conditions if it determines that a homogeneous object cannot be fabricated with the initial molding conditions described above. The prediction unit 12 adjusts the molding conditions by adjusting at least one of the various conditions and the molding path. Possible methods for adjusting the molding conditions include empirically selecting the contents of the various conditions or the molding path. Alternatively, the prediction unit 12 may adjust the molding conditions by using a trained model. In this case, the prediction unit 12 adjusts the molding conditions by predicting the molding conditions at which the material properties obtained using the trained model fall within an acceptable range.

[0089] For example, the prediction unit 12 sets a loss function for the trained model generated based on the actual manufacturing data in the memory unit 9, as described above. Here, the prediction unit 12 sets a loss function such that the variation in Vickers hardness is 10 HV. The prediction unit 12 adjusts the manufacturing conditions by finding manufacturing conditions that conform to this loss function. If the variation in material properties is the cause of heat accumulation during manufacturing, one way to adjust the manufacturing conditions is to decrease the laser output as the height in the Z direction of the manufacturing position increases. Other adjustments include changing the order of manufacturing at each manufacturing point within a layer, or increasing the rest time between layers.

[0090] For example, after the prediction unit 12 adjusts the molding conditions, suppose the prediction unit 12 predicts the Vickers hardness based on the adjusted molding conditions and the variation in Vickers hardness is 8 HV. In this case, since the variation in Vickers hardness falls within the acceptable range of 10 HV or less, the prediction unit 12 determines that a homogeneous object can be fabricated with these adjusted molding conditions. The prediction unit 12 outputs these adjusted molding conditions to the molding condition generation unit 6. On the other hand, if the variation in Vickers hardness predicted based on the adjusted molding conditions still does not satisfy the acceptable range, the prediction unit 12 resets the acceptable range. The processing unit 11 then determines whether the physical properties satisfy the acceptable range based on the reset acceptable range.

[0091] When the molding conditions adjusted by the prediction unit 12 are input to the molding condition generation unit 6, the molding condition generation unit 6 reflects the molding conditions on the CAM in order to output the molding conditions to the molding unit 7. The molding condition generation unit 6 confirms on the CAM that the molding unit 7 can operate without problems with the molding conditions input to the molding condition generation unit 6. Since the molding conditions input to the molding condition generation unit 6 through adjustment by the calculation unit 10 are optimized molding conditions for predicting characteristics, it is necessary to confirm on the CAM that there are no problems in the operation of the molding unit 7.

[0092] The molding condition generation unit 6 outputs molding conditions to the molding unit 7 when it is confirmed on the CAM that the molding unit 7 can operate without problems based on the molding conditions input to the molding condition generation unit 6. The molding unit 7 then performs molding according to the molding conditions input to the molding unit 7, i.e., the adjusted molding conditions.

[0093] As a result, the metal additive manufacturing system 1 is capable of producing homogeneous objects. Furthermore, in order to confirm the accuracy of the manufacturing conditions adjusted as described above, the Vickers hardness of the manufactured object may be measured and the acquired data stored in the storage unit 9. This makes it possible to predict material properties with even greater accuracy in subsequent manufacturing processes.

[0094] According to Embodiment 1, the metal additive manufacturing system 1 includes: an object information input unit 5 into which object information is input; a molding condition generation unit 6 that generates molding conditions based on the object information; a calculation unit 10 that predicts physical property values ​​or temperature history for each of a plurality of positions of the object when the object indicated in the object information is manufactured according to the molding conditions; a processing unit 11 that determines whether the predicted results of the physical property values ​​or temperature history fall within a set tolerance range; and a molding unit 7 that manufactures the object according to the molding conditions adjusted based on the determination result by the processing unit 11. The metal additive manufacturing system 1 can manufacture homogeneous objects by predicting the characteristics of each of a plurality of positions of the object and adjusting the molding conditions.

[0095] Furthermore, the metal additive manufacturing system 1 includes a storage unit 9 that stores manufacturing performance data, which is data in which the manufacturing conditions and manufactured object information from past manufacturing processes are associated with physical properties or temperature history at each of multiple locations, or thermal simulation data, which is the result of simulating the temperature history at each of multiple locations. The calculation unit 10 predicts physical properties or temperature history based on the manufacturing performance data or thermal simulation data. By predicting physical properties or temperature history based on the manufacturing performance data or thermal simulation data stored in the storage unit 9, the metal additive manufacturing system 1 can make more accurate predictions in the prediction unit 12 as the amount of information stored increases.

[0096] Furthermore, the calculation unit 10 predicts physical properties or temperature history based on the manufacturing performance data by inputting the manufacturing conditions into a trained model, which is the result of learning from the manufacturing performance data. The trained model is generated by machine learning. Neural networks, random forests, linear regression, or support vectors are used for machine learning. As a result, the metal additive manufacturing system 1 can predict physical properties or temperature history based on manufacturing performance data that shows the results of past manufacturing.

[0097] Furthermore, the molding unit 7 fabricates the object by point molding, supplying material to the processing point while the processing point to which the heat source is supplied is stopped. The calculation unit 10 predicts the physical properties or temperature history using a trained model generated by machine learning, with the physical properties or temperature history of each position of the object as the objective variable and the molding conditions as the explanatory variable. In point molding, the temperature history associated with the coordinates can be obtained for each of the multiple positions in the object. As a result, the metal additive manufacturing system 1 can predict the characteristics of each of the multiple positions in the object.

[0098] Embodiment 2. Embodiment 2 describes in detail the case in which a trained model is generated using the metal additive manufacturing system 1 shown in Figure 1, based on manufacturing performance data which is training data, with temperature history as the target variable. In Embodiment 2, the same content as in Embodiment 1, where physical properties are used as the target variable, will be omitted from the explanation. In Embodiment 2, as in Embodiment 1, the manufacturing unit 7 will manufacture the object by point manufacturing.

[0099] The learning unit 13 reads the above-mentioned fabrication performance data, which is stored in a database. Based on the fabrication performance data, which is the learning data, the learning unit 13 generates a trained model with temperature history as the target variable. The explanatory variables of the trained model are the values ​​of x-coordinate, y-coordinate, z-coordinate, material, laser output, time from the previous fabrication point, rest time, maximum temperature, interlayer temperature, and cooling rate. The items to be used as explanatory variables are not limited to those exemplified here. One or more of the items exemplified here may be excluded from the explanatory variables. Alternatively, one or more items other than those exemplified here may be included in the explanatory variables. For example, the explanatory variables may include the melting point of the material, the material supply rate, or the laser beam irradiation time.

[0100] The machine learning performed by the learning unit 13 can employ general-purpose algorithms such as neural networks, random forests, linear regression, or support vectors. Here, the learning unit 13 will perform machine learning using a neural network.

[0101] When the prediction unit 12 predicts the temperature history, various conditions and the printing path, which are the printing conditions for the printing process for which the temperature history is to be predicted, are entered into the database in the storage unit 9. The temperature history predicted by the prediction unit 12 consists of at least one of the following: cooling rate, maximum temperature, interlayer temperature, and holding time.

[0102] The processing unit 11 determines whether the physical property value predicted by the prediction unit 12 exceeds the acceptable range. Here, for example, suppose that a cooling rate of 180°C / sec or higher is set as the acceptable range. The processing unit 11 determines whether the cooling rate predicted by the prediction unit 12 is outside the acceptable range. The operation after this determination by the processing unit 11 is the same as in Embodiment 1 when the physical property value is the target variable.

[0103] According to Embodiment 2, the metal additive manufacturing system 1 generates a trained model with temperature history as the target variable, and from the predicted temperature history, it is possible to determine the manufacturing conditions necessary to obtain a homogeneous object. However, when predicting temperature history using a trained model, the material properties are not directly predicted, so the accuracy of predicting the material properties of the object is considered to be inferior to when predicting material properties using a trained model. However, as described in Embodiment 1, when predicting temperature history using a trained model, a large amount of data can be easily acquired, so an improvement in accuracy can be expected by using a large amount of data. In addition, temperature history data for each manufacturing point can be acquired in real time during manufacturing. Therefore, when acquiring temperature history data, there is an advantage in that it does not take as much time to acquire data compared to acquiring material property data.

[0104] Embodiment 3. Embodiment 3 describes a case in which even more accurate predictions are possible by utilizing temperature history data that can be acquired in real time during molding. In Embodiment 3, the same reference numerals are used for the same components as in Embodiment 1 or 2, and the differences from Embodiment 1 or 2 are mainly described. In Embodiment 3, as in Embodiment 1 or 2, the molding unit 7 performs molding of the object by point molding.

[0105] Embodiment 2 describes a case in which the temperature history is predicted. The predicted temperature history may deviate from the predicted temperature history if an unexpected abnormality occurs during the molding process. For example, if the wire 21 is supplied in a bent state during molding, the amount of material supplied to the processing point may be less than the amount previously assumed. If powder is used as the material, the amount of material supplied to the processing point may be reduced due to powder clogging. The problems described here are just examples, and various problems can occur during molding.

[0106] For these reasons, while the prediction results before fabrication are suitable for determining the general direction, it is difficult to guarantee sufficient accuracy. Therefore, in Embodiment 3, we will utilize the temperature history information acquired during fabrication to perform feedback control during the fabrication process.

[0107] Figure 9 shows an example of the configuration of a metal additive manufacturing system 1A according to Embodiment 3. The metal additive manufacturing system 1A comprises a metal additive manufacturing apparatus 2 and an information processing apparatus 3, which have the same configuration as the metal additive manufacturing system 1 shown in Figure 1. Furthermore, the metal additive manufacturing system 1A is equipped with a sensor 4. The sensor 4 functions as a monitor unit that monitors the temperature history when the manufactured object is fabricated. The sensor 4 is a device that measures the temperature of the manufactured object, and is, for example, a radiation thermometer.

[0108] Furthermore, the molding unit 7 includes a molding command unit that modifies the molding conditions for the molded object based on the result of comparing the temperature history acquired by the sensor 4, which is the monitoring unit, with the temperature history predicted by the calculation unit 10. In Embodiment 3, the control device 35 within the molding unit 7 is provided with the functions of the molding command unit. The molding command unit may be implemented using an external device to the control device 35. This device may be provided inside the molding unit 7 or outside the molding unit 7.

[0109] Next, the operation of the metal additive manufacturing system 1A will be described. Figure 10 is a flowchart showing an example of the operation procedure of the metal additive manufacturing system 1A according to Embodiment 3.

[0110] In step S11, the molding unit 7 begins molding the object. Before starting to mold the object, the metal additive manufacturing system 1A performs the same operations as in steps S1 to S7 shown in Figure 3. As a result, the molding unit 7 molds the object according to the initial molding conditions generated by the molding condition generation unit 6, or according to the molding conditions adjusted by the calculation unit 10.

[0111] The control device 35 acquires temperature history data predicted by the prediction unit 12 while the object is being fabricated. The control device 35 also acquires measurement results from the sensor 4 and calculates the temperature history based on the acquired measurement results to obtain actual temperature history values.

[0112] In step S12, the control device 35 determines whether the measured temperature history deviates from the predicted temperature history. For example, the control device 35 calculates the difference between the measured temperature history and the predicted temperature history, and determines that the measured temperature history deviates from the predicted temperature history if the difference exceeds a preset threshold. Note that the method for determining whether the measured temperature history deviates from the predicted temperature history is not limited to the method described here and is arbitrary. If it is determined that the measured temperature history does not deviate from the predicted temperature history (step S12, No), the metal additive manufacturing system 1A proceeds to step S15.

[0113] On the other hand, if it is determined that the measured temperature history deviates from the predicted temperature history (step S12, Yes), in step S13, the control device 35 modifies the molding conditions. For example, the control device 35 calculates the predicted value by inputting the measured temperature history into the learned model.

[0114] In step S14, the molding unit 7 continues to fabricate the object based on the fabrication conditions modified in step S13. In step S15, the molding unit 7 determines whether or not to terminate the fabrication of the object. If the fabrication of the object is not terminated (step S15, No), the metal additive manufacturing system 1A returns to step S12. On the other hand, if the fabrication of the object is terminated (step S15, Yes), the metal additive manufacturing system 1A terminates the operation according to the procedure shown in Figure 10. According to the procedure shown in Figure 10, by repeatedly modifying the fabrication conditions in step S13, the discrepancy between the predicted temperature history and the measured temperature history at each fabrication point can be reduced.

[0115] According to Embodiment 3, the molding unit 7 includes a molding command unit that modifies the molding conditions for the molded object based on the result of comparing the temperature history acquired by the monitoring unit with the temperature history predicted by the calculation unit 10. As a result, the metal additive manufacturing system 1A can predict the characteristics of the molded object with high accuracy, thereby enabling the production of even more homogeneous molded objects.

[0116] Embodiment 4. In Embodiments 1 to 3, in the molding performance data used to predict physical properties or temperature history, the molding conditions, temperature history, and physical property data were associated with coordinates indicating the position on the molded object. Embodiment 4 describes a case in which, in the molding performance data, the molding conditions, temperature history, and physical property data are associated with the sum of the heat transfer indices at each of multiple positions on the molded object. The heat transfer index is defined as the amount of heat uniformly transferred from an arbitrary point (any point on the molded object) to an adjacent point (an adjacent point), assuming that heat is uniformly transferred from that arbitrary point to an adjacent point. Here, "point" refers to the molding point. Strictly speaking, the heat transfer is a non-steady state and heat is not transferred uniformly, but here, in order to simply consider the ease of heat transfer, we assume that heat is uniformly transferred to adjacent points, and the purpose is to grasp the trend of the magnitude of the heat transfer index depending on the number of adjacent points.

[0117] The operation described in Embodiment 4 can be realized with the same configuration as the metal additive manufacturing system 1 shown in Figure 1. Embodiment 4 mainly describes an operation that differs from Embodiments 1 to 3. In Embodiment 4, as in Embodiments 1 to 3, the manufacturing unit 7 will manufacture the object by point manufacturing.

[0118] When data on molding conditions, temperature history, and material properties are mapped to coordinates, it becomes necessary to accumulate molding data every time the shape of the molded object changes. In contrast, when data on molding conditions, temperature history, and material properties are mapped to the sum of heat transfer indices, it becomes possible to reuse molding data in various cases where the shape of the molded object differs.

[0119] In Embodiment 4, the data held by the storage unit 9 includes molding performance data or thermal simulation data. The molding performance data is data that associates the molding conditions and molded object information from past molding with the sum of the heat transfer indices from an arbitrary point (each of a plurality of positions) to adjacent points adjacent to that arbitrary point in past molding. The thermal simulation data is the simulation result of the sum of the heat transfer indices. Thus, the storage unit 9 stores either molding performance data, which is data that associates the molding conditions and molded object information from past molding with the sum of the heat transfer indices from an arbitrary point (each of a plurality of positions) to adjacent points adjacent to that arbitrary point in past molding, or thermal simulation data, which is the simulation result of the sum of the heat transfer indices. The calculation unit 10 predicts the physical properties or temperature history of an arbitrary point based on the molding performance data or thermal simulation data.

[0120] Here, with reference to Figures 6, 11, and 12, the manner of heat transfer in a manufactured object produced by the metal additive manufacturing system 1 will be described. Figure 11 is the first diagram illustrating the manner of heat transfer in a manufactured object produced by the metal additive manufacturing system 1 according to Embodiment 4. Figure 12 is the second diagram illustrating the manner of heat transfer in a manufactured object produced by the metal additive manufacturing system 1 according to Embodiment 4.

[0121] Here, the object with the shape shown in Figures 5 to 7 is referred to as the first object. Figure 6 shows the arrangement of dot beads in the X and Y directions of the first object. In the first object, four dot beads are arranged in the X direction and four in the Y direction.

[0122] Figure 11 shows the arrangement of dot beads in the X and Y directions for a second object, which is different from the first object. In the second object, six dot beads are arranged in the X direction and six in the Y direction. Figure 12 shows the arrangement of dot beads in the X and Y directions for a third object, which is different from the first and second objects. In the third object, each row x=1 and x=2 has six dot beads arranged in the Y direction. In the third object, each row x=3, x=4, x=5 and x=6 has two dot beads arranged in the Y direction.

[0123] It is empirically known that in block-shaped objects, heat accumulation at each corner of the block is significantly greater than at other locations. In the 4x4 block shape of the first object shown in Figure 6, heat accumulation is high at the corners (1,1), (4,1), (1,4), and (4,4) in the XY plane. This is because the number of adjacent point beads in the X and Y directions of the point beads at each corner is smaller than that of point beads at other locations, making it more difficult for heat to be conducted to the surroundings.

[0124] In the 6x6 block shape of the second fabricated object shown in Figure 11, the three points (4,1), (1,4), and (4,4) are not corners. In the case of the third fabricated object shown in Figure 12, the two points (4,1) and (1,4) are not corners, and there is no point bead at (4,4). From this, it can be seen that the heat transfer patterns for each coordinate are completely different for the first, second, and third fabricated objects, which are all different fabricated objects. If machine learning is performed using fabrication data obtained from fabricated objects of various shapes without considering that the heat transfer patterns for each coordinate are completely different, it will not be possible to generate a trained model that can accurately predict material properties or temperature history.

[0125] Therefore, in Embodiment 4, instead of using data associated with coordinates as the actual build data, data associated with the heat transfer index of each build point is used as the actual build data. The heat conduction of a built object changes moment by moment depending on the position and time, and can be said to be non-steady. For this reason, rigorous thermal analysis of the built object is difficult.

[0126] Here, we do not strictly consider the changes in the heat transfer index in the fabricated object, and assume that heat is conducted uniformly from any point. The amount of heat conducted from any point to one point adjacent to that point is defined as the heat transfer index q. Furthermore, the sum of the heat conducted from any point to all points surrounding that point is defined as the sum of the heat transfer indices Q. According to this definition, the fewer points surrounding an arbitrary point, the lower the Q of that point, and the less heat is transferred there. In the case of a 4x4 block shape, the corners (1,1), (4,1), (1,4), and (4,4) have lower Qs than other positions, and can be said to be prone to high heat storage. In Embodiment 4, the heat transfer index data is data that represents the trend of heat transfer and is included in the fabrication performance data, and does not need to be strictly accurate data.

[0127] Here, adjacent points refer to adjacent points in three dimensions. That is, adjacent points include adjacent points in the X, Y, and Z directions, as well as adjacent points in diagonal directions between these directions. For any given point, there can be up to 26 adjacent points. Furthermore, since the heat transfer index depends on distance, it is possible to obtain a more realistic Q by using the value obtained by dividing the heat transfer index by the distance between the points.

[0128] In this way, the metal additive manufacturing system 1 can accumulate manufacturing performance data obtained from various shapes of objects by using the sum of the heat transfer indices Q as an explanatory variable, rather than using coordinate data. In other words, the metal additive manufacturing system 1 can use the manufacturing performance data not as data for a specific shape, but as data for various shapes in a general sense. Furthermore, the number of adjacent points for each manufacturing point of an object can be counted before manufacturing. For this reason, the sum of the heat transfer indices Q can be easily used as an explanatory variable.

[0129] Figure 13 is a first diagram illustrating a method for calculating the sum of heat transfer indices in the metal additive manufacturing system 1 according to Embodiment 4. Figure 14 is a second diagram illustrating a method for calculating the sum of heat transfer indices in the metal additive manufacturing system 1 according to Embodiment 4.

[0130] Figures 13 and 14 show the construction process of a block-shaped object. Figure 13 shows the arrangement of dot beads in the X and Y directions of the object being constructed. Figure 14 shows the arrangement of dot beads in the Y and Z directions of the same object as in Figure 13. In Figures 13 and 14, circles with dark halftones represent dot beads at z = N-2. Circles with hatched diagonal lines represent dot beads at z = N-1. Circles with light halftones represent dot beads at z = N.

[0131] Here, let's consider (1, 3, N) as an arbitrary point. In the X, Y, or Z directions, there are two build points adjacent to this arbitrary point: (1, 2, N) and (1, 3, N-1). In the diagonal directions of the XY, YZ, or XZ planes, there are four build points adjacent to this arbitrary point: (2, 2, N), (1, 2, N-1), (2, 3, N-1), and (1, 4, N-1). In the diagonal directions on the front or back of the paper in Figure 13, or on the front or back of the paper in Figure 14, there are two build points adjacent to this arbitrary point: (2, 2, N-1) and (2, 4, N-1).

[0132] For any point adjacent to a build point in the X, Y, or Z direction, a heat transfer index q is assumed to be transferred from that point. In this case, the sum of the heat transfer indices Q at the arbitrary point (1, 3, N) is expressed by the following formula: Q = 2 × q + 4 × q × 2 -(1 / 2) +2 × q × 3 -(1 / 2)

[0133] According to Embodiment 4, the storage unit 9 stores printing performance data, which is data associated with the printing conditions and printed object information from past printing processes, and the sum of heat transfer indices from each of multiple locations in past printing processes to adjacent points adjacent to that point, or thermal simulation data, which is the simulation result of the sum of heat transfer indices. The calculation unit 10 predicts the physical properties or temperature history of the arbitrary point based on the printing performance data or thermal simulation data. The metal additive manufacturing system 1 can use the printing performance data not as data for a specific shape, but as data for various shapes in a general sense. As a result, the metal additive manufacturing system 1 can easily predict the characteristics of each of multiple locations of the printed object.

[0134] Embodiment 5. Embodiment 5 describes a case in which the molding unit 7 fabricates an object by linear molding. Linear molding is a fabrication method in which material is supplied to a processing point while moving the processing point to which a heat source is supplied. The operation described in Embodiment 5 can be realized with the same configuration as the metal additive manufacturing system 1 shown in Figure 1. Embodiment 5 mainly describes an operation that differs from Embodiments 1 to 4.

[0135] In linear fusion, the processing head 26 moves continuously, causing the radiation thermometer 32, which is coaxial with the processing point, to move continuously as well. The metal additive manufacturing system 1 can measure the maximum temperature at each printing point using the radiation thermometer 32, but it cannot obtain the interlayer temperature or cooling rate in the same way as in point fusion. The metal additive manufacturing system 1 can determine the interlayer temperature at the starting point where fusion begins and the cooling rate at the ending point where fusion ends for a linear bead using the radiation thermometer 32. In this case, it is not possible to determine the interlayer temperature and cooling rate between the starting and ending points as a temperature history.

[0136] Therefore, in Embodiment 5, the metal additive manufacturing system 1 acquires the temperature history using a radiation thermometer 32 mounted at a position other than coaxial with the laser beam. The calculation unit 10 predicts the physical properties or temperature history using a trained model generated by machine learning, with the physical properties or temperature history at each position of the manufactured object as the objective variable and the manufacturing conditions as the explanatory variable.

[0137] Alternatively, the metal additive manufacturing system 1, when using physical properties as the objective variable as in Embodiment 1, may predict physical properties without using temperature history as an explanatory variable. In this case, the calculation unit 10 predicts physical properties using a trained model generated by machine learning, where the physical properties at each position of the manufactured object are the objective variable and the manufacturing conditions are the explanatory variables. The metal additive manufacturing system 1 may also perform predictions while utilizing data from point manufacturing in order to perform machine learning based on a large amount of manufacturing performance data. For example, by narrowing the spacing between manufacturing points in point manufacturing and considering it as zero, theoretically, the data from point manufacturing becomes the same as the data from line manufacturing. The metal additive manufacturing system 1 may also utilize the data from point manufacturing, when the spacing between manufacturing points in point manufacturing is narrowed and considered as zero, for predictions in line manufacturing.

[0138] According to Embodiment 5, the molding unit 7 fabricates an object by linear molding. The calculation unit 10 predicts the physical properties or temperature history of each position of the object using a trained model generated by machine learning, with the physical properties or temperature history of each position of the object as the objective variable and the molding conditions as the explanatory variable. As a result, the metal additive manufacturing system 1 can predict the characteristics of each of the multiple positions of the object when performing linear molding.

[0139] Embodiment 6. Embodiment 6 describes the improvement of prediction accuracy when performing linear lithography. In Embodiment 5, the temperature history was not fully utilized, and a large amount of data was required to improve prediction accuracy. In Embodiment 6, the metal additive manufacturing system 1 supplements the temperature history data with thermal simulation data of linear lithography in order to utilize the temperature history. The operation described in Embodiment 6 can be realized with the same configuration as the metal additive manufacturing system 1 shown in Figure 1. Embodiment 6 mainly describes operations that differ from Embodiments 1 to 5.

[0140] The metal additive manufacturing system 1 does not obtain all of the temperature history used for predicting properties from thermal simulation data, but rather uses corrected thermal simulation data, which is thermal simulation data corrected based on measured data. The calculation unit 10 creates corrected thermal simulation data by correcting the temperature history of each position shown in the thermal simulation data based on the temperature history of the starting point where manufacturing begins and the ending point where manufacturing ends among the layers that make up the manufactured object. The calculation unit 10 predicts physical properties or temperature history based on the corrected thermal simulation data. By correcting the thermal simulation data, the metal additive manufacturing system 1 can correct the difference between the actual values ​​during manufacturing and the data obtained by simulation.

[0141] As described in Embodiment 4, the heat conduction of a fabricated object changes constantly with position and time, and can be said to be non-steady. It is difficult to precisely simulate the temperature history of a fabricated object. For this reason, thermal simulation data is calculated with the aim of finding trends in the temperature history by making the mesh, which is the smallest unit for data collection, coarser, or by simplifying the calculation. The calculated thermal simulation data is corrected by comparing it with the acquired temperature history or past fabrication performance data. The metal additive manufacturing system 1 predicts the temperature history or material properties based on the thermal simulation data corrected in this way.

[0142] According to Embodiment 6, the metal additive manufacturing system 1 can perform machine learning by supplementing the temperature history data using thermal simulation data of the manufacturing process, thereby supplementing the temperature history between the start and end points of the line bead. This allows the metal additive manufacturing system 1 to improve the accuracy of predicting properties.

[0143] In addition, in embodiments 1 to 5, the metal additive manufacturing systems 1 and 1A may also use thermal simulation data to supplement the temperature history data.

[0144] Next, the hardware that realizes the information processing device 3 will be described. The information processing device 3 is realized by using a processing circuit. The processing circuit is a circuit in which the processor executes software. The processing circuit is, for example, the control circuit shown in Figure 15.

[0145] Figure 15 shows an example configuration of a control circuit 40 according to embodiments 1 to 6. The control circuit 40 comprises an input unit 41, a processor 42, a memory 43, and an output unit 44. The input unit 41 is an interface circuit that receives data from outside the control circuit 40 and provides it to the processor 42. The output unit 44 is an interface circuit that sends data from the processor 42 or the memory 43 to the outside of the control circuit 40.

[0146] The arithmetic unit 10 and the processing unit 11 are implemented by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in memory 43. The processing circuit implements the arithmetic unit 10 and the processing unit 11 by having the processor 42 read and execute the program stored in memory 43. In other words, the processing circuit includes memory 43 for storing the program that will ultimately be executed as a result of the processing of the information processing device 3. The program stored in memory 43 can also be said to be a program that causes the computer to execute the procedures and methods of the information processing device 3. The data input unit 8 is implemented using the input unit 41. The storage unit 9 is implemented using memory 43. Memory 43 is also used as temporary memory when the processor 42 executes various processes.

[0147] The processor 42 is a CPU (Central Processing Unit). The processor 42 may also be a central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, processor, or DSP (Digital Signal Processor). The memory 43 may be, for example, a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM® (Electrically Erasable Programmable Read Only Memory), magnetic disk, flexible disk, optical disk, compact disk, minidisc, or DVD (Digital Versatile Disc).

[0148] The object information input unit 5 and the molding condition generation unit 6 of the metal additive manufacturing apparatus 2 are implemented with a configuration similar to the hardware configuration shown in Figure 15. The control device 35 within the molding unit 7 of the metal additive manufacturing apparatus 2 is implemented with a configuration similar to the hardware configuration shown in Figure 15.

[0149] The processing circuits of the information processing device 3 and the metal additive manufacturing apparatus 2 may be dedicated circuits. Dedicated circuits include single circuits, composite circuits, programmed processors, parallel programmed processors, ASICs (Application Specific Integrated Circuits), FPGAs (Field Programmable Gate Arrays), or circuits combining these.

[0150] Each component of the metal additive manufacturing systems 1 and 1A does not need to be physically configured as shown in the figures. The specific forms of dispersion and integration of each component are not limited to those shown. Each component may be functionally or physically dispersed in any unit, or it may be integrated. For example, the calculation unit 10 and processing unit 11 shown in Figure 1 may be realized by a single component. Alternatively, the calculation unit 10 may be incorporated into the molding condition generation unit 6 shown in Figure 1.

[0151] The configurations shown in each of the embodiments described above are examples of the content of this disclosure. The configurations of each embodiment can be combined with other known technologies. The configurations of each embodiment may be combined with each other as appropriate. It is possible to omit or modify parts of the configurations of each embodiment without departing from the gist of this disclosure.

[0152] The various aspects of this disclosure are summarized below as an appendix.

[0153] (Note 1) A metal additive manufacturing system comprising: an object information input unit into which object information indicating the material and shape of the object is input; a molding condition generation unit that generates molding conditions for the object based on the object information; a calculation unit that predicts physical property values ​​indicating the characteristics of the object at each of a plurality of positions of the object or the temperature history of each of the plurality of positions when the object indicated in the object information is manufactured according to the molding conditions; a processing unit that determines whether the predicted results of the physical property values ​​or the temperature history fall within a set tolerance range; and a molding unit that manufactures the object according to the molding conditions adjusted based on the result of the determination by the processing unit. (Note 2) The metal additive manufacturing system according to Note 1, comprising a storage unit that stores manufacturing performance data, which is data associated with the manufacturing conditions and manufactured object information in past manufacturing processes, and the physical properties or temperature history at each of the multiple positions in the past manufacturing process, or thermal simulation data, which is the result of simulating the temperature history at each of the multiple positions, wherein the calculation unit predicts the physical properties or temperature history based on the manufacturing performance data or the thermal simulation data. (Note 3) The metal additive manufacturing system according to Note 1, comprising a storage unit that stores manufacturing performance data, which is data associated with the manufacturing conditions and manufactured object information in past manufacturing processes, and the sum of heat transfer indices from an arbitrary point, which is each of the multiple positions in the past manufacturing process, to an adjacent point adjacent to that arbitrary point, or thermal simulation data, which is the result of simulating the sum of heat transfer indices, wherein the calculation unit predicts the physical properties or temperature history at the arbitrary point based on the manufacturing performance data or the thermal simulation data. (Note 4) The metal additive manufacturing system according to Note 2 or 3, characterized in that the calculation unit predicts the physical properties or temperature history based on the manufacturing performance data by inputting the manufacturing conditions into a trained model which is the result of learning the manufacturing performance data.(Note 5) The metal additive manufacturing system according to Note 4, characterized in that the molding unit manufactures the object by point molding, supplying material to the processing point while the processing point to which the heat source is supplied is stopped, and the calculation unit predicts the physical properties or temperature history using a trained model generated by machine learning, in which the physical properties or temperature history of each position of the object are the objective variables and the molding conditions are the explanatory variables. (Note 6) The metal additive manufacturing system according to Note 4, characterized in that the molding unit manufactures the object by line molding, supplying material to the processing point while moving the processing point to which the heat source is supplied, and the calculation unit predicts the physical properties or temperature history using a trained model generated by machine learning, in which the physical properties or temperature history of each position of the object are the objective variables and the molding conditions are the explanatory variables. (Note 7) The metal additive manufacturing system according to Note 4, characterized in that the molding unit molds the molded object by linear molding, supplying material to the processing point while stopping the processing point to which the heat source is supplied, and the calculation unit creates corrected thermal simulation data by correcting the temperature history of each position shown in the thermal simulation data based on the respective temperature histories of the starting point where molding begins and the ending point where molding ends among the layers constituting the molded object, and predicts the physical properties or the temperature history based on the corrected thermal simulation data. (Note 8) The metal additive manufacturing system according to Note 5 or 6, characterized in that a neural network, random forest, linear regression, or support vectors are employed for machine learning. (Note 9) A metal additive manufacturing system according to any one of Notes 1 to 8, comprising a monitoring unit for monitoring the temperature history when the molded object is being manufactured, and the molding unit comprising a molding command unit for modifying the molding conditions for the molded object being manufactured based on the result of comparing the temperature history acquired by the monitoring unit with the temperature history predicted by the calculation unit.(Note 10) A metal additive manufacturing apparatus comprising: an input unit into which information relating to the material of the molded object and the shape of the molded object is input; a molding condition generation unit that generates molding conditions for the molded object based on the molded object information; and a molding unit that performs the molding of the molded object, wherein it is determined whether the physical property values ​​indicating the characteristics of the molded object at each of a plurality of positions of the molded object or the predicted results of the temperature history of each of the plurality of positions fall within a set tolerance range when the molded object indicated in the molded object information is molded according to the molding conditions, and the molding condition generation unit outputs the molding conditions adjusted based on the result of the determination, and the molding unit performs the molding of the molded object according to the molding conditions output by the molding condition generation unit. (Note 11) A method for manufacturing a metal additive manufacturing product, comprising: receiving manufacturing product information indicating the material of the manufacturing product and the shape of the manufacturing product; generating manufacturing conditions for the manufacturing product based on the manufacturing product information; predicting physical property values ​​indicating the characteristics of the manufacturing product at each of a plurality of positions of the manufacturing product or the temperature history of each of the plurality of positions when the manufacturing product indicated in the manufacturing product information is manufactured according to the manufacturing conditions; determining whether the predicted results of the physical property values ​​or the temperature history fall within a set tolerance range; adjusting the manufacturing conditions based on the result of determining the prediction results; and manufacturing the manufacturing product according to the adjusted manufacturing conditions.

[0154] 1, 1A Metal additive manufacturing system, 2 Metal additive manufacturing apparatus, 3 Information processing apparatus, 4 Sensor, 5 Molded object information input unit, 6 Molding condition generation unit, 7 Molding unit, 8 Data input unit, 9 Storage unit, 10 Calculation unit, 11 Processing unit, 12 Prediction unit, 13 Learning unit, 21 Wire, 22 Wire nozzle, 23 Laser oscillator, 24 Beam nozzle, 25 Fiber cable, 26 Processing head, 27 Gas nozzle, 28 Gas supply device, 29 Piping, 30 Rotation mechanism, 31 Stage, 32 Radiation thermometer, 33 Head drive device, 34 Material supply mechanism, 35 Control device, 40 Control circuit, 41 Input unit, 42 Processor, 43 Memory, 44 Output unit.

Claims

1. A metal additive manufacturing system comprising: an object information input unit that receives object information indicating the material and shape of the object; a molding condition generation unit that generates molding conditions for the object based on the object information; a calculation unit that predicts physical property values ​​indicating the characteristics of the object at each of a plurality of positions of the object or the temperature history of each of the plurality of positions when the object indicated in the object information is molded according to the molding conditions; a processing unit that determines whether the predicted results of the physical property values ​​or the temperature history fall within a set tolerance range; and a molding unit that molds the object according to the molding conditions adjusted based on the result of the determination by the processing unit.

2. The metal additive manufacturing system according to claim 1, comprising a storage unit that stores manufacturing performance data, which is data in which the manufacturing conditions and information of the manufactured object in past manufacturing processes are associated with the physical properties or temperature history at each of the multiple locations, or thermal simulation data, which is the result of simulating the temperature history at each of the multiple locations, wherein the calculation unit predicts the physical properties or temperature history based on the manufacturing performance data or the thermal simulation data.

3. The metal additive manufacturing system according to claim 1, comprising a storage unit that stores, in addition to the printing conditions and printed object information in past printing, printing performance data which is data relating the sum of heat transfer indices from an arbitrary point, which is each of a plurality of positions in past printing, to adjacent points adjacent to the arbitrary point, or thermal simulation data which is the simulation result of the sum of heat transfer indices, wherein the calculation unit predicts the physical property value at the arbitrary point or the temperature history at the arbitrary point based on the printing performance data or the thermal simulation data.

4. The metal additive manufacturing system according to claim 2 or 3, characterized in that the calculation unit predicts the physical properties or the temperature history based on the manufacturing performance data by inputting the manufacturing conditions into a trained model which is the result of learning the manufacturing performance data.

5. The metal additive manufacturing system according to claim 4, characterized in that the molding unit performs the molding of the object by point molding, supplying material to the processing point while the processing point to which the heat source is supplied is stopped, and the calculation unit predicts the physical properties or temperature history of each position of the object using the trained model generated by machine learning, in which the physical properties or temperature history of each position of the object are the target variables and the molding conditions are the explanatory variables.

6. The metal additive manufacturing system according to claim 4, characterized in that the molding unit fabricates the object by linear molding, supplying material to the processing point while moving the processing point to which a heat source is supplied, and the calculation unit predicts the physical properties or temperature history of each position of the object using a trained model generated by machine learning, in which the physical properties or temperature history of each position of the object are the target variables and the molding conditions are the explanatory variables.

7. The metal additive manufacturing system according to claim 4, characterized in that the molding unit performs the molding of the object by linear molding, supplying material to the processing point while stopping the processing point to which a heat source is supplied, and the calculation unit creates corrected thermal simulation data by correcting the temperature history of each position shown in the thermal simulation data based on the respective temperature histories of the starting point where molding is started and the ending point where molding is finished among the layers constituting the object, and predicts the physical properties or the temperature history based on the corrected thermal simulation data.

8. The metal additive manufacturing system according to claim 5 or 6, characterized in that the machine learning employs a neural network, random forest, linear regression, or support vectors.

9. A metal additive manufacturing system according to any one of claims 1 to 8, comprising a monitoring unit for monitoring the temperature history when the object is being manufactured, and the manufacturing unit comprising a manufacturing command unit for modifying the manufacturing conditions for the object being manufactured based on the result of comparing the temperature history acquired by the monitoring unit with the temperature history predicted by the calculation unit.

10. A metal additive manufacturing apparatus comprising: an input unit into which information relating to the material and shape of a molded object is input; a molding condition generation unit that generates molding conditions for the molded object based on the molded object information; and a molding unit that performs molding of the molded object, wherein it is determined whether the physical property values ​​indicating the characteristics of the molded object at each of a plurality of positions of the molded object or the predicted results of the temperature history of each of the plurality of positions fall within a set tolerance range when the molded object indicated in the molded object information is molded according to the molding conditions, and the molding condition generation unit outputs the molding conditions adjusted based on the result of the determination, and the molding unit performs molding of the molded object according to the molding conditions output by the molding condition generation unit.

11. A method for manufacturing a metal additive manufacturing product, comprising: receiving manufacturing product information indicating the material of the manufacturing product and the shape of the manufacturing product; generating manufacturing conditions for the manufacturing product based on the manufacturing product information; predicting physical property values ​​indicating the characteristics of the manufacturing product at each of a plurality of positions of the manufacturing product or the temperature history of each of the plurality of positions when the manufacturing product indicated in the manufacturing product information is manufactured according to the manufacturing conditions; determining whether the predicted results of the physical property values ​​or the temperature history fall within a set tolerance range; adjusting the manufacturing conditions based on the result of determining the prediction results; and manufacturing the manufacturing product according to the adjusted manufacturing conditions.

Images