JP2022024333A - Additive manufacturing device and additive manufacturing method - Google Patents

Abstract

To provide an additive manufacturing device capable of shortening a time required for manufacturing and capable of manufacturing a shaped article having high shape accuracy.SOLUTION: An additive manufacturing device 100 melts a filler metal by irradiation with a beam, and manufactures a shaped article by stacking layers which are solidified products of the melted filler metal. The additive manufacturing device 100 includes an objective function updating unit 14 which updates the value of the objective function by inputting a measured value 25 which is a result of measuring the shape of the layer of the manufactured, shaped article to the objective function having the shape error of the layer as a variable, and a search processing unit 15 which searches for a processing condition based on an evaluation having the value of the objective function as the reference.SELECTED DRAWING: Figure 1

Description

The present disclosure relates to an addition manufacturing apparatus and an addition manufacturing method for manufacturing a three-dimensional model.

As one of the techniques for manufacturing a three-dimensional model, the technique of additive manufacturing (AM) is known. According to the Direct Energy Deposition (DED) method, which is one of several methods in the additive manufacturing technology, the additive manufacturing device supplies the filler metal to the processing point where the beam is irradiated. By moving the processing point while forming a bead. The bead is a solidified product obtained by solidifying the molten filler material. The additive manufacturing apparatus manufactures a model by sequentially stacking layers of beads.

In the modeled object, there may be a place where an error is likely to occur between the bead shape determined based on the design data and the bead shape formed. Whether or not an error occurs can be estimated to some extent based on the experience of the user who uses the additional manufacturing apparatus. However, it is difficult to estimate the location where the error occurs or the degree of the error with high accuracy because the occurrence of the error involves various physical phenomena during the manufacture of the modeled object.

Patent Document 1 includes a processing head that sprays metal powder, which is a filler material, at the same time as irradiating a beam, and detects an error that is the difference between the actual distance and the assumed distance between the spraying surface of the metal powder and the processing head. An additional manufacturing apparatus for adjusting the supply amount of metal powder in some cases is disclosed. When an error is detected in one layer, the additional manufacturing apparatus according to Patent Document 1 calculates processing conditions such as a supply amount, a laser output, and a processing head speed for the layers stacked on the layer.

Japanese Unexamined Patent Publication No. 2017-190505

In the conventional additive manufacturing apparatus disclosed in Patent Document 1, when there is a shape error in which the height of the layer is insufficient, post-processing is performed to compensate for the insufficient height in the next layer to be formed. Will be. For this reason, in the conventional additional manufacturing apparatus, there is a problem that the time required for manufacturing becomes long because post-processing is required every time a shape error which is insufficient in height occurs. Further, in the conventional additive manufacturing apparatus, when a shape error occurs due to a large deviation of the processing conditions from the appropriate processing conditions, the shape error cannot be corrected in the next layer to be formed, and the shape accuracy is improved. There was a problem that it may decrease.

The present disclosure has been made in view of the above, and an object of the present invention is to obtain an additional manufacturing apparatus capable of shortening the time required for manufacturing and manufacturing a modeled object having high shape accuracy.

In order to solve the above-mentioned problems and achieve the object, the additional manufacturing apparatus according to the present disclosure melts the filler metal by irradiation with a beam, and forms by stacking layers which are solidified materials of the melted filler metal. Manufacture things. The additional manufacturing apparatus according to the present disclosure updates the value of the objective function by inputting the measured value which is the result of measuring the layer of the manufactured model into the objective function whose variable is the shape error of the layer. It includes an update unit and a search processing unit that searches for machining conditions based on evaluation based on the value of the objective function.

The additional manufacturing apparatus according to the present disclosure has an effect that the time required for manufacturing can be shortened and a modeled object having high shape accuracy can be manufactured.

The block diagram which shows the structure of the addition manufacturing apparatus which concerns on Embodiment 1. The figure which shows the example of the shaped object manufactured by the addition manufacturing apparatus which concerns on Embodiment 1. The figure for demonstrating the search of the processing condition by the addition manufacturing apparatus which concerns on Embodiment 1. A flowchart showing an operation procedure of the addition manufacturing apparatus according to the first embodiment. A block diagram showing a configuration of a search processing unit when searching for machining conditions using machine learning in the first embodiment. The block diagram which shows the functional structure of the learning apparatus which the addition manufacturing apparatus which concerns on Embodiment 1 has. A flowchart showing a processing procedure of the learning device according to the first embodiment. The block diagram which shows the functional structure of the inference apparatus which the addition manufacturing apparatus which concerns on Embodiment 1 has. A flowchart showing a processing procedure of the inference device according to the first embodiment. The block diagram which shows the structure of the processing condition search apparatus which the addition manufacturing apparatus which concerns on Embodiment 2 has. The figure for demonstrating the operation of the addition manufacturing apparatus which concerns on Embodiment 3. The figure for demonstrating the modification of Embodiment 3 The figure for demonstrating the operation of the addition manufacturing apparatus which concerns on Embodiment 4. The figure which shows the hardware configuration example which the processing condition search apparatus which concerns on Embodiment 1 to 4 has.

Hereinafter, the addition manufacturing apparatus and the addition manufacturing method according to the embodiment will be described in detail with reference to the drawings.

Embodiment 1.
FIG. 1 is a block diagram showing a configuration of an addition manufacturing apparatus according to the first embodiment. The addition manufacturing apparatus 100 according to the first embodiment is a machine tool that manufactures a modeled object by adding a molten filler material to a workpiece. The additional manufacturing apparatus 100 melts the filler metal by irradiation with a beam, and manufactures a modeled object by stacking layers that are solidified materials of the melted filler metal. The filler material is a material such as metal wire or metal powder. The filler material may be other than metal. In Embodiment 1, the beam is a laser beam. The beam may be a beam such as an electron beam or an arc beam.

The additional manufacturing apparatus 100 includes a machining condition search device 10 for searching for machining conditions, a numerical control device 11 for controlling the additional manufacturing apparatus 100, a machining unit 12 for performing machining for manufacturing a modeled object, and a formed layer. It has a measuring instrument 13 for measuring the shape of the above. In the first embodiment, the machining condition search device 10 is built in the additional manufacturing device 100. The processing condition search device 10 may be an external device of the additional manufacturing device 100. The machining condition search device 10 may be a device that can be connected to the addition manufacturing device 100 via a network. The processing condition search device 10 may be a device existing on the cloud server. The machining conditions searched by the machining condition search device 10 may be applied to a plurality of additional manufacturing devices 100 connected to the machining condition search device 10 via a network.

The processing unit 12 includes a laser oscillator 20 that oscillates a laser beam, a material supply unit 21 that supplies a filler material to an irradiation position of the laser beam, and a shaft drive unit 22 that drives a processing head. The illustration of the processing head is omitted. The processing head emits a laser beam toward the workpiece. The shaft drive unit 22 moves the machining head in each of the X-axis direction, the Y-axis direction, and the Z-axis direction. The X-axis, Y-axis, and Z-axis are three axes perpendicular to each other. The X-axis and the Y-axis are horizontal axes. The Z axis is a vertical axis. In the addition manufacturing apparatus 100, the shaft drive unit 22 moves the machining head to move the irradiation position of the laser beam on the workpiece.

The measuring instrument 13 measures the three-dimensional shape of the layer by measuring the height of the layer in the Z-axis direction and the shape of the layer in the X-axis direction and the Y-axis direction. For example, a camera is used as the measuring instrument 13. The measuring instrument measures the three-dimensional shape of the layer by analyzing the captured image. As the measuring instrument 13, a laser displacement meter, an optical coherence tomography (OCT) for performing optical coherence tomography, or the like may be used.

The numerical control device 11 controls the addition manufacturing device 100 according to the machining program. The numerical control device 11 includes a machining program analysis unit 16 that analyzes a machining program, a command generation unit 17 that generates a command sent to the machining unit 12, and a machining condition output unit 18 that outputs machining conditions to the machining unit 12. Have.

The processing program analysis unit 16 obtains a movement path of the irradiation position based on the content of the processing described in the processing program. The machining program analysis unit 16 outputs the movement route data to the command generation unit 17. The machining program analysis unit 16 sets the machining conditions by acquiring information for setting the machining conditions from the machining program. The machining program analysis unit 16 may acquire the machining program by reading the machining conditions described in the machining program from the machining program.

The command generation unit 17 generates a position command based on the data of the movement route. The position command is a command indicating a group of interpolation points on the moving path of the laser beam. The command generation unit 17 controls the shaft drive unit 22 by sending the generated position command to the shaft drive unit 22.

The command generation unit 17 generates a beam output command according to the machining program and machining conditions. The beam output command is a command indicating the output of the laser beam. The command generation unit 17 controls the laser oscillator 20 by sending the generated laser output command to the laser oscillator 20.

The command generation unit 17 generates a supply command according to the machining program and machining conditions. The supply command is a command indicating the supply speed of the filler metal. The command generation unit 17 controls the material supply unit 21 by sending the generated supply command to the material supply unit 21.

Information on machining conditions, which is a search result by the machining condition search device 10 described later, is input to the machining condition output unit 18. The initial machining conditions are the machining conditions initially set according to the machining program, and are the preset machining conditions. The switching unit 19 switches between the search result and the initial machining conditions. By such switching, the machining condition output unit 18 outputs one of the machining condition information which is the search result and the initial machining condition information to the machining section 12. As a result, the machining conditions input to the machining unit 12 are switched between the search result and the initial machining conditions.

The machining condition search device 10 has an objective function update unit 14 and a search processing unit 15. The objective function update unit 14 holds an objective function whose variable is the shape error of the layer, and updates the value of the objective function by inputting the measured value of the shape error in the manufactured model to the objective function. The search processing unit 15 searches for machining conditions based on an evaluation based on the calculation result.

The command generation unit 17 outputs the command position information 24, which is information indicating the command position, to the objective function update unit 14 and the search processing unit 15. The command position is the position of each interpolation point constituting the interpolation point group. The measuring instrument 13 outputs the measured value 25, which is the measurement result, to the objective function updating unit 14 and the search processing unit 15. The machining condition output unit 18 outputs the machining condition information 26, which is one of the information of the machining condition which is the search result and the information of the initial machining condition, to the machining section 12 and the search processing section 15.

In the first embodiment, the objective function is expressed as "objective function = Σ | (target shape of layer)-(measured layer shape) |". “| (Target shape of layer)-(measured shape of layer) |” represents a shape error of the layer. The objective function update unit 14 updates the objective function by inputting the measured value 25 at each command position into the objective function. Further, the objective function update unit 14 updates the objective function by inputting the measured values 25 for each of the plurality of layers constituting the modeled object into the objective function.

The search processing unit 15 searches for machining conditions for each command position, which are machining conditions that minimize the value of the objective function. The search processing unit 15 searches for processing conditions such as a material supply speed, a laser beam output, a processing head feed speed, and a pitch. The pitch is the distance between the emission position of the laser beam and the workpiece.

The search processing unit 15 obtains the machining conditions that are the search results by adding the correction amount to the machining conditions used for the current machining. In the first search after starting the production of the modeled object, the search processing unit 15 searches for the processing conditions based on the initial processing conditions. After performing the initial search, the search processing unit 15 searches for the machining conditions based on the machining conditions that are the search results.

As the correction amount, a randomly determined correction amount can be used. The correction amount may be determined based on the correlation between the shape of the modeled object and the processing conditions. For the correlation, a relationship empirically obtained in additive manufacturing can be used. The correction amount may be determined based on a model of a physical phenomenon associated with modeling. A model of a physical phenomenon refers to a model in which the supply of filler metal is increased in order to increase the volume of the layer.

When one layer is formed during the production of one modeled object, the measuring instrument 13 measures the formed layer and sends the measured value 25 to the processing condition search device 10. The machining condition search device 10 acquires the command position information 24 corresponding to the measured value 25 from the command generation unit 17. Further, the machining condition search device 10 acquires target shape data indicating the target shape of the layer. The machining condition search device 10 acquires target shape data from, for example, design data of a modeled object. The objective function update unit 14 updates the value of the objective function by inputting the target shape data and the measured value 25 into the objective function.

The objective function update unit 14 acquires the measured value 25 associated with the command position. The objective function update unit 14 acquires the target shape data and the measured value 25 for each layer of the modeled object, and updates the value of the target function. Further, the objective function updating unit 14 updates the value of the target function in the manufacture of a plurality of shaped objects having the same shape as each other. The same shape as each other means that the three-dimensional shapes are the same as each other. A plurality of shaped objects having the same shape as each other are, for example, a plurality of shaped objects manufactured based on the same design data.

The objective function update unit 14 determines whether or not to end the update of the objective function value by evaluating whether or not the objective function value has been minimized. When it is determined that the value of the objective function is minimized, the objective function update unit 14 determines that the update is completed. The objective function update unit 14 sends a signal 23 indicating the presence / absence of update to the search processing unit 15. In the first embodiment, the minimization of the value of the objective function means a state in which the difference between the target shape and the formed shape is no longer detected by the measuring instrument 13.

The search processing unit 15 searches for machining conditions each time the value of the objective function is updated. The search processing unit 15 ends the search for machining conditions when the objective function updating unit 14 ends updating the value of the objective function. In this way, the search processing unit 15 searches for machining conditions based on the evaluation based on the value of the objective function.

Next, the search for machining conditions by the machining condition search device 10 will be described with reference to FIGS. 2 and 3. FIG. 2 is a diagram showing an example of a modeled object manufactured by the addition manufacturing apparatus according to the first embodiment. FIG. 3 is a diagram for explaining a search for processing conditions by the addition manufacturing apparatus according to the first embodiment. The model 30 shown in FIG. 2 has a three-layer wall shape manufactured by sequentially stacking the first layer 31, the second layer 32, and the third layer 33. Here, a search for processing conditions in the case of manufacturing a plurality of the modeled objects 30 shown in FIG. 2 will be described. The height of the first layer 31 is z1, the height of the second layer 32 is z2, and the height of the third layer 33 is z3. Such a height is a height in the Z-axis direction.

The modeled object 30-1 is a modeled object 30 manufactured by the first production among the plurality of modeled objects 30. The model 30-1 is a model 30 manufactured using the initial processing conditions. FIG. 3 shows the shape of the modeled object 30-1 on the XZ surface and the supply amount of the filler material when forming each layer. The supply amount is an example of the processing conditions to be searched. The vertical axis of the graph showing the supply amount is W, which is the supply amount, and the horizontal axis is the command position in the X-axis direction.

When the first layer 31 of the model 30-1 is formed, the measuring instrument 13 measures the shape of the first layer 31 at each of the command positions x1, x2, ..., Xn. When the second layer 32 of the model 30-1 is formed, the measuring instrument 13 measures the shape of the second layer 32 at each of the command positions x1, x2, ..., Xn. When the third layer 33 of the modeled object 30-1 is formed, the measuring instrument 13 measures the shape of the third layer 33 at each of the command positions x1, x2, ..., Xn. Let n be an integer.

The objective function update unit 14 updates the objective function using the measured value 25 for the model 30-1. The objective function update unit 14 inputs the absolute value of Δz1 at each command position x1, x2, ..., Xn to the objective function. Δz1 is a shape error of the first layer 31, which is the difference between the height of the formed first layer 31 and z1. The objective function update unit 14 inputs the absolute value of Δz2 at each command position x1, x2, ..., Xn to the objective function. Δz2 is a shape error of the second layer 32, which is the difference between the height of the formed second layer 32 and z2. The objective function update unit 14 inputs the absolute value of Δz3 at each command position x1, x2, ..., Xn to the objective function. Δz3 is a shape error of the third layer 33, which is the difference between the height of the formed third layer 33 and z3. The search processing unit 15 searches for the processing conditions of each command position x1, x2, ..., Xn as the objective function is updated.

The shape of the end of the bead at the processing start position and the end of the bead at the processing end position are liable to sag. In the modeled object 30-1, since the machining conditions have not been changed from the initial machining conditions, a shape error occurs between the bead machining start position and the bead machining end position in each layer. Each of the first layer 31, the second layer 32, and the third layer 33 of the modeled object 30-1 has a shape error in which the height in the Z-axis direction becomes low. When the additional manufacturing apparatus 100 determines that the height of the modeled object 30-1 is insufficient at the time when the production of the modeled object 30-1 is completed, the material 34-2,34 for compensating for the insufficient height is used. Perform post-processing to add -3.

As described above, the addition manufacturing apparatus 100 supplements the material by a post-process when the material added in the production of the model 30 is insufficient. The additional manufacturing apparatus 100 may remove a part of the added material by post-processing when the added material is excessive in the manufacturing of the model 30.

The additional manufacturing apparatus 100 performs the second manufacturing using the processing conditions which are the search results when the manufacturing of the modeled product 30-1 is completed, and manufactures the modeled product 30-2. The measuring instrument 13 measures in the same manner as in the case of the modeled object 30-1. When the production of the modeled object 30-2 is completed, the objective function updating unit 14 updates the objective function by inputting a shape error into the objective function as in the case of the modeled object 30-1. The search processing unit 15 searches for the processing conditions of each command position x1, x2, ..., Xn as the objective function is updated.

Since the processing conditions used for the model 30-2 are more appropriate than the initial processing conditions, the shape error in each layer of the model 30-2 is smaller than that of the model 30-1. .. Specifically, the insufficient height of each layer is compensated by increasing the supply amount W at the bead processing start position and the bead processing end position in each layer. The amount of material 34-3 added to the model 30-2 by post-processing is less than that of the model 30-1.

The modeled object 30-m is a modeled object 30 manufactured by the m-th manufacturing by the additional manufacturing apparatus 100. Let m be an integer. The processing condition search device 10 searches for processing conditions each time the production of the modeled object 30 having the same shape is repeated. The additional manufacturing apparatus 100 can reduce the shape error by searching for appropriate processing conditions each time the manufacturing of the modeled object 30 is repeated. The additional manufacturing apparatus 100 can efficiently reduce the shape error in the place where the shape error is likely to occur by optimizing the processing conditions for each command position. By repeating the production a plurality of times, as shown in FIG. 3, it is possible to obtain a modeled product 30-m in which the shape error is eliminated.

The additional manufacturing apparatus 100 searches for machining conditions each time the manufacturing of the modeled object 30 having the same shape is repeated, so that shape correction by post-machining becomes unnecessary. The additional manufacturing apparatus 100 can shorten the time required for manufacturing by eliminating the time required for manufacturing that does not require shape correction by post-processing.

FIG. 4 is a flowchart showing an operation procedure of the addition manufacturing apparatus according to the first embodiment. In step S1, the addition manufacturing apparatus 100 forms a layer of the model 30. In step S2, the addition manufacturing apparatus 100 measures the shape of the layer by the measuring instrument 13. In step S3, the addition manufacturing apparatus 100 determines whether or not the modeling of the modeled object 30 is completed.

The addition manufacturing apparatus 100 determines that the modeling of the modeled object 30 is completed when all the layers of the modeled object 30 are formed. If there is a layer of the modeled object 30 that has not yet been formed, the additional manufacturing apparatus 100 determines that the modeling of the modeled object 30 has not been completed. When the modeling of the modeled object 30 is completed (steps S3 and Yes), the addition manufacturing apparatus 100 proceeds to step S4 described below. When the modeling of the modeled object 30 is not completed (steps S3 and No), the additional manufacturing apparatus 100 returns to step S3 to form a layer that has not yet been formed.

In step S4, the addition manufacturing apparatus 100 updates the objective function with the measured value 25 acquired by the measurement in step S2. In step S5, the addition manufacturing apparatus 100 searches for processing conditions in the search processing unit 15. As a result, the addition manufacturing apparatus 100 ends the operation according to the procedure shown in FIG. When manufacturing a modeled object 30 having the same shape, the additional manufacturing apparatus 100 updates the objective function and searches for processing conditions by the operation according to the procedure shown in FIG.

The search processing unit 15 may search for machining conditions using machine learning. FIG. 5 is a block diagram showing a configuration of a search processing unit when searching for machining conditions using machine learning in the first embodiment. The search processing unit 15 includes a learning device 41 that performs machine learning, an inference device 42 that infers processing conditions that reduce the shape error of the modeled object 30, and a model storage unit 43 that stores the learned model. The learning device 41 learns the processing conditions that minimize the objective function and generates a trained model. The inference device 42 infers the machining conditions for each command position, which are the machining conditions that minimize the objective function, using the trained model.

FIG. 6 is a block diagram showing a functional configuration of a learning device included in the addition manufacturing device according to the first embodiment. The learning device 41 has a data acquisition unit 44 and a model generation unit 45. The data acquisition unit 44 acquires the command position information 24, the measured value 25, and the machining condition information 26. In addition, the data acquisition unit 44 acquires target shape data indicating the target shape. The data acquisition unit 44 acquires command position information 24, measured value 25, machining condition information 26, and target shape data as learning data.

The data acquisition unit 44 creates a data set in which the measured value 25 for each command position, the machining condition information 26 for each command position, and the target shape data for each command position are combined. That is, the data acquisition unit 44 creates a data set including the measured value 25 associated with the command position, the machining condition information 26, and the target shape data. The data acquisition unit 44 outputs the created data set to the model generation unit 45.

The model generation unit 45 generates a trained model using the data set. The learning algorithm used by the model generation unit 45 may be any. As an example, the case where reinforcement learning is applied will be described. Reinforcement learning is that an action subject who is an agent in a certain environment observes the current state and decides an action to be taken. Agents get rewards from the environment by choosing actions and learn how to get the most rewards through a series of actions. Q-learning and TD-Learning are known as typical methods of reinforcement learning. For example, in the case of Q-learning, the action value table, which is a general update expression of the action value function Q (s, a), is expressed by the following equation (1). The action value function Q (s, a) represents the action value Q, which is the value of the action of selecting the action “a” under the environment “s”.

In the update formula expressed by the above formula (1), if the action value of the best action "a" at the time "t + 1" is larger than the action value Q of the action "a" executed at the time "t". , The action value Q is increased, and in the opposite case, the action value Q is decreased. In other words, the action value function Q (s, a) is updated so that the action value Q of the action “a” at the time “t” approaches the best action value at the time “t + 1”. As a result, the best behavioral value in one environment is sequentially propagated to the behavioral value in the previous environment.

The model generation unit 45 has a reward calculation unit 46 and a function update unit 47. The reward calculation unit 46 calculates the reward based on the value of the objective function. The function update unit 47 updates the function for determining the processing conditions according to the reward calculated by the reward calculation unit 46. The function update unit 47 outputs the trained model created by updating the function to the model storage unit 43.

The reward calculation unit 46 calculates the value of the target function based on the measured value 25 associated with the command position and the target shape data. The reward calculation unit 46 calculates the reward "r" based on the value of the objective function. The reward calculation unit 46 increases the reward "r" when the value of the objective function becomes small. The reward calculation unit 46 increases the reward "r" by giving the reward value "1". The reward value is not limited to "1". Further, the reward calculation unit 46 reduces the reward "r" when the difference becomes large. The reward calculation unit 46 reduces the reward "r" by giving the reward value "-1". The reward value is not limited to "-1".

FIG. 7 is a flowchart showing a processing procedure of the learning device according to the first embodiment. A reinforcement learning method for updating the action value function Q (s, a) will be described with reference to the flowchart of FIG. 7.

In step S11, the learning device 41 acquires learning data. The data acquisition unit 44 creates a data set using the learning data. In step S12, the learning device 41 calculates the reward. In step S13, the learning device 41 updates the action value function Q (s, a) based on the reward. In step S14, the learning device 41 determines whether or not the action value function Q (s, a) has converged. The learning device 41 determines that the action value function Q (s, a) has converged because the action value function Q (s, a) is not updated in step S14.

When it is determined that the action value function Q (s, a) has not converged (step S14, No), the learning device 41 returns the operation procedure to step S11. When it is determined that the action value function Q (s, a) has converged (steps S14, Yes), the learning device 41 ends the learning according to the procedure shown in FIG. 7. The learning device 41 may continue learning by returning the operation procedure from step S13 to step S11 without performing the determination in step S14. The model storage unit 43 stores the trained model which is the generated action value function Q (s, a).

FIG. 8 is a block diagram showing a functional configuration of an inference device included in the addition manufacturing device according to the first embodiment. The inference device 42 infers the machining conditions for each command position, which are the machining conditions that minimize the value of the target function, based on the trained model. The inference device 42 has a data acquisition unit 48 and an inference unit 49.

The data acquisition unit 48 acquires the command position information 24 as inference data. The data acquisition unit 48 outputs the command position information 24 to the inference unit 49. The inference unit 49 infers the processing conditions for each command position that minimizes the value of the target function by inputting the command position information 24 into the trained model read from the model storage unit 43.

FIG. 9 is a flowchart showing a processing procedure of the inference device according to the first embodiment. In step S15, the inference device 42 acquires the command position information 24, which is the inference data, in the data acquisition unit 48. In step S16, the inference device 42 inputs the command position information 24, which is the acquired data, to the trained model in the inference unit 49, and infers the processing conditions for each command position. In step S17, the inference device 42 outputs the machining condition information indicating the machining condition which is the inference result to the numerical control device 11. As a result, the inference device 42 ends the process according to the procedure shown in FIG.

The addition manufacturing apparatus 100 can efficiently search for machining conditions that minimize the value of the objective function by a machine learning method.

In the first embodiment, the case where reinforcement learning is applied to the learning algorithm used by the learning device 41 has been described, but learning other than reinforcement learning may be applied to the learning algorithm. The learning device 41 may execute machine learning using a known learning algorithm other than reinforcement learning, for example, a learning algorithm such as deep learning, neural network, genetic programming, functional logic programming, or a support vector machine. good.

In the first embodiment, the learning device 41 is built in the additional manufacturing device 100. The learning device 41 is not limited to the device included in the addition manufacturing device 100, and may be an external device of the addition manufacturing device 100. The learning device 41 may be a device that can be connected to the addition manufacturing device 100 via a network. The learning device 41 may be a device existing on the cloud server.

The learning device 41 may learn the parameters according to the data set created for the plurality of additional manufacturing devices 100. The learning device 41 may acquire a data set from a plurality of additional manufacturing devices 100 used at the same site, or may acquire a data set from a plurality of additional manufacturing devices 100 used at different sites. Is also good. The data set may be collected from a plurality of additional manufacturing devices 100 that operate independently of each other at a plurality of sites. After starting the collection of the data set from the plurality of additional manufacturing devices 100, a new additional manufacturing device 100 may be added to the target for which the data set is collected. Further, after starting the collection of the data set from the plurality of additional manufacturing devices 100, a part of the plurality of additional manufacturing devices 100 may be excluded from the target for which the data set is collected.

The learning device 41 that has learned about one additional manufacturing device 100 may learn about other additional manufacturing devices 100 other than the additional manufacturing device 100. The learning device 41 that learns about the other additional manufacturing apparatus 100 can update the output prediction model by re-learning in the other additional manufacturing apparatus 100.

According to the first embodiment, the additional manufacturing apparatus 100 updates the value of the target function by inputting the measured value 25 into the objective function whose variable is the shape error of the layer, and evaluates based on the value of the target function. Search for machining conditions based on. Since the additional manufacturing apparatus 100 does not require post-machining by searching for machining conditions based on the evaluation based on the value of the target function, the time required for manufacturing can be shortened. Further, the additional manufacturing apparatus 100 can manufacture a modeled object having high shape accuracy by searching for machining conditions based on the evaluation based on the value of the target function. As described above, the additional manufacturing apparatus 100 has an effect that the time required for manufacturing can be shortened and a modeled object having high shape accuracy can be manufactured.

Embodiment 2.
In the second embodiment, an example of updating the value of the target function and searching for the processing condition based on the result of simulating the production of the modeled object 30 will be described. FIG. 10 is a block diagram showing a configuration of a processing condition search device included in the addition manufacturing device according to the second embodiment. In the second embodiment, the same components as those in the first embodiment are designated by the same reference numerals, and the configurations different from those in the first embodiment will be mainly described.

The machining condition search device 50 includes an objective function update unit 14, a search processing unit 15, and a simulation unit 51. The simulation unit 51 simulates the production of the model 30.
In the second embodiment, the simulation unit 51 is built in the machining condition search device 50. The simulation unit 51 may be an external device of the machining condition search device 50. The simulation unit 51 may be an external device of the additional manufacturing device 100.

Each time the simulation unit 51 simulates the production of the modeled object 30, the simulation unit 51 outputs shape data representing the shape of each layer constituting the modeled object 30 to the objective function updating unit 14. The search processing unit 15 searches for the machining conditions used in the simulation, and outputs the machining condition information representing the machining conditions as the search result to the simulation unit 51. The simulation unit 51 simulates the production of the modeled object 30 according to the input processing condition information. When the simulation by the simulation unit 51 is completed, the search processing unit 15 outputs the processing condition information representing the processing conditions as the search result to the numerical control device 11.

According to the second embodiment, the additional manufacturing apparatus 100 searches for the machining conditions by performing a simulation without consuming the filler metal and electric power required for the actual molding. can.

Embodiment 3.
In the third embodiment, an example of searching for the processing conditions for the target shape will be described using the search results for the processing conditions for the modeled object having a shape similar to the target shape. FIG. 11 is a diagram for explaining the operation of the addition manufacturing apparatus according to the third embodiment. In the third embodiment, the same components as those in the first or second embodiment are designated by the same reference numerals, and the configurations different from those in the first or second embodiment will be mainly described.

The modeled object 30b shown in FIG. 11 is a first modeled object that is subject to processing conditions. As for the modeled object 30a, it is assumed that the search for processing conditions has already been performed as in the case shown in the first or second embodiment. The modeled object 30a is a second modeled object whose processing conditions have already been searched for. It is assumed that the target shape SA of the modeled object 30b and the shape of the modeled object 30a are similar to each other, and the target shape SA is a scaled-up shape of the modeled object 30a. The similarity ratio R1 is a similarity ratio between the shape of the modeled object 30a and the target shape SA.

FIG. 11 shows a graph of the supply amount W which is the searched processing condition of the modeled object 30a and a graph of the supply amount W which is the processing condition of the modeled object 30b. The search processing unit 15 determines the association between the supply amount W and the command position in the initial processing condition by changing the association between the supply amount W and the command position in the searched processing condition by using the similarity ratio R1. do. As described above, the search processing unit 15 uses the processing conditions in which the searched processing conditions of the modeled object 30a are adjusted based on the similarity ratio R1 as the initial processing conditions when searching for the processing conditions for the modeled object 30b. The search processing unit 15 searches for the processing conditions for the modeled object 30b based on the searched processing conditions for the modeled object 30a.

According to the third embodiment, the additional manufacturing apparatus 100 can shorten the time required for the search by searching for the machining conditions based on the searched machining conditions. Further, the additional manufacturing apparatus 100 can search for the processing conditions for the modeled object 30a, which is a scaled-down version of the modeled object 30a, in advance, and then search for the processing conditions for the modeled object 30a. In addition, the number of times to manufacture the full-scale model 30a can be reduced. As a result, the addition manufacturing apparatus 100 can reduce the consumption of filler metal, electric power, and the like.

In the third embodiment, even if the shape of the first modeled object and the shape of the second modeled object are not similar to each other, the first aspect is based on the searched processed shape of the second modeled object. You may search for the processed shape of the modeled object. A search for a second modeled object even when the shape of the first modeled object is similar to the shape of the second modeled object and the height in the Z-axis direction or the length in the X-axis direction is different. You may search for the processed shape of the first modeled object based on the processed shape. The shape of the first modeled object is a shape obtained by changing the shape of the second modeled object at a constant size ratio. In such a case, the shape of the first modeled object shall be similar to the shape of the second modeled object.

FIG. 12 is a diagram for explaining a modification of the third embodiment. The modeled object 30d shown in FIG. 12 is a first modeled object that is subject to processing conditions. As for the modeled object 30c, it is assumed that the search for processing conditions has already been performed as in the case shown in the first or second embodiment. The modeled object 30c is a second modeled object whose processing conditions have already been searched for. The shape of the modeled object 30d and the shape of the modeled object 30c are similar to each other. The shape of the modeled object 30d is a reduced shape SB of the modeled object 30c with a size ratio R2 in the X-axis direction and the Z-axis direction.

The search processing unit 15 determines the association between the supply amount W and the command position in the initial processing condition by changing the association between the supply amount W and the command position in the searched processing condition by using the size ratio R2. do. As described above, the search processing unit 15 uses the processing conditions in which the searched processing conditions of the modeled object 30c are adjusted based on the size ratio R2 as the initial processing conditions when searching for the processing conditions for the modeled object 30d. The search processing unit 15 searches for the processing conditions for the modeled object 30d based on the searched processing conditions for the modeled object 30c. Also in the case of the modification, the additional manufacturing apparatus 100 can shorten the time required for the search by searching for the machining conditions based on the searched machining conditions.

Embodiment 4.
In the fourth embodiment, an example in which the search for machining conditions is terminated when the shape error is within the permissible range will be described. FIG. 13 is a diagram for explaining the operation of the addition manufacturing apparatus according to the fourth embodiment. In the fourth embodiment, the same components as those in the first to third embodiments are designated by the same reference numerals, and the configurations different from those in the first to third embodiments will be mainly described.

In the fourth embodiment, the objective function updating unit 14 ends updating the objective function value when the objective function value is converged within a preset range. The search processing unit 15 searches for machining conditions when each of the shape errors Δz1, Δz2, and Δz3 becomes less than a preset threshold value in response to the completion of the update by the objective function update unit 14. finish. Assuming that the value of the objective function is minimized by repeating the manufacturing and the search for the processing conditions m times, the modeled object 30- (mk) which is the modeled object 30 manufactured in the mkth time. When it is determined that the update of the value of the objective function is completed, the additional manufacturing apparatus 100 ends the search for the processing conditions for the modeled object 30. k is an integer less than m.

According to the fourth embodiment, the additional manufacturing apparatus 100 can shorten the time required for the search by ending the search for the machining conditions when the shape error is within the allowable range.

Next, the hardware configuration of the machining condition search devices 10 and 50 according to the first to fourth embodiments will be described. FIG. 14 is a diagram showing a hardware configuration example of the machining condition search device according to the first to fourth embodiments. FIG. 14 shows a hardware configuration when the functions of the machining condition search devices 10 and 50 are realized by using the hardware for executing the program.

The machining condition search devices 10 and 50 include a processor 61 that executes various processes, a memory 62 that is a built-in memory, a storage device 63 that stores information, and input of information to the machining condition search devices 10 and 50 and machining conditions. It has an interface circuit 64 for outputting information from the search devices 10 and 50.

The processor 61 is a CPU (Central Processing Unit). The processor 61 may be a processing device, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor). The memory 62 is a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory) or an EEPROM (registered trademark) (Electrically Erasable Programmable Read Only Memory).

The storage device 63 is an HDD (Hard Disk Drive) or an SSD (Solid State Drive). The program that causes the computer to function as the processing condition search device 10 is stored in the storage device 63. The processor 61 reads the program stored in the storage device 63 into the memory 62 and executes it.

The program may be stored in a storage medium that can be read by a computer system. The processing condition search devices 10 and 50 may store the program recorded in the storage medium in the memory 62. The storage medium may be a portable storage medium that is a flexible disk, or a flash memory that is a semiconductor memory. The program may be installed in a computer system from another computer or server device via a communication network.

The functions of the objective function update unit 14, the search processing unit 15, and the simulation unit 51 are realized by a combination of the processor 61 and software. Each of the functions may be realized by a combination of the processor 61 and the firmware, or may be realized by a combination of the processor 61, the software and the firmware. The software or firmware is described as a program and stored in the storage device 63. Each function of the model storage unit 43 is realized by using the storage device 63. The interface circuit 64 receives a signal from an external device connected to the hardware. The external devices are the numerical control device 11, the processing unit 12, and the measuring instrument 13.

The configuration shown in each of the above embodiments shows an example of the contents of the present disclosure. The configurations of each embodiment can be combined with other known techniques. The configurations of the respective embodiments may be appropriately combined. It is possible to omit or change a part of the configuration of each embodiment without departing from the gist of the present disclosure.

10,50 Machining condition search device, 11 Numerical control device, 12 Machining unit, 13 Measuring instrument, 14 Objective function update unit, 15 Search processing unit, 16 Machining program analysis unit, 17 Command generation unit, 18 Machining condition output unit, 19 Switching unit, 20 laser oscillator, 21 material supply unit, 22 axis drive unit, 23 signal, 24 command position information, 25 measured value, 26 processing condition information, 30,30-1,30-2,30- (m-k) ), 30-m, 30a, 30b, 30c, 30d Model, 31 1st layer, 32 2nd layer, 33 3rd layer, 34-2, 34-3 Material, 41 Learning device, 42 Reasoning device, 43 model Storage unit, 44,48 data acquisition unit, 45 model generation unit, 46 reward calculation unit, 47 function update unit, 49 inference unit, 51 simulation unit, 61 processor, 62 memory, 63 storage device, 64 interface circuit, 100 additional manufacturing Device.

Claims (11)

It is an additional manufacturing apparatus that melts the filler metal by irradiation with a beam and stacks layers that are solidified materials of the melted filler metal to manufacture a modeled object.
An objective function updater that updates the value of the objective function by inputting a measured value that is the result of measuring the layer of the manufactured object into the objective function that uses the shape error of the layer as a variable.
An additional manufacturing apparatus including a search processing unit that searches for machining conditions based on an evaluation based on the value of the objective function. A command generator for generating a command indicating a group of interpolation points on the moving path of the beam is provided.
The objective function update unit updates the objective function by inputting the measured value at each command position, which is the position of each interpolation point included in the interpolation point group, into the objective function.
The additional manufacturing apparatus according to claim 1, wherein the search processing unit searches for the processing conditions for each command position. Claim 1 or claim 1, wherein the objective function updating unit updates the value of the objective function by inputting the measured value for each of the plurality of layers constituting the modeled object into the objective function. 2. The additional manufacturing apparatus according to 2. The objective function update unit updates the value of the objective function using the measured value obtained by simulating the production of the modeled object.
The additional manufacturing apparatus according to any one of claims 1 to 3, wherein the search processing unit searches for the processing conditions used in the simulation. When the shape of the first modeled object for which the processing conditions are searched and the shape of the second modeled object for which the processing conditions have already been searched are similar to each other, the search processing unit may perform the search processing unit. The addition according to any one of claims 1 to 3, wherein the processing conditions for the first modeled object are searched based on the search result of the processing conditions for the second modeled object. Manufacturing equipment. The additional manufacturing apparatus according to any one of claims 1 to 5, wherein the search processing unit searches for the processing conditions that minimize the value of the objective function. One of claims 1 to 5, wherein the objective function update unit ends updating the value of the objective function when the value of the objective function is converged within a preset range. The additional manufacturing equipment described in 1. The search processing unit is characterized by having an inference device that infers the machining conditions for each command position, which are the machining conditions that minimize the objective function, based on the command position information indicating the command position. The additional manufacturing apparatus according to claim 2. The search processing unit has a learning device that generates a trained model for inferring the machining conditions for each command position, which are the machining conditions that minimize the objective function.
The additional manufacturing according to claim 8, wherein the inference device infers the machining conditions for each command position, which are the machining conditions that minimize the objective function, using the trained model. Device. The learning device includes a data acquisition unit that acquires the command position information, the measured value, the processing condition information indicating the processing condition, and the target shape data indicating the target shape of the modeled object.
A model generation unit that generates the trained model using a data set that combines the measured values for each command position, the processing condition information for each command position, and the target shape data for each command position. The additional manufacturing apparatus according to claim 9, further comprising: It is an additional manufacturing method in which a filler metal is melted by irradiation with a beam, and a modeled product is manufactured by an additive manufacturing apparatus that stacks layers that are solidified of the melted filler metal.
A step of updating the value of the objective function by inputting a measured value which is a result of measuring the layer of the manufactured object into the objective function having the shape error of the layer as a variable.
An addition manufacturing method comprising a step of searching for machining conditions based on an evaluation based on the value of the objective function.

Images