Machining permissibility determination device, machining permissibility learning device, machining permissibility determination method, machining permissibility learning method, machining permissibility determination program, and machining permissibility learning program

Abstract

A machining permissibility determination device (100) is provided with: a three-dimensional shape extraction unit (10) that extracts a three-dimensional shape from three-dimensional shape data indicating a predetermined machining shape; a three-dimensional shape rotation unit (12) that rotates the three-dimensional shape such that a machining surface in the three-dimensional shape faces a desired machining axis direction; a depth map conversion unit (14) that generates a depth map obtained by adding depth information in the desired machining axis direction extracted from the three-dimensional shape data to a two-dimensional image obtained by orthogonally projecting the machining surface to a plane perpendicular to the desired machining axis direction; and an estimation model (16) constructed by machine learning using a learning machining axis direction instruction, a learning depth map extracted from the learning three-dimensional shape data, and machining permissibility information on which the actual machining permissibility can be performed in accordance with the learning machining axis direction instruction and the learning three-dimensional shape data is recorded. A determination is made as to whether or not machining of the predetermined shape is possible by using an estimation using the desired machining axis and the depth map generated by the depth map conversion unit (14).

Description

Technical Field

[0001] This disclosure relates to a processing capability determination device, a processing capability learning device, a processing capability determination method, a processing capability learning method, a processing capability determination procedure, and a processing capability learning procedure. Background Technology

[0002] Cutting of raw materials (work objects) using cutting tools such as drills, milling cutters, or turning tools (hereinafter referred to as "tools") can sometimes be difficult to perform using tools due to their design shape. In addition, defects may occur due to expansion of the raw materials or deformation of the tools caused by heat generated during machining, as well as machining defects caused by tool vibration and deflection.

[0003] Patent Document 1 discloses an electrode manufacturing method for determining whether an electrode for electrical discharge machining (EDM) can be manufactured. Additionally, Non-Patent Document 1 discloses a technique that uses a device to learn a 3D-CNN (three-dimensional convolutional neural network) model by inputting voxel data representing a three-dimensional shape into the model to determine whether cutting can be performed.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2010-105080

[0007] Non-patent literature

[0008] Non-patent literature 1: Computer Aided Geometric Design 62 (2018) 263-275 Summary of the Invention

[0009] The problem that the invention aims to solve

[0010] However, the technology described in Patent Document 1 has the following problems: It extracts all the edges constituting the electrode shape based on the input three-dimensional electrode data and shape recognition information, and then determines whether the electrode contour determined based on the extracted edges can be processed. Therefore, the computational load for extracting the electrode contour from the three-dimensional electrode data increases. Furthermore, in technologies not described in Patent Document 1, large-scale 3D-CNN models are required to process the voxel data as three-dimensional data, thus increasing the computational load and the amount of data required for learning.

[0011] The purpose of this disclosure is to provide a machining capability determination device, a machining capability determination method, and a machining capability determination program for determining whether cutting is permissible (whether cutting is possible or not). The machining capability determination device, machining capability learning method, and machining capability learning program are constructed by suppressing the learning data number and computational load.

[0012] Methods for solving problems

[0013] The processing feasibility determination apparatus disclosed herein comprises: a three-dimensional shape extraction unit that extracts a three-dimensional shape from three-dimensional shape data representing a predetermined processing shape; a three-dimensional shape rotation unit that rotates the three-dimensional shape such that the processing surface in the three-dimensional shape is aligned with a desired processing axis direction; a depth map conversion unit that generates a depth map, which is obtained by orthographically projecting a two-dimensional image onto a plane perpendicular to the desired processing axis direction, and adding depth information of the desired processing axis direction extracted from the three-dimensional shape data; and an inference unit that constructs a processing axis direction indication that includes learning processing axis direction indication information, a learning depth map extracted from learning three-dimensional shape data according to the learning processing axis direction indication, and processing feasibility information recorded by machine learning that describes the actual processing feasibility based on the learning processing axis direction indication and the learning three-dimensional shape data, and determines the processing feasibility of the predetermined processing shape by inference using the desired processing axis direction and the depth map generated by the depth map conversion unit.

[0014] The processing feasibility determination method disclosed herein is a computer-executed processing feasibility determination method, which includes the following steps: extracting a three-dimensional shape from three-dimensional shape data representing a predetermined processing shape; rotating the three-dimensional shape such that the processing surface in the three-dimensional shape is aligned with a desired processing axis direction; generating a depth map, which is a two-dimensional image obtained by orthographically projecting the processing surface onto a plane perpendicular to the desired processing axis direction, and adding depth information of the desired processing axis direction extracted from the three-dimensional shape data; and constructing a processing feasibility determination method by using a learning processing axis direction indicator, a learning depth map extracted from the learning three-dimensional shape data according to the learning processing axis direction indicator, and processing feasibility information recorded by machine learning that describes the actual processing feasibility based on the learning processing axis direction indicator and the learning three-dimensional shape data, and determining the processing feasibility of the predetermined processing shape by inference from the desired processing axis direction and the depth map.

[0015] The machining feasibility determination procedure disclosed herein causes a computer to perform the following steps: extracting a three-dimensional shape from three-dimensional shape data representing a predetermined machining shape; rotating the three-dimensional shape such that the machining surface in the three-dimensional shape is aligned with a desired machining axis direction; generating a depth map, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the desired machining axis direction, and appending depth information of the desired machining axis direction extracted from the three-dimensional shape data; and constructing a machining feasibility determination procedure by using a machine learning method that learns a machining axis direction indication, a learning depth map extracted from the learning three-dimensional shape data according to the learning machining axis direction indication, and machining feasibility information recorded in the learning machining axis direction indication and the learning three-dimensional shape data, and by using inference from the desired machining axis direction and the depth map, to determine the machining feasibility of the predetermined machining shape.

[0016] Invention Effects

[0017] According to this disclosure, a machining feasibility determination apparatus, a machining feasibility determination method, and a machining feasibility determination program are provided that determine whether a cutting process is feasible by using a depth map from the machining axis direction to suppress the number of learning data and the computational load. Furthermore, a machining feasibility learning apparatus, a machining feasibility learning method, and a machining feasibility learning program are provided that perform the learning. Attached Figure Description

[0018] [ Figure 1 (a) is a functional structure diagram of the processing feasibility determination device of Embodiment 1, and (b) is a functional structure diagram of the processing feasibility learning device of Embodiment 1.

[0019] [ Figure 2 [Illustration] is a hardware structure diagram showing the processing feasibility determination device and processing feasibility learning device of Embodiment 1.

[0020] [ Figure 3 (a), (b), and (c) are illustrations illustrating cutting processes performed using milling machines or machining centers.

[0021] [ Figure 4 (a), (b), (c), (d), and (e) are explanatory diagrams illustrating situations where processing is not possible.

[0022] [ Figure 5 [ ] is a flowchart illustrating an example of the determination process of the processing capability determination device after learning is completed in Implementation 1.

[0023] [ Figure 6 (a) is a schematic diagram showing an example of a 3D shape extracted from 3D CAD data, and (b) is a diagram based on... Figure 6 An example of a depth map generated from a 3D shape, shown in (a).

[0024] [ Figure 7 (a) is a schematic diagram showing an example of a tool accessing the raw material from the direction of the machining axis, and (b) is an explanatory diagram showing that the shape information of the machining surface can be represented by a depth map through orthographic projection.

[0025] [ Figure 8 [] is a functional structure diagram of the processing feasibility determination device in Embodiment 2.

[0026] [ Figure 9 (a) is an explanatory diagram illustrating a part that is difficult to cut in the current machining axis direction, and (b) is an explanatory diagram showing a situation where it can be machined by changing the machining axis direction.

[0027] [ Figure 10 [ ] is a flowchart illustrating an example of the determination process of the processing capability determination device after learning is completed in Implementation 2.

[0028] [ Figure 11 (a) is a functional structure diagram of the processing feasibility determination device of Embodiment 3, and (b) is a functional structure diagram of the processing feasibility learning device of Embodiment 3.

[0029] [ Figure 12 [] is a functional structure diagram representing the processing capability learning device of embodiment 4.

[0030] [ Figure 13 (a) is a schematic diagram showing an example of a depth map, and (b) is a schematic diagram showing a depth map. Figure 13 The depth map of (a) is a schematic diagram of an example of a heat map of a part that is not suitable for processing, which corresponds to the processing availability information.

[0031] [ Figure 14 [ ] is a functional structure diagram of the processing feasibility determination device in embodiment 5.

[0032] [ Figure 15 (a) is an explanatory diagram showing the predetermined shape for processing relative to the shape of the raw material; (b) is an explanatory diagram showing the case where the processing surface / processing axis direction recording unit records the processing axis direction in the processing surface division unit's list of processing surfaces; (c) is a diagram showing the shape obtained by the three-dimensional shape rotation unit. Figure 15 (a) is an explanatory diagram showing the raw material shape rotated 90° to the left in the figure, and (d) is an explanatory diagram showing the machining surface / machining axis direction recording unit recording the machining axis direction in the machining surface list when the raw material shape is rotated 90°.

[0033] [ Figure 16[ ] is a flowchart illustrating an example of the determination process of the processing capability determination device after learning is completed in Implementation 5. Detailed Implementation

[0034] Hereinafter, the processing feasibility determination apparatus and processing feasibility learning apparatus according to embodiments will be described with reference to the accompanying drawings. The following embodiments are merely examples, and the embodiments can be appropriately combined and modified.

[0035] Implementation Method 1

[0036] Figure 1 (a) is a functional structure diagram showing the processing feasibility determination device 100 of Embodiment 1. The processing feasibility determination device 100 includes: a three-dimensional shape extraction unit 10, which extracts the processed shape information (processing predetermined shape) of the raw material as the processing object from the input three-dimensional CAD data; a three-dimensional shape rotation unit 12, which, according to the input processing axis instruction, orients the surface to be processed (hereinafter referred to as "processing surface") in the extracted shape information toward the desired processing axis direction (e.g., the Z-axis direction); and a depth map conversion unit 14, which converts the shape information that orients the processing surface toward the processing axis direction to generate a depth map. Figure 6 (b) shows a depth map along the machining axis direction; and an inference model 16 that determines whether machining is permissible based on the generated depth map and outputs the determination result. As described later, the inference model 16 is constructed by learning a mathematical model such as a CNN through machine learning, which uses a learning machining axis direction indicator prepared in advance for learning, a learning depth map extracted from the learning three-dimensional shape data according to the learning machining axis direction indicator, and machining permissibility information recording whether actual machining is permissible according to the learning machining axis direction indicator and the learning three-dimensional shape data.

[0037] Figure 1 (b) is a functional structure diagram showing the processing feasibility learning device 110 of Embodiment 1. In addition to the three-dimensional shape extraction unit 10, three-dimensional shape rotation unit 12, depth map conversion unit 14 and inference model 16 described above, the processing feasibility learning device 110 also includes a shape / processing axis direction database (DB) 20, a processing feasibility information DB24 storing processing feasibility information as supervision data, and a learner 22 that uses the processing feasibility information to update the inference model 16.

[0038] The shape / machining axis direction DB20 includes shape information DB20A, which stores 3D CAD data as 3D shape data for learning, and machining axis direction indication DB20B, which stores the machining axis direction as an indication for learning.

[0039] The shape information DB20A, machining axis direction indicator DB20B, and machining success / failure information DB24 each store data from past machining examples. For instance, the 3D CAD data stored in shape information DB20A is the data of the 3D shape used for learning in actual machining, and the machining axis direction indicator stored in machining axis direction indicator DB20B is also data of the machining axis direction indicator used for learning in actual machining. Furthermore, machining success / failure information DB24 stores the success or failure of the machining result using the 3D CAD data stored in shape information DB20A and the machining axis direction indicator stored in machining axis direction indicator DB20B. In addition, the machining success / failure information for learning stored in machining success / failure information DB24 corresponds to the 3D CAD data stored in shape information DB20A and the machining axis direction indicator stored in machining axis direction indicator DB20B, respectively. In Implementation 1, the learner 22 compares the result of the inference model 16 judging the learning depth map obtained by using the three-dimensional CAD data and machining axis direction indication as past machining case data, with the machining feasibility information corresponding to the three-dimensional CAD data and machining axis direction indication used by the inference model 16 for judgment, thereby evaluating the inference model 16 and updating the inference model 16.

[0040] Figure 2 This is a hardware structure diagram showing the processing approval / disapproval determination device 100 and processing approval / disapproval learning device 110 according to Embodiment 1. The processing approval / disapproval determination device 100 and the processing approval / disapproval learning device 110 are composed of a computer including a processor 210, a storage device 220, and an input / output interface 230. The processing approval / disapproval determination device 100 may also be composed of multiple computers.

[0041] The processor 210 is an integrated circuit (IC) that performs computational processing. Specifically, the processor 210 may be a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or a GPU (Graphics Processing Unit). The processor 210 can render 3D CAD data, change the orientation of the imaged 3D CAD data to any machining axis direction, and function as the 3D shape extraction unit 10, the 3D shape rotation unit 12, and the depth map conversion unit 14 by executing a program that generates a depth map based on the 3D CAD data. Furthermore, the processor 210 functions as a learner 22 by executing a machine learning program that uses supervised data; the result of this machine learning is the construction of an inference model 16 that determines whether machining is feasible. In Embodiment 1, by executing the machine learning program that determines whether machining is feasible, the processor 210 functions as a learner 22 that updates the 3D shape extraction unit 10, the 3D shape rotation unit 12, the depth map conversion unit 14, and the inference model 16. Furthermore, in Embodiment 1, after machine learning, the processor 210 functions as a 3D shape extraction unit 10, a 3D shape rotation unit 12, a depth map conversion unit 14, and an inference model 16 by processing the approval / disapproval procedure. Additionally, the approval / disapproval procedure and the approval / disapproval learning procedure are provided, for example, by a recording medium on which they are recorded.

[0042] The storage device 220 is composed of volatile storage devices such as RAM (Random Access Memory), non-volatile storage devices such as ROM (Read Only Memory), HDD (Hard Disk Drive), or flash memory.

[0043] Input / output interface 230 is a port connecting input device 300 and output device 310. Specifically, input / output interface 230 may be a USB (Universal Serial Bus) terminal, an IEEE 1394 terminal, or a Thunderbolt terminal, and may also include communication interfaces such as Ethernet. Input device 300 may be a touch panel, keyboard, mouse, etc. Output device 310 may be a monitor or printer, etc. Furthermore, during learning, shape / machining axis direction DB20 and machining availability information DB24 are connected to the aforementioned input / output interface 230. Shape / machining axis direction DB20 and machining availability information DB24 may also be stored in storage device 220.

[0044] Figure 3 (a), (b), and (c) are illustrations illustrating cutting processes performed using cutting machines such as milling machines or machining centers. Figure 3 (a) is an explanatory diagram of planar machining of a planar cutting surface 42 of a raw material 40 using a tool 30 such as a milling cutter. In planar machining, the tool 30 is moved in a direction parallel to the plane of the raw material 40, which is perpendicular to the machining axis, as shown by the arrow, while cutting the cutting surface 42.

[0045] Figure 3 (b) is an explanatory diagram of side machining of the raw material 40 using a tool 32 such as a corner milling cutter. In side machining, the cutting surface 44 is cut while the tool 32 is moved in the direction of the arrow.

[0046] Figure 3 (c) is an explanatory diagram of hole / cavity machining of raw material 40 by cutting a tool 34, such as an end mill, along the machining axis. In hole / cavity machining, the tool 34 is fed out along the machining axis direction indicated by the arrow to cut the cutting surface 46 of the raw material 40.

[0047] In Implementation 1, it is also determined whether other processing methods described above are permissible. For example, it can also be determined whether a three-dimensional shape can be formed by electrical discharge processing, which processes the raw material by using an electric arc discharge between an electrode and the raw material.

[0048] Figure 4 (a), (b), (c), (d), and (e) are illustrations showing cases where processing is not possible. Figure 4 (a) indicates tool interference where the portion of tool 34 other than the cutting edge interferes with the raw material 40. Figure 4 (b) indicates a situation where tool 34 is unsuitable for processing the shape of raw material 40. Figure 4 In the cases shown in (a) and (b), the problem can be solved by changing tool 34 to a tool suitable for the processing rules, so that the processing is possible can be determined even without using the processing possibility determination device 100 of embodiment 1.

[0049] Figure 4 (c) indicates a situation where the cutting surface 44 cannot be accurately cut due to vibration of tool 32, etc. Figure 4 (d) indicates a situation where tool 34 is bent and cannot accurately cut raw material 40. Figure 4 (e) represents an example where the processed shape of raw material 40 is complex, making it difficult to make judgments based on rules during processing. Figure 4In the cases shown in (c), (d), and (e), the problem cannot be solved based on processing rules such as tool changes. Therefore, in the past, skilled technicians judged whether processing was feasible based on their experience. In Implementation 1, the feasibility of processing is determined by a processing feasibility determination device 100 that has been trained, without relying on skilled technicians. Figure 4 Can the processing be carried out under the conditions shown in (c), (d), and (e)?

[0050] Figure 5 This is a flowchart illustrating an example of the determination process of the processing capability determination device 100 after learning is completed in Embodiment 1. In step S101, three-dimensional CAD data as shape information is input to the three-dimensional shape extraction unit 10.

[0051] In step S102, the three-dimensional shape extraction unit 10 renders the input three-dimensional CAD data and extracts the three-dimensional shape. Figure 6 (a) is a schematic diagram showing an example of a three-dimensional shape extracted from three-dimensional CAD data.

[0052] In step S103, the three-dimensional shape rotation unit 12 rotates the three-dimensional shape extracted in step S102 according to the machining axis direction indication. In Embodiment 1, as an example, the machining surface of the raw material 40 is oriented towards the Z-axis direction. More specifically, the machining surface of the raw material 40 is directly aligned with the Z-axis direction. The machining axis direction indication can be preset to the Z-axis or input in step S101.

[0053] In step S104, the depth map conversion unit 14 generates a depth map that makes the machining surface oriented in the Z-axis direction, forming a three-dimensional shape. Figure 6 (b) is based on Figure 6 Example of a depth map generated from a 3D shape shown in (a). A depth map is 2D image data generated by orthographically projecting a 3D shape with the machining surface facing the Z-axis onto a plane perpendicular to the Z-axis. The depth map includes Z-axis depth information in the 2D image data representing the machining surface. The Z-axis depth information is extracted from the 3D CAD data, which is the input shape information. Figure 6 In (b), depth information is either hue or brightness. In a depth map, lighter areas are represented by brighter hues such as yellow, while darker areas are represented by darker hues such as dark blue. When depth is represented by a single color such as gray, lighter areas are represented by high brightness, and darker areas by low brightness.

[0054] In milling machines and other cutting machines, such as Figure 7As shown in (a), tool 32 approaches the raw material 40 from the machining axis direction (Z-axis direction) to perform machining. Therefore, the important information needed to determine whether machining is feasible is the shape information of the machining surface as viewed from the Z-axis direction; the shape information of the sides or back of the raw material 40 other than the machining surface is not needed at this time. Figure 7 As shown by the arrow in (b), the shape information of the processed surface can be represented by a depth map through orthographic projection. Therefore, the depth map, which is a two-dimensional data, can be used instead of the data representing the three-dimensional shape to determine whether the processing is feasible.

[0055] In step S105, the inference model 16 infers whether machining is feasible. As described above, in Embodiment 1, the inference model 16 learns from 3D CAD data, machining axis direction indicators, and machining feasibility information corresponding to the 3D CAD data and machining axis direction indicators, respectively. After learning, the inference model 16 infers whether machining is feasible based on the learned machining feasibility information and outputs the inference result.

[0056] In step S106, the inference model 16 determines whether the inference result output in step S105 is manufacturable. If the result is manufacturable in step S106, the process moves to step S107; if the result is unmanufacturable, the process moves to step S108.

[0057] In step S107, the inference model 16 outputs a processable determination to the output device 310 and ends the process. In step S108, the inference model 16 outputs an unprocessable determination to the output device 310 and ends the process.

[0058] As explained above, in Embodiment 1, two-dimensional data, namely a depth map, generated based on the three-dimensional shape is used to determine whether processing is feasible. Since the data is two-dimensional, its size is smaller than the three-dimensional shape; therefore, the amount of learning data and computational load during the learning of the inference model 16 can be reduced. Furthermore, the computational load for determining whether processing is feasible in the learned processing feasibility determination device 100 can be reduced.

[0059] When implementing the inference model 16 for determining whether a cutting operation is feasible, in models that input voxel data as in the past, processing data representing three-dimensional shapes requires a large amount of computation and learning data. However, in Implementation 1, by inputting a depth map as two-dimensional data, the amount of computation and learning data can be reduced. In addition, if the number of learning data is the same, the number of parameters of the neural network is smaller, and correspondingly, higher determination accuracy can be achieved than before.

[0060] Implementation Method 2

[0061] Next, the processing feasibility determination device 120 of Embodiment 2 will be described. Figure 8The difference between the machining feasibility determination device 120 of Embodiment 2 and Embodiment 1 is that it includes a machining axis direction adjustment unit 26 that adjusts the machining axis direction based on the machining feasibility determination output by the inference model 16, and a three-dimensional shape rotation unit 28 that rotates the three-dimensional shape in the machining axis direction adjusted by the machining axis direction adjustment unit 26. However, other structures are the same as in Embodiment 1, so structures identical to those in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted. In addition, the hardware structure of Embodiment 2 is the same as that of Embodiment 1, so detailed descriptions are omitted. However, in Embodiment 2, the processor 210 functions as the three-dimensional shape extraction unit 10, the three-dimensional shape rotation unit 28, the depth map conversion unit 14, and the inference model 16, and also functions as the machining axis direction adjustment unit 26.

[0062] In Implementation Method 1, the machining axis is set to the Z-axis to determine whether machining is feasible. However, when determining whether machining is feasible, there are sometimes... Figure 9 The undercut shape 38 shown in (a) is a part that is difficult to cut in the current machining axis direction. Even in Figure 9 In the case shown in (a), as Figure 9 As shown in (b), machining can also be performed according to the machining axis direction. In Embodiment 2, when design data is provided, the machining axis direction (orientation of the raw material 40) is automatically changed while repeatedly determining whether machining is feasible, thereby verifying all machining axis directions. For example, as Figure 8 As shown, the machining axis direction is changed by rotating the raw material 40 in the vertical direction 50 or the horizontal direction 52. In some cutting machines, there are models that can change the machining axis while the raw material 40 is fixed; however, in these models, rotating the raw material 40 to change the machining surface allows for simpler and faster machining control compared to changing the machining axis. In Embodiment 2 and Embodiment 5 (described later), the actual machining axis on which the tool is installed is fixed, for example, in the Z-axis direction, and the three-dimensional shape is rotated in the same way as when changing the machining axis from the Z-axis, so that the machining surface is rotated.

[0063] Figure 10 This is a flowchart illustrating an example of the determination process of the processing capability determination device 120 after learning is completed in Implementation 2. Figure 10 The flowchart shown is Figure 5 The flowchart of Embodiment 1 differs in that step S203 is provided instead of step S103 in Embodiment 1, and includes step S204 for determining whether the machining axis direction has been changed in all directions and step S205 for changing the machining axis direction indication. However, the other steps are the same as in Embodiment 1, therefore, the processes that are the same as in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted.

[0064] In step S203, the three-dimensional shape rotation unit 28 rotates the three-dimensional shape extracted in step S102 according to the machining axis direction indication. In Embodiment 2, the machining axis direction indication is preset to the Z-axis in the initial determination process. The machining axis direction indication can also be input in step S101. As will be described later, in Embodiment 2, if the determination result is that machining cannot be performed using the current machining axis direction indication, the machining axis direction adjustment unit 26 changes the machining axis direction indication, and in step S203, the three-dimensional shape is rotated according to the changed machining axis direction indication.

[0065] If the determination result in step S106 is that the material cannot be processed (unmanufacturable case), the machining axis direction adjustment unit 26 determines in step S204 whether the feasibility of processing has been studied in all machining axis directions. If the raw material 40 is considered as a cuboid, there are a total of 6 machining axis direction indications. The machining axis direction adjustment unit 26 registers the Z-axis setting in the initial determination and the machining axis direction indication changed in the subsequent step S205 among these 6 machining axis direction indications to the storage device 220. In step S204, the machining axis direction adjustment unit 26 determines, by referring to the storage device 220, whether the feasibility of processing has been studied in all machining axis directions.

[0066] If, in step S204, it is determined that the feasibility of machining has been investigated in all machining axis directions, the process is transferred to step S108. In step S108, similar to Embodiment 1, the unmachinable determination is output to the output device 310 and the process ends.

[0067] If it is determined in step S204 that the feasibility of machining has not been studied in all machining axis directions, the process is transferred to step S205. In step S205, the machining axis direction adjustment unit 26 changes the machining axis direction indication and inputs the changed machining axis direction indication to the three-dimensional shape rotation unit 28.

[0068] In step S203, the three-dimensional shape is rotated according to the modified machining axis direction indication. Specifically, the three-dimensional shape is rotated such that the new machining surface when machining the three-dimensional shape in the machining axis direction indicated by the machining axis direction indication input from the machining axis direction adjustment unit 26 is aligned with the desired machining axis direction (Z-axis direction in Embodiment 2). In the subsequent process, the depth map conversion unit 14 generates a new depth map by orthographically projecting the new machining surface onto a plane perpendicular to the Z-axis direction. The inference model 16 uses the new depth map to determine whether machining the predetermined shape is feasible. If the determination result in step S106 is that it is feasible, the feasible determination is output to the output device 310 in step S107 and the process ends.

[0069] As explained above, in Embodiment 2, when design data is provided, the machining axis direction is automatically changed while repeatedly determining whether machining is feasible, thereby verifying all machining axis directions. As a result, even in cases where the feasibility of machining varies depending on the machining axis direction, by determining all machining axis directions, it is possible to determine at what angle the part shape shown in the design data cannot be machined. Furthermore, in Embodiment 2, the feasibility of machining can also be determined for the shapes of the sides and back, which are lost as information when converting the three-dimensional shape into a depth map.

[0070] Implementation Method 3

[0071] Next, the processing feasibility determination device 130 and processing feasibility learning device 140 of Embodiment 3 will be described. Figure 11 The difference between the machining feasibility learning device 140 of Embodiment 3 shown in (b) and Embodiment 1 is that the learner 22 compares the result obtained by the inference model 58 in judging the depth map obtained by using 3D CAD data as past machining example data and machining axis direction indication, the machining axis direction indicated by the machining axis direction indication, the material information of the raw material 40 stored in the machining information DB20C included in the shape / machining axis direction / machining information DB48, tool information (tool type, tool material, tool diameter, and tool length, etc.), and cutting parameters (tool feed rate and tool rotation speed, etc.) with the machining feasibility information stored in the machining feasibility information DB24. Therefore, the inference model 58 is evaluated and updated. Furthermore, Figure 11 The difference between the machining feasibility determination device 130 of Embodiment 3 shown in (a) and Embodiment 1 is that the inference model 58 determines the depth map obtained by the learned inference model 58 using 3D CAD data and machining axis direction indication, the machining axis direction indicated by the machining axis direction indication, and machining information including material information of raw material 40, tool information, and cutting parameters. However, other structures are the same as in Embodiment 1, therefore, structures identical to those in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted. In addition, the hardware structure of Embodiment 3 is the same as that of Embodiment 1, therefore detailed descriptions are omitted.

[0072] The material of raw material 40, tool information, and cutting parameters all significantly influence the determination of whether machining is feasible. For example, using inappropriate tools for difficult-to-machine materials such as stainless steel, or having an inappropriate feed rate or spindle speed relative to the difficult-to-machine material, can all negatively impact the machining process. Figure 4 As shown in (c), it is possible that the cutting surface 44 cannot be accurately cut due to vibration of tool 32, or as... Figure 4As shown in (d), it is possible that tool 34 may bend and fail to accurately cut the raw material 40.

[0073] Thus, even with the same design data, the feasibility of machining varies depending on the material information of the raw material 40, the tool information, and the cutting parameters. In the machining feasibility determination device 130 and machining feasibility learning device 140 of Embodiment 3, at least one of the material information of the raw material 40, the tool information, and the cutting parameters is added to the input of the inference model 58 to determine whether machining is feasible.

[0074] As explained above, according to Embodiment 3, even when the feasibility of processing varies not only based on the design data but also on the material of the raw material 40, the tool, and the cutting parameters, an accurate determination of feasibility can be made.

[0075] Implementation Method 4

[0076] Next, the processing capability learning device 150 of Embodiment 4 will be described. Figure 12 The difference between the processing feasibility learning device 150 in Embodiment 4 and Embodiment 1 is that a heat map representing the unsuitable processing area is stored in the processing feasibility information DB56, and the learner 54 uses this heat map as supervisory data to update the inference model 16. However, the other structures are the same as in Embodiment 1; therefore, structures identical to those in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted. Furthermore, the processing feasibility determination device constructed by the processing feasibility learning device 150 is identical in structure to the processing feasibility determination device 100 in Embodiment 1, except that it updates the learning of the inference model 16 by using the heat map representing the unsuitable processing area as processing feasibility information and outputs a determination result that the inference model 16 has determined the unsuitable processing area; therefore, detailed descriptions are omitted. Additionally, the hardware structure of Embodiment 4 is the same as that of Embodiment 1; therefore, detailed descriptions are omitted.

[0077] Figure 13 (a) is a schematic diagram showing an example of a depth map. Figure 13 (b) indicates that it is related to Figure 13 The depth map of (a) is a schematic diagram of an example of a heat map of a part that is not suitable for processing, which corresponds to the processing availability information.

[0078] exist Figure 13 In the depth diagram shown in (a), a sharp edge 60, known as a pin angle, is recorded at the location corresponding to the concave portion. During machining, the concave portion is cut using tools such as end mills. However, since the end mills and other tools rotate while cutting, it is impossible to avoid rounding the corners of the raw material 40, and the pin angle like edge 60 cannot be reproduced.

[0079] In implementation 4, a heatmap corresponding to the depth map is used as supervision data for the inference model 16 to learn. The heatmap is a mapping of the depth map to the depth map. Figure 13 The depth map, as shown in (a), corresponds to a range divided into a grid with specified cell sizes, and the values ​​representing processing possibilities are associated with each cell. For example, in Figure 13 In (b), the cells shown in white represent workable parts 62 with values ​​close to 0, and the cells shown in black represent unworkable parts 66 with values ​​close to 1. Furthermore, in Figure 13 In (b), the cells shown in gray represent the slightly more difficult middle section 64, which is associated with values ​​between 0 and 1. The values ​​associated with each cell in the heatmap, which serves as information on whether the processing is feasible, are determined based on the results of actual manual verification of whether the work is feasible.

[0080] In Implementation 4, the learner 54 compares the result of the inference model 16 judging the depth map obtained by the inference model 16 using the 3D CAD data and the machining axis direction indication as past machining case data, with the heat map as machining feasibility information corresponding to the 3D CAD data and machining axis direction indication used by the inference model 16 to judge, thereby evaluating the inference model 16 and updating the inference model 16.

[0081] The processing suitability determination device constructed through the above learning outputs a heat map, which quantifies the degree of processing unsuitability in cell units, as the processing suitability determination result. As a result, according to Embodiment 4, it is possible to accurately indicate which part of the three-dimensional shape or depth map is the processing unsuitability part.

[0082] Implementation Method 5

[0083] Next, the processing feasibility determination device 160 of Embodiment 5 will be described. Figure 14The processing feasibility determination device 160 of Embodiment 5 differs from Embodiment 1 in that it includes: a processing surface segmentation unit 70, which lists all processing surfaces of the raw material 40 and creates a processing surface list based on the three-dimensional shape extracted by the three-dimensional shape extraction unit 10; a processing surface / processing axis direction recording unit 74, which is input into the processing surface list created by the processing surface segmentation unit 70 and records the processing axis directions of the processing surfaces determined to be processable in the processing feasibility determination output by the inference model 16 in the processing surface list; an end determination unit 76, which performs an end determination when all processing surfaces recorded in the processing surface list are determined to be processable, or when the processing feasibility in all processing axis directions has been verified, and outputs the processing feasibility determination for the three-dimensional shape and the processing axis directions of each processing surface; and a processing axis direction adjustment unit 78, which outputs an indication of processing axis directions other than the processing axis directions that have been verified when the end determination unit 76 determines that all processing axis directions have not been verified; and a three-dimensional shape rotation unit 72 rotates the three-dimensional shape in the processing axis direction adjusted by the processing axis direction adjustment unit 78. However, the other structures are the same as in Embodiment 1. Therefore, structures identical to those in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted. Furthermore, the hardware structure of Embodiment 5 is the same as that of Embodiment 1, so detailed descriptions are omitted. However, in Embodiment 5, the processor 210 functions as a three-dimensional shape extraction unit 10, a three-dimensional shape rotation unit 72, a depth map conversion unit 14, and an inference model 16. It also functions as a machining surface segmentation unit 70, a machining surface / machining axis direction recording unit 74, an end determination unit 76, and a machining axis direction adjustment unit 78.

[0084] Figure 15 (a) is an explanatory diagram showing the pre-processed shape 82 relative to the raw material shape 80. The pre-processed shape 82 includes processing surfaces (1), (2), (3), (4), (5), (6), and (7), but... Figure 15 In the state shown in (a), the surface that can be processed by tool 34 becomes machinable surface 84, and the machinable surfaces (5), (6), and (7) cannot be processed.

[0085] Figure 15 (b) is an explanatory diagram showing the case where the machining surface / machining axis direction recording unit 74 records the machining axis direction in the machining surface division unit 70's machining surface list. Figure 15 In the case shown in (a), machining of surfaces (1) to (4) is possible; therefore, the machining axis direction in the machining surface list is, for example, 0° representing the Z-axis direction. However, in Figure 15 In the case shown in (a), machining of surfaces (5) to (7) is not possible, therefore, the machining axis direction in the machining surface list is empty.

[0086] Figure 15 (c) is achieved through the three-dimensional shape rotation part 72. Figure 15 The illustration shows the raw material shape 80 as shown in (a) being rotated 90° to the left in the figure. By rotating the raw material shape 80 by 90°, it is possible to process the processing surfaces (5) to (7) that could not be processed before the rotation.

[0087] Figure 15 (d) is an explanatory diagram showing the machining surface / machining axis direction recording unit 74 recording the machining axis direction in the machining surface list when the raw material shape is rotated 80° by 90°. (See diagram below.) Figure 15 As shown in (c), with the raw material shape rotated 80° by 90°, the machining surfaces (5) to (7) can be machined. Therefore, the machining axis direction of the machining surface list records the rotation angle relative to the Z-axis, which is +90°.

[0088] Figure 16 This is a flowchart illustrating an example of the determination process of the processing capability determination device 160 after learning is completed in Implementation 5. Figure 16 The flowchart shown is Figure 5 The flowchart of Embodiment 1 differs in that it includes step S302 instead of step S103 in Embodiment 1, comprising: step S301, extracting machining surfaces from the three-dimensional shape extracted by the three-dimensional shape extraction unit 10 and creating a machining surface list; step S304, recording the machining surfaces and machining axis directions determined as machineable by the machining surface / machining axis direction recording unit 74 in the machining surface list; step S305, the end determination unit 76 determining whether all machining surfaces are determined to be machineable; step S306, outputting the machineability determination and the machining axis direction of each machining surface; step S307, determining whether the machining axis direction has been changed for all directions; and step S308, changing the machining axis direction indication. However, the other processes are the same as in Embodiment 1, therefore, the processes that are the same as in Embodiment 1 are labeled with the same reference numerals as in Embodiment 1, and detailed descriptions are omitted.

[0089] In step S301, the machining surface segmentation unit 70 extracts the machining surface from the three-dimensional shape extracted by the three-dimensional shape extraction unit 10 in step S102, and manufactures the machining surface. Figure 15 (b) and Figure 15 The list of machined surfaces shown in (d) is as follows.

[0090] In step S302, the three-dimensional shape rotation unit 72 rotates the three-dimensional shape extracted in step S102 according to the machining axis direction indication. In Embodiment 5, the machining axis direction indication is preset to the Z-axis in the initial determination process. The machining axis direction indication can also be input in step S101. As will be described later, in Embodiment 5, if the determination result is that machining cannot be performed using the current machining axis direction indication, the machining axis direction adjustment unit 78 changes the machining axis direction indication, and in step S302, the three-dimensional shape is rotated according to the changed machining axis direction indication.

[0091] In step S304, based on the machining feasibility determination output by the inference model 16 in step S105, the machining surface / machining axis direction recording unit 74 records the machining surfaces and machining axis directions that are determined to be machineable in the machining surface list.

[0092] In step S305, the end determination unit 76 determines whether all machining surfaces are deemed machinable. The end determination unit 76 is, for example, as follows: Figure 15 As shown in (d), if all columns in the machining axis direction of the machining surface list are filled with meaningful angle information such as 0° or +90°, it is determined that all machining surfaces are machinable. If it is determined in step S305 that all machining surfaces are machinable, the process moves to step S306; if it is determined that none of the machining surfaces are machinable, the process moves to step S307.

[0093] In step S306, the final machinability determination of the three-dimensional shape and the machining axis direction of each machining surface are output to the output device 310, and the processing ends. The machinability determination and machining axis direction information output in step S306 are, for example,... Figure 15 The “d” indicates a list of machining surfaces corresponding to the machining axis direction of each machining surface.

[0094] If, in step S305, it is determined that none of the machining surfaces are deemed machinable, the termination determination unit 76 determines in step S307 whether the machining axis direction has been changed in all directions. If the raw material 40 is considered a cuboid, there are a total of six machining axis direction indications. The machining axis direction adjustment unit 78, similar to Embodiment 2, registers the Z-axis setting in the initial determination and the machining axis direction indication changed in the subsequent step S308 among these six machining axis direction indications in the storage device 220. In step S307, the termination determination unit 76 determines, by referring to the storage device 220, whether machining feasibility has been studied in all machining axis directions. Alternatively, a machining axis change history column can be set up separately in the machining surface list, and the termination determination unit 76 can determine whether machining feasibility has been studied in all machining axis directions by referring to this column.

[0095] If it is determined in step S307 that the machining axis direction has been changed in all directions, the process moves to step S108. In step S108, similar to Embodiment 1, the unmachinable determination is output to the output device 310 and the process ends.

[0096] If it is determined in step S307 that no machining axis direction has been changed in all directions, the process proceeds to step S308. In step S308, the machining axis direction adjustment unit 78 changes the machining axis direction indication and inputs the changed machining axis direction indication to the three-dimensional shape rotation unit 72.

[0097] In step S302, the three-dimensional shape rotation unit 72 rotates the three-dimensional shape according to the changed machining axis direction indication. In the subsequent process, the depth map is converted and the machining feasibility is inferred. If it is determined in step S305 that all machining surfaces are machined, in step S306, the machining feasibility determination and the machining axis direction of each machining surface are output to the output device 310 and the process ends.

[0098] As explained above, in Embodiment 5, similarly to Embodiment 2, even when the shape's machining capability varies depending on the machining axis direction, by determining all machining axis directions, it is possible to determine at what angle the part shape shown in the design data cannot be machined. Furthermore, by clarifying not only whether machining is possible but also from which direction each machining surface can be machined, preparation for actual machining can be easily made.

[0099] Furthermore, the inference section in the claims is equivalent to the inference models 16 and 58 described in the detailed description of the invention.

[0100] Label Explanation

[0101] 10: 3D shape extraction unit; 12: 3D shape rotation unit; 14: Depth map conversion unit; 16: Inference model; 20A: Shape information DB; 20B: Machining axis direction indicator DB; 20C: Machining information DB; 22: Learner; 24: Machining availability information DB; 26: Machining axis direction adjustment unit; 30, 32, 34: Tools; 40: Raw materials; 54: Learner; 56: Machining availability information DB; 58: Inference model; 70: Machining surface segmentation unit; 72: 3D shape rotation unit; 74: Machining surface / machining axis direction recording unit; 76: End determination unit; 78: Machining axis direction adjustment unit; 100: Machining availability determination device; 110: Machining availability learning device; 120: Machining availability determination device; 130: Machining availability determination device; 140: Machining availability learning device; 150: Machining availability learning device; 160: Machining availability determination device.

Claims

1. A processing feasibility determination device, the processing feasibility determination device comprising: The three-dimensional shape extraction unit extracts the three-dimensional shape from the three-dimensional shape data representing the predetermined shape to be processed; A three-dimensional shape rotation unit that rotates the three-dimensional shape in such a way that the machining surface in the three-dimensional shape is aligned with the desired machining axis direction; The depth map conversion unit generates a depth map, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the desired machining axis direction, and adds depth information of the desired machining axis direction extracted from the three-dimensional shape data. as well as The inference unit is constructed using a learning machining axis direction indicator, a learning depth map extracted from the learning three-dimensional shape data according to the learning machining axis direction indicator, and machine learning that records information on whether actual machining is possible according to the learning machining axis direction indicator and the learning three-dimensional shape data. It determines whether machining the predetermined shape is possible by inferring the desired machining axis direction and the depth map generated by the depth map conversion unit.

2. The processing feasibility determination device according to claim 1, wherein, The machining feasibility determination device further includes a machining axis direction adjustment unit. If the inference unit outputs a determination that a machining surface directly opposite the desired machining axis direction is unmachinable, the machining axis direction adjustment unit outputs an indication of a machining axis direction different from the desired machining axis direction. The three-dimensional shape rotation unit rotates the three-dimensional shape such that the new machining surface when machining the three-dimensional shape in the machining axis direction indicated by the machining axis direction indicator input from the machining axis direction adjustment unit is aligned with the desired machining axis direction. The depth map conversion unit generates a new depth map by orthographically projecting the new machining surface onto a plane perpendicular to the desired machining axis direction. The inference unit uses the new depth map to determine whether the processing of the predetermined shape is feasible.

3. The processing feasibility determination device according to claim 1, wherein, The inference unit is constructed using machine learning, which employs the learning machining axis direction indicator, a learning depth map extracted from the learning 3D shape data according to the learning machining axis direction indicator, learning machining information, and machine capability information recording the actual machining capability based on the learning machining axis direction indicator, the learning 3D shape data, and the learning machining information. It determines the machining capability of a predetermined shape by inferring the desired machining axis direction, the depth map generated by the depth map conversion unit, and newly input machining information. The learning machining information comprises tool information including tool type, tool material, tool diameter, and tool length, information about the material of the workpiece, and cutting parameters including tool feed rate and tool rotation speed. The newly input machining information comprises tool information including tool type, tool material, tool diameter, and tool length, information about the material of the workpiece, and cutting parameters including tool feed rate and tool rotation speed.

4. The processing feasibility determination device according to claim 1, wherein, The processing feasibility information is a heat map representing the areas on the processing surface that are unsuitable for processing. The inference unit constructs a machine learning model using a learning machining axis direction indicator, a learning depth map extracted from the learning 3D shape data according to the learning machining axis direction indicator, and a heat map recording whether actual machining is feasible according to the learning machining axis direction indicator and the learning 3D shape data. The inference unit outputs the following determination result: by inferring the desired machining axis direction and the depth map generated by the depth map conversion unit, the unsuitable machining parts in the predetermined machining shape are determined.

5. The processing feasibility determination device according to claim 1, wherein, The processing feasibility determination device also includes: The processing surface segmentation unit lists all processing surfaces of the processing object from the three-dimensional shape extracted by the three-dimensional shape extraction unit and creates a processing surface list; The machining surface / machining axis direction recording unit records the machining axis direction of the machining surface that is determined to be machinable in the machining feasibility determination output by the inference unit in the machining surface list; The termination determination unit performs a termination determination in any of the following cases: all machined surfaces listed in the machined surface list have been determined to be machineable; or the machinedability in all machined axis directions has been verified. It then outputs the machinedability determination for the three-dimensional shape and the machined axis directions for each machined surface. The machining axis direction adjustment unit, if determined by the end determination unit that the feasibility of machining in all machining axis directions has not been verified, outputs an indication of the machining axis directions other than those whose machining feasibility has been verified. The three-dimensional shape rotation unit rotates the three-dimensional shape such that the new machining surface when machining the three-dimensional shape in the machining axis direction indicated by the machining axis direction indicator input from the machining axis direction adjustment unit is aligned with the desired machining axis direction. The depth map conversion unit generates a new depth map by orthographically projecting the new machining surface onto a plane perpendicular to the desired machining axis direction. The inference unit uses the new depth map to determine whether the processing of the predetermined shape is feasible.

6. A processability-learning device, the processability-learning device comprising: The 3D shape extraction unit extracts the learning 3D shape from the input learning 3D shape data; A three-dimensional shape rotation unit rotates the three-dimensional shape such that the machining surface in the three-dimensional shape is aligned with the learning machining axis direction indicated by the input learning machining axis direction indicator. The depth map conversion unit generates a learning depth map, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the learning machining axis direction, and adding depth information in the learning machining axis direction extracted from the learning three-dimensional shape data. as well as The inference unit is updated by a learner that performs machine learning using the learning machining axis direction indicator, the learning depth map, and machining feasibility information recorded according to the learning machining axis direction indicator and the learning 3D shape data. The processing feasibility learning device constructs the processing feasibility determination device of claim 1 through machine learning.

7. A processability-learning device, the processability-learning device comprising: The 3D shape extraction unit extracts the learning 3D shape from the input learning 3D shape data; A three-dimensional shape rotation unit rotates the three-dimensional shape such that the machining surface in the three-dimensional shape is aligned with the learning machining axis direction indicated by the input learning machining axis direction indicator. The depth map conversion unit generates a learning depth map, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the learning machining axis direction, and adding depth information in the learning machining axis direction extracted from the learning three-dimensional shape data. as well as The inference unit is updated via a learner that performs machine learning using the learning machining axis direction indicator, the learning depth map, the learning machining information, and machining feasibility information recorded based on the learning machining axis direction indicator, the learning 3D shape data, and the learning machining information. The learning machining information comprises tool information including tool type, tool material, tool diameter, and tool length; material information of the workpiece; and cutting parameters including tool feed rate and tool spindle speed. The processing feasibility learning device constructs the processing feasibility determination device as described in claim 3 through machine learning.

8. A processability-learning device, the processability-learning device comprising: The 3D shape extraction unit extracts the learning 3D shape from the input learning 3D shape data; A three-dimensional shape rotation unit rotates the three-dimensional shape such that the machining surface in the three-dimensional shape is aligned with the learning machining axis direction indicated by the input learning machining axis direction indicator. The depth map conversion unit generates a learning depth map, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the learning machining axis direction, and adding depth information in the learning machining axis direction extracted from the learning three-dimensional shape data. as well as The inference unit is updated via a learner that performs machine learning using the learned machining axis direction indicator, the learned depth map, and heatmaps representing unsuitable machining areas in actual machining according to the learned machining axis direction indicator and the learned 3D shape data. The processing feasibility learning device constructs the processing feasibility determination device of claim 4 through the machine learning.

9. A method for determining whether a process is permissible, executed by a computer, comprising the following steps: Extracting a 3D shape from 3D shape data representing a predetermined shape for processing; The three-dimensional shape is rotated so that the machining surface in the three-dimensional shape is aligned with the desired machining axis direction; A depth map is generated, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the desired machining axis direction, and adding depth information of the desired machining axis direction extracted from the three-dimensional shape data. as well as The machining feasibility of the predetermined shape is determined by using a machine learning method that employs a learning machining axis direction indicator, a learning depth map extracted from the learning 3D shape data according to the learning machining axis direction indicator, and a machine learning method that records information on whether actual machining is possible according to the learning machining axis direction indicator and the learning 3D shape data.

10. A processability determination procedure that causes a computer to perform the following steps: Extracting a 3D shape from 3D shape data representing a predetermined shape for processing; The three-dimensional shape is rotated so that the machining surface in the three-dimensional shape is aligned with the desired machining axis direction; A depth map is generated, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the desired machining axis direction, and adding depth information of the desired machining axis direction extracted from the three-dimensional shape data. as well as The machining feasibility of the predetermined shape is determined by using a machine learning method that employs a learning machining axis direction indicator, a learning depth map extracted from the learning 3D shape data according to the learning machining axis direction indicator, and a machine learning method that records information on whether actual machining is possible according to the learning machining axis direction indicator and the learning 3D shape data.

11. A computer-executed method for determining whether a process is learnable, the method comprising the following steps: Extract the learning 3D shape from the input learning 3D shape data; The three-dimensional shape is rotated such that the machining surface in the three-dimensional shape is aligned with the learning machining axis direction indicated by the input learning machining axis direction indicator. A learning depth map is generated, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the learning machining axis direction, and adding depth information in the learning machining axis direction extracted from the learning three-dimensional shape data; as well as The learner updates the inference unit that determines whether a predetermined shape can be machined. This learner performs machine learning using the learning machining axis direction indicator, the learning depth map, and machining feasibility information recorded based on the learning machining axis direction indicator and the learning 3D shape data. The processing feasibility determination device of claim 1 is constructed using the machine learning method described above.

12. A processability learning program that causes a computer to perform the following steps to construct the processability determination apparatus of claim 1: Extract the learning 3D shape from the input learning 3D shape data; The three-dimensional shape is rotated such that the machining surface in the three-dimensional shape is aligned with the learning machining axis direction indicated by the input learning machining axis direction indicator. A learning depth map is generated, which is a two-dimensional image obtained by orthographically projecting the machining surface onto a plane perpendicular to the learning machining axis direction, and adding depth information in the learning machining axis direction extracted from the learning three-dimensional shape data; as well as The learner updates the inference unit that determines whether a predetermined shape can be machined. The learner performs machine learning using the learning machining axis direction indicator, the learning depth map, and machining feasibility information recorded according to the learning machining axis direction indicator and the learning three-dimensional shape data.

Images