阅读说明：本技术 用于生成工具路径的方法以及装置 (Method and device for generating tool path ) 是由 中本圭一 桥本真由 高野和雅 武井胜彦 猪狩真二 于 2019-02-14 设计创作，主要内容包括：本方法具备进行机器学习的工序以及生成新的工具路径的工序。进行机器学习的工序包括：关于多个已知的加工物的各个,得到形状数据；得到多个加工面各自的几何信息；从多个工具路径模式中,得到针对多个加工面的各个选择的工具路径模式；以及使用多个已知的加工物的几何信息及工具路径模式,进行输入是加工面的几何信息且输出是加工面的工具路径模式的机器学习。生成新的工具路径的工序包括：得到对象加工物的形状数据；得到对象加工物的多个加工面各自的几何信息；以及使用对象加工物的几何信息,根据机器学习的结果,关于对象加工物的多个加工面的各个,生成工具路径模式。(The method includes a step of performing machine learning and a step of generating a new tool path. The process of performing machine learning includes: obtaining shape data for each of a plurality of known processed products; obtaining respective geometric information of a plurality of processing surfaces; obtaining a tool path pattern selected for each of the plurality of machining surfaces from the plurality of tool path patterns; and performing machine learning in which geometric information of the machining surface is input and a tool path pattern of the machining surface is output, using the geometric information and the tool path pattern of the plurality of known machined objects. The process of generating a new tool path includes: obtaining shape data of the object processing object; obtaining geometric information of each of a plurality of processing surfaces of the target processing object; and generating a tool path pattern for each of the plurality of processing surfaces of the target processing object based on the result of the machine learning using the geometric information of the target processing object.)

1. A method for generating a tool path in NC machining,

the method comprises:

performing machine learning based on information of a plurality of known processed objects having the generated tool path; and

generating a new tool path for the target workpiece based on the result of the machine learning,

the plurality of known processed products and the target processed product each have a plurality of processed surfaces,

the step of performing the machine learning includes:

obtaining shape data for each of the plurality of known processed objects;

obtaining geometric information of each of the plurality of machined surfaces for each of the plurality of known machined objects;

obtaining, for each of the plurality of known machined objects, a tool path pattern selected for each of the plurality of machined surfaces from a plurality of tool path patterns; and

performing machine learning in which geometric information of a machining surface is input and a tool path pattern of the machining surface is output, using the geometric information and the tool path pattern of the plurality of known machined objects,

the step of generating the new tool path includes:

obtaining shape data of the object processing object;

obtaining geometric information of each of the plurality of processing surfaces of the target processing object; and

generating a tool path pattern for each of the plurality of machined surfaces of the target machined object from a result of the machine learning using the geometric information of the target machined object.

2. The method of claim 1,

the plurality of tool path patterns includes at least a contour path, a scan line path, and a face path.

3. The method of claim 1,

the shape data of the plurality of known processed objects and the shape data of the target processed object are defined by an XYZ-axis coordinate system which is a three-dimensional orthogonal coordinate system,

the geometric information includes at least 1 of a type of the machining surface, a ratio of a maximum length in an X-axis direction to a maximum length in a Z-axis direction in each machining surface, a ratio of a maximum length in a Y-axis direction to a maximum length in a Z-axis direction in each machining surface, a ratio of a maximum length in a Z-axis direction in the entirety of the plurality of machining surfaces to a maximum length in a Z-axis direction in each machining surface, a ratio of a surface area in the entirety of the plurality of machining surfaces to a surface area in each machining surface, a long radius of the machining surface, a short radius of the machining surface, a Z component of a normal vector at a center of gravity of the machining surface, a maximum curvature of the machining surface, and a minimum curvature of the machining surface.

4. The method of claim 1,

a neural network is used in the machine learning.

5. An apparatus for generating a tool path in NC machining,

the device is provided with:

a processor; and

a display part for displaying the display position of the display part,

the processor is configured to perform:

performing machine learning based on information of a plurality of known processed objects having the generated tool path; and

generating a new tool path for the target workpiece based on the result of the machine learning,

the plurality of known processed products and the target processed product each have a plurality of processed surfaces,

performing the machine learning includes:

obtaining shape data for each of the plurality of known processed objects;

obtaining geometric information of each of the plurality of machined surfaces for each of the plurality of known machined objects;

obtaining, for each of the plurality of known machined objects, a tool path pattern selected for each of the plurality of machined surfaces from a plurality of tool path patterns; and

performing machine learning in which geometric information of a machining surface is input and a tool path pattern of the machining surface is output, using the geometric information and the tool path pattern of the plurality of known machined objects,

generating the new tool path comprises:

obtaining shape data of the object processing object;

obtaining geometric information of each of the plurality of processing surfaces of the target processing object; and

generating a tool path pattern with respect to each of the plurality of machined surfaces of the target machined object from a result of the machine learning using the geometric information of the target machined object,

assigning visually recognizable predetermined features to the plurality of tool path patterns respectively,

the processor recognizes the plurality of tool path patterns as the predetermined feature,

the display unit displays each of the machining surfaces together with the predetermined feature corresponding to the generated tool path pattern.

6. The apparatus of claim 5,

the processor is configured to calculate a certainty factor with respect to a tool path pattern generated for each of the plurality of processing surfaces of the target processing object,

the display unit emphasizes the corresponding machining surface when the certainty factor is lower than a predetermined threshold value.

Technical Field

The application relates to a method and a device for generating a tool path.

Background

Various techniques for supporting NC (Numerical Control) processing have been proposed in the past. For example, patent document 1 discloses a method and an apparatus for supporting the design of a mold using NC data. In patent document 1, when designing a mold using NC data of an existing mold, CAM data of the existing mold and CAM data of a mold to be designed are compared, and whether or not the existing mold data can be used is determined for each machining part. The ratio of the number of processing sites and the total number of processing sites, which can use the data of the existing mold, is calculated as a steal rate. In the calculation of the steal rate, a neural network is used.

Patent document 2 discloses a device for supporting generation of tool path data in an NC machine tool. In patent document 2, tool path data is automatically generated based on feature data about the three-dimensional shape of a product, raw material data, data of each machining step, data about the shape after each machining step, and data of available tools.

Disclosure of Invention

Tool paths in NC machining are sometimes generated by inputting various data into CAM software according to experience and expertise of an operator. However, when the operator is not skilled, it is difficult to generate a desired tool path particularly when the workpiece has a complicated shape.

The invention aims to provide a method and a device for generating a new tool path according to a plurality of cases.

One aspect of the present disclosure is a method for generating a tool path in NC machining, the method including: performing machine learning based on information of a plurality of known processed objects having the generated tool path; and a step of generating a new tool path for the target workpiece based on the result of the machine learning, the plurality of known workpieces and the target workpiece each having a plurality of machining surfaces, the step of performing the machine learning including: obtaining shape data for each of a plurality of known processed products; obtaining geometric information of each of a plurality of machined surfaces for each of a plurality of known machined objects; obtaining a tool path pattern selected for each of the plurality of machining surfaces from among the plurality of tool path patterns for each of the plurality of known machined products; and performing machine learning in which geometric information of the machining surface is input and a tool path pattern of the machining surface is output using the geometric information and the tool path pattern of the plurality of known machined objects, and generating a new tool path includes: obtaining shape data of the object processing object; obtaining geometric information of each of a plurality of processing surfaces of the target processing object; and generating a tool path pattern for each of the plurality of processing surfaces of the target processing object based on the result of the machine learning using the geometric information of the target processing object.

In the method according to one aspect of the present disclosure, machine learning is performed in which geometric information of a machining surface is input and a tool path pattern of the machining surface is output, based on geometric information and the tool path pattern of a plurality of known machining objects, and the tool path pattern is automatically generated for each of the plurality of machining surfaces of the target machining object, based on a result of the machine learning. Therefore, a new tool path can be generated from a plurality of instances.

It is also possible that the plurality of tool path patterns include at least a contour path, a scan line path, and a face path.

The shape data of the plurality of known processed objects and the shape data of the target processed object may be defined by an XYZ-coordinate system which is a three-dimensional orthogonal coordinate system, and the geometric information may include at least 1 of a type of the processed surface, a ratio of a maximum length in an X-axis direction to a maximum length in a Z-axis direction in each processed surface, a ratio of a maximum length in a Y-axis direction to a maximum length in a Z-axis direction in each processed surface, a ratio of a maximum length in a Z-axis direction in the whole of the plurality of processed surfaces to a maximum length in a Z-axis direction in each processed surface, a ratio of a surface area in the whole of the plurality of processed surfaces to a surface area in each processed surface, a long radius of the processed surface, a short radius of the processed surface, a Z-component of a normal vector at a center of gravity of the processed surface, a maximum curvature of the processed surface, and a.

Neural networks may also be used in machine learning.

Another aspect of the present disclosure is an apparatus for generating a tool path in NC machining, the apparatus including a processor and a display unit, the processor being configured to execute: performing machine learning based on information of a plurality of known processed objects having the generated tool path; and generating a new tool path for the target workpiece based on the result of the machine learning, the plurality of known workpieces and the target workpiece each having a plurality of machined surfaces, the performing the machine learning including: obtaining shape data for each of a plurality of known processed products; obtaining geometric information of each of a plurality of machined surfaces for each of a plurality of known machined objects; obtaining a tool path pattern selected for each of the plurality of machining surfaces from among the plurality of tool path patterns for each of the plurality of known machined products; and performing machine learning in which geometric information of the machining surface is input and a tool path pattern of the machining surface is output using the geometric information and the tool path pattern of the plurality of known machined objects, and generating a new tool path includes: obtaining shape data of the object processing object; obtaining geometric information of each of a plurality of processing surfaces of the target processing object; and generating a tool path pattern for each of a plurality of processing surfaces of the target workpiece based on a result of the machine learning using the geometric information of the target workpiece, wherein a predetermined feature that is visually recognizable is assigned to each of the plurality of tool path patterns, the processor recognizes the plurality of tool path patterns as the predetermined feature, and the display unit displays each of the processing surfaces together with the predetermined feature corresponding to the generated tool path pattern. Such an apparatus can provide the same effects as the above-described method. In such an apparatus, since each of the predetermined features of the machining surface corresponding to the generated tool path pattern is displayed on the display unit, the operator can easily recognize which tool path pattern is selected in the machining surface.

The processor may be configured to calculate the certainty factor with respect to the tool path pattern generated for each of the plurality of processing surfaces of the target processing object, or the display unit may emphasize the corresponding processing surface when the certainty factor is lower than a predetermined threshold value. In this case, the operator can change the tool path pattern of the machining surface with low certainty as necessary.

According to an aspect of the present disclosure, a method and apparatus capable of generating a new tool path from a plurality of instances can be provided.

Drawings

Fig. 1 is a diagrammatic view illustrating the method and apparatus of the present disclosure.

Fig. 2 is a schematic diagram showing several tool path patterns.

Fig. 3 is a perspective view showing an example of training data.

Fig. 4 is a perspective view showing an example of training data.

Fig. 5 is a perspective view showing an example of training data.

Fig. 6 is a perspective view showing an example of training data.

Fig. 7 is a flowchart illustrating machine learning performed by the apparatus of fig. 1.

Fig. 8 is a flowchart illustrating generation of a new tool path for the target workpiece executed by the apparatus of fig. 1.

Fig. 9 is a schematic diagram showing a network configuration.

Fig. 10 is a perspective view showing a tool path in the case of the past process design, which is a processed product in the case of the past process design and is used as a target processed product for generating a new tool path.

Fig. 11 is a perspective view illustrating a tool path generated for a target machined object by the method of the present disclosure.

(description of reference numerals)

1: a storage device; 2: a processor; 3: a display unit; 10: a device; 50: a CAD system; 60: a CAM system; 70: a work machine; 100: provided is a system.

Detailed Description

Hereinafter, a method and an apparatus for generating a tool path in NC machining according to an embodiment will be described with reference to the drawings. The same or corresponding elements are denoted by the same reference numerals, and redundant description thereof is omitted.

Fig. 1 is a diagrammatic view illustrating the method and apparatus of the present disclosure. The method of the present disclosure is implemented within a system 100 that includes a CAD (Computer Aided Design) system 50, a CAM (Computer Aided manufacturing) system 60, a device 10 for generating a tool path, and a work machine 70. The system 100 may also include other components.

In the CAD system 50, CAD data of a workpiece is created. Examples of the processed product include a mold. The workpiece represented by the CAD data has a target shape after being machined by the tool. The CAD system 50 creates CAD data of a "known workpiece" (hereinafter, may also be referred to as "training data") that becomes training data when the device 10 performs machine learning, and CAD data of a "target workpiece" that creates a new tool path from the result of the machine learning. The "known workpiece (training data)" may be a workpiece actually created in the past or a workpiece created only as electronic data and having a tool path set by a skilled operator.

The CAD data includes shape data such as a vertex, an edge, and a surface included in the workpiece. The CAD data can be defined by, for example, an XYZ coordinate system which is a three-dimensional orthogonal coordinate system. CAD data can also be defined with other coordinate systems. The workpiece includes a plurality of processing surfaces surrounded (or divided) by the characteristic line. The CAD data includes various geometric information (for example, a type of the machined surface, a ratio of a maximum length in an X-axis direction to a maximum length in a Z-axis direction in each machined surface, a ratio of a maximum length in a Y-axis direction to a maximum length in a Z-axis direction in each machined surface, a ratio of a maximum length in a Z-axis direction in the entire plurality of machined surfaces to a maximum length in a Z-axis direction in each machined surface, a ratio of a surface area in the entire plurality of machined surfaces to a surface area in each machined surface, a long radius of the machined surface, a short radius of the machined surface, a Z component of a normal vector at a center of gravity of the machined surface, a maximum curvature of the machined surface, a minimum curvature of the machined surface, and the like) with respect to each of the plurality of. The CAD data may also include other geometric information.

CAD data of a known workpiece is input to the CAM system 60. In the CAM system 60, an operator (particularly a skilled operator) P selects, for each of a plurality of machining surfaces of a known machined product, a tool path pattern to be used for the machining surface in actual machining in the past or a tool path pattern to be considered suitable for machining the machining surface, for example, from among a plurality of tool path patterns. By combining a plurality of tool path patterns selected for a plurality of machining surfaces, tool paths for 1 machined object are generated.

Fig. 2 is a schematic diagram showing several tool path patterns. The plurality of tool path modes can include various tool path modes. For example, the left side of fig. 2 shows the contour path. In the contour path, the tool T machines a machining surface in a contour motion. The central diagram of fig. 2 shows the scan line path. In the scanning line path, the tool T machines the machined surface so as to bury the region while following the machined surface. The right-hand graph of fig. 2 represents the creeping path. In the creeping path, the tool T machines the machining surface in an action along the boundary line of the machining surface. The plurality of tool path patterns may include a tool path pattern other than the 3 tool path patterns (for example, a spiral machining in which a tool moves spirally in a height direction while machining a machined surface).

In order to visually recognize which tool path pattern is selected for the machining surface, a predetermined color is assigned to each of the plurality of tool path patterns. As shown in fig. 2, for example, in the present embodiment, "red" is assigned to the machining surface selected as the contour path, "green" is assigned to the machining surface selected as the scanning path, and "blue" is assigned to the machining surface selected as the planar path. The display unit of the CAM system 60 displays each of the machined surfaces in a predetermined color corresponding to the selected tool path pattern. This allows the operator P to easily recognize which tool path pattern is selected on the machining surface. In addition, in order to visually recognize which tool path pattern is selected for the machining surface, a predetermined visually recognizable feature (for example, a color, a pattern, and/or a character) can be assigned to each of the plurality of tool path patterns.

Fig. 3 to 6 are perspective views showing examples of training data. The training data includes shape data produced by the CAD system 50, geometric information calculated by the CAD system 50, and tool path patterns selected by the CAM system 60. The tool path pattern is included in the training data as a color assigned to the tool path pattern.

Referring again to fig. 1, training data is input to the device 10. As described above, the training data input to the apparatus 10 already has a tool path generated.

The device 10 includes a storage device 1, a processor 2, and a display unit 3, and these components are connected to each other via a bus (not shown) or the like. The device 10 may include other components such as a ROM (Read Only Memory), a RAM (Random access Memory), and/or an input device (e.g., a mouse, a keyboard, and/or a touch panel). The apparatus 10 may be, for example, a computer, server, tablet, or the like.

The storage device 1 may be one or more hard disk drives or the like. The storage device 1 stores inputted training data.

The processor 2 may be, for example, a CPU (Central Processing Unit) or the like. The processor 2 may be constituted by 1 CPU or a plurality of CPUs, for example. The processor 2 is configured to execute a plurality of processes described below, and a program for executing each process can be stored in the storage device 1, for example. The processor 2 is configured to perform machine learning (details will be described later) based on information of a plurality of pieces of training data stored in the storage device 1. In machine learning, for example, a neural network can be used.

The processor 2 is configured to generate a new tool path for the target workpiece based on the result of the machine learning, using CAD data of the target workpiece created by the CAD system 50 (details will be described later). The processor 2 is configured to calculate the certainty factor with respect to the tool path pattern generated for each of the plurality of processing surfaces of the target processing object.

The display section 3 may be a liquid crystal display, a touch panel, or the like. The display unit 3 displays each machined surface together with predetermined visually recognizable features (for example, colors, patterns, and/or characters) corresponding to the generated tool path pattern, as in the display unit of the CAM system 60.

The CAD system 50, the CAM system 60, and the device 10 may be configured as independent devices, or may be embedded in the same device (for example, CAD software and CAM software may be embedded in the device 10).

The new tool path generated by the apparatus 10 can be converted into NC data and input to the machine tool 70.

Next, the operation performed by the apparatus 10 will be described.

First, machine learning performed by the apparatus 10 is explained. Fig. 7 is a flowchart illustrating machine learning performed by the apparatus of fig. 1.

The processor 2 acquires information from the storage device 1 regarding each of the plurality of training data (step S100). The acquired information includes shape data of each piece of training data, geometric information of each of the plurality of processing surfaces, and a tool path (color) selected for each of the plurality of processing surfaces. Next, the processor 2 performs machine learning in which geometric information of a machined surface is input and a tool path pattern (color) of the machined surface is output, using geometric information and the tool path pattern (color) of the plurality of training data (step S102). With the above, a series of operations ends. The above steps may be repeated until a desired convergence result is obtained.

Next, generation of a new tool path for the target workpiece performed by the apparatus 10 will be described. Fig. 8 is a flowchart illustrating generation of a new tool path for the target workpiece executed by the apparatus of fig. 1.

The processor 2 acquires CAD data of the target workpiece created by the CAD system 50 (step S200). The obtained CAD data includes shape data of the target workpiece and geometrical information of each of the plurality of machined surfaces.

Next, the processor 2 generates a tool path pattern for each of the plurality of processing surfaces of the target workpiece from the result of the machine learning by using the geometric information of the target workpiece (step S202).

Specifically, in step S202, the processor 2 calculates, for each of the plurality of tool path patterns, a degree of probability (or may also be referred to as a certainty) with which a certain machining surface is machined by the tool path pattern. Specifically, the processor 2 calculates, for each of the plurality of colors, how much certainty is given to selecting the color for a certain processing surface. For example, in the present embodiment, the processor 2 calculates the certainty factor of selecting red, the certainty factor of selecting green, and the certainty factor of selecting blue for a certain machining surface, and then selects the color having the highest certainty factor (i.e., the tool path pattern) for the machining surface. The processor 2 combines the selected plurality of tool path patterns to generate a tool path of the target workpiece, and transmits the generated tool path to the display unit 3.

Next, the display unit 3 displays the tool path of the generated target workpiece (step S204). Specifically, the display unit 3 displays each of the plurality of processing surfaces of the target processing object in the selected color. At this time, the display unit 3 displays the selected color (tool path mode) according to the certainty factor. Specifically, when the certainty factor of the tool path pattern selected for a certain machining surface is lower than a predetermined threshold value, the display unit 3 emphasizes the machining surface (for example, displays the machining surface in a light color or a dark color). For example, when red is selected for a certain machined surface but the reliability thereof is lower than a predetermined threshold value, the machined surface is displayed in light red. This makes it easy for the operator P to recognize that the machined surface has low certainty.

Next, the processor 2 receives an input from the operator P as to whether or not a change is necessary (step S206). Specifically, when the operator P determines that the tool path pattern (color) of a certain machining surface (for example, a machining surface with low certainty degree displayed in step S204) is not suitable, the operator P can change the tool path pattern (color) of the corresponding machining surface by inputting a change command via the input device.

When an input indicating that a change is necessary is made in step S206, the processor 2 changes the tool path pattern (color) of the corresponding processing surface in accordance with the change command input from the operator P (step S208), and returns to step S204 again.

If there is an input indicating that no change is required in step S206, the processor 2 stores the generated tool path of the target workpiece in the storage device 1 (step S210), and ends the series of operations. The new tool path stored in the storage device 1 may be used as one of the training data in the machine learning to be performed next time. The generated new tool path may be actually used for machining in the machine tool 70 after being converted into NC data, or the operator P may change the tool path generated by the apparatus 10 based on the result of the actual machining. The tool path changed according to the result of the machining may be stored in the storage device 1 and used as one of the training data in the machine learning to be performed next time.

As described above, the method and apparatus 10 according to the present embodiment perform machine learning in which the geometric information of the machining surface is input and the tool path pattern of the machining surface is output, based on the geometric information and the tool path pattern of a plurality of known machining objects, and automatically generate the tool path pattern for each of the plurality of machining surfaces of the target machining object based on the result of the machine learning. Therefore, a new tool path can be generated from a plurality of instances.

In the device 10 according to the present embodiment, since each machined surface is displayed on the display unit 3 together with a predetermined color corresponding to the generated tool path pattern, the operator P can easily recognize which tool path pattern has been selected for the machined surface.

In the device 10 according to the present embodiment, the processor 2 is configured to calculate the certainty factor with respect to the tool path pattern generated for each of the plurality of machining surfaces of the target machining object, and the display unit 3 is configured to emphasize the corresponding machining surface when the certainty factor is lower than a predetermined threshold value. Therefore, the operator can change the tool path pattern of the machining surface with low certainty as necessary.

Although the embodiments of the method and apparatus for generating a tool path in NC machining have been described, the present invention is not limited to the above embodiments. Those skilled in the art will appreciate that various modifications of the above-described embodiments can be made. Further, those skilled in the art will appreciate that the above methods need not be performed in the above order, and may be performed in other orders as long as there is no contradiction.

For example, in the above embodiment, a neural network is used in machine learning. However, in other embodiments, other methods (e.g., decision trees, etc.) may be used in machine learning.