Mold creation support system, mold creation support method, and learning data set generation method

Abstract

A mold manufacturing support system is provided with: an acquisition unit that acquires mold actual shape data indicating the actual shape of a mold, and molded article difference data indicating the difference between the actual shape of a molded article molded by the mold and a target shape; and an estimation unit that uses the learning mold shape data and the learning molded product difference data as input data and the learning mold target shape data as teaching data to learn a previously generated learning model by a machine. On the basis of the actual mold shape data and the molded product difference data, mold target shape data indicating the target shape of the mold is estimated.

Description

Technical Field

[0001] This invention relates to a mold making support system, a mold making support method, and a method for generating a learning dataset. Background Technology

[0002] Patent document 1 discloses a technique for obtaining estimated shape data of a mold through so-called CAE (Computer Aided Engineering).

[0003] Prior art literature

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2012-119010 Summary of the Invention

[0006] The problem that the invention aims to solve

[0007] In the aforementioned conventional techniques, the accuracy of prediction was insufficient, which sometimes led to an increase in the number of mold shape corrections. In particular, in recent years, in addition to problems such as cracks and wrinkles caused by the increasing strength of materials, it has become difficult to meet the dimensional accuracy of molded products due to springback after molding, resulting in a trend of increasingly more mold shape corrections.

[0008] The present invention was made in view of the above-mentioned problems, and its main objective is to provide a mold manufacturing support system, a mold manufacturing support method, and a method for generating a learning dataset that can reduce the number of mold shape corrections.

[0009] Solution for solving the problem

[0010] To address the aforementioned issues, one aspect of the present invention relates to a mold manufacturing support system, comprising: an acquisition unit that acquires actual mold shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and a target shape of a molded article formed by the mold; and an estimation unit that uses a pre-generated, learned model obtained through machine learning, taking the learned mold shape data and the learned molded article difference data as input data and the learned mold target shape data as teaching data, to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded article difference data. Accordingly, it is possible to reduce the number of mold shape corrections required.

[0011] In the above scheme, the mold making support system may also include a transformation unit that transforms the original data representing the actual shape of the mold into actual shape data of the mold for input to the learned model. Accordingly, the actual shape data of the mold can be obtained from the original data.

[0012] In the above scheme, the mold making support system may also include a CAM unit that transforms the target shape data of the mold into NC data, and a machine tool that corrects the actual shape of the mold based on the NC data. Accordingly, the actual shape of the mold can be corrected based on the target shape data of the mold estimated by learning the model.

[0013] In the above-described scheme, the mold-making support system may also include a shape measuring device that measures the actual shape of the molded article formed by the mold after correction by the machine tool. Accordingly, the actual shape of the molded article formed by the corrected mold can be measured.

[0014] In the above-described scheme, the mold-making support system may also include a subtraction calculation unit that calculates corrected difference data representing the difference between the actual shape of the molded article as measured by the shape measuring instrument and the target shape. Based on this, the difference between the actual shape and the target shape of the molded article formed by the corrected mold can be calculated.

[0015] In the above scheme, the estimation unit may use the target shape data of the mold as the new actual shape data of the mold and the corrected difference data as the new difference data of the molded product to estimate new target shape data of the mold. Accordingly, the estimation of the target shape data of the mold can be performed repeatedly.

[0016] In the above-described scheme, the mold-making support system may also include: a display unit that displays the target shape data of the mold estimated from the learned model; and a receiving unit that receives the user's decision on the target shape of the mold. Accordingly, the user can observe the displayed target shape data to determine the target shape of the mold.

[0017] In the above scheme, the mold making support system may also include a learning unit that uses the target shape data of the mold estimated by the learned model and the actual shape data of the molded article representing the actual shape of the molded article measured by the shape measuring device to relearn the learned model. This further improves the estimation accuracy of the learned model.

[0018] Furthermore, another aspect of the present invention relates to a mold-making support method, wherein the mold-making support method performs the following processing: obtaining actual mold shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and the target shape of the molded article formed by the mold; and using a pre-generated learning model generated through machine learning, with learning mold shape data and learning molded article difference data as input data and learning mold target shape data as teaching data, to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded article difference data. Accordingly, it is possible to reduce the number of mold shape corrections.

[0019] Furthermore, another aspect of the present invention relates to a method for generating a learning dataset, wherein the method comprises the following steps: preparing multiple mold shape data and multiple molded product shape data corresponding to the multiple mold shape data; extracting first mold shape data and second mold shape data from the multiple mold shape data; extracting first molded product shape data and second molded product shape data corresponding to the first mold shape data and second mold shape data from the multiple molded product shape data; and setting the first mold shape data as learning mold shape data, setting the difference between the first molded product shape data and the second molded product shape data as learning molded product difference data, and setting the second mold shape data as learning mold target shape data. Accordingly, by setting the learning mold shape data and learning molded product difference data as input data and the learning mold target shape data as teaching data, a learned model can be generated through machine learning, enabling the estimation of the mold target shape without complex user GUI operations.

[0020] Furthermore, another aspect of the present invention relates to a mold-making support system, wherein the mold-making support system comprises: an acquisition unit that acquires mold shape data representing the shape of a mold; and an estimation unit that uses a pre-generated, learned model obtained through machine learning, taking the mold shape data for learning as input data and the molded article shape data for learning as teaching data, to estimate molded article predicted shape data representing the predicted shape of the molded article based on the mold shape data. Accordingly, it is possible to reduce the number of mold shape corrections.

[0021] In the above-described solution, the mold-making support system may also include a display unit that shows the mold shape data and the predicted shape data of the molded product. Accordingly, the user can visually compare the mold shape data with the predicted shape data of the molded product.

[0022] In the above scheme, the display unit may also display target shape data of the molded article, representing the target shape of the molded article. Accordingly, the user can visually compare the predicted shape data of the molded article with the target shape data of the molded article.

[0023] In the above-described scheme, the mold-making support system may also include a receiving unit that accepts corrections to the mold shape data made by the user, an estimation unit that estimates new predicted shape data for the molded product based on the corrected mold shape data, and a display unit that displays the new predicted shape data for the molded product. Accordingly, new molded product shape data can be estimated and displayed based on the mold shape data corrected by the user.

[0024] Furthermore, another aspect of the present invention relates to a mold-making support method, wherein the mold-making support method performs the following processing: obtaining mold shape data representing the shape of the mold; and using a pre-generated, learned model generated through machine learning with the mold shape data as input data and the molded article shape data as teaching data, estimating molded article predicted shape data representing the predicted shape of the molded article based on the mold shape data. Accordingly, it is possible to reduce the number of mold shape corrections. Attached Figure Description

[0025] Figure 1 This is a diagram illustrating an example of a mold-making support system according to the first embodiment.

[0026] Figure 2 It is a diagram used to illustrate the estimation of the target shape data of the mold.

[0027] Figure 3 This diagram shows an example of a supporting device and a GUI terminal.

[0028] Figure 4 This is a diagram illustrating an example of a mold-making support method.

[0029] Figure 5 This is a diagram used to illustrate the creation of a learning dataset.

[0030] Figure 6 This is a diagram used to illustrate the creation of a learning dataset.

[0031] Figure 7 This is a diagram illustrating an example of a mold-making support system according to the second embodiment.

[0032] Figure 8 This is a diagram illustrating an example of a mold-making support system according to the third embodiment.

[0033] Figure 9This is a diagram illustrating an example of a mold-making support system according to the fourth embodiment.

[0034] Figure 10 It is a diagram used to illustrate the estimation of the shape data of a molded product.

[0035] Figure 11A This is a diagram showing an example of a GUI terminal display.

[0036] Figure 11B This is a diagram showing an example of a GUI terminal display.

[0037] Figure 11C This is a diagram showing an example of a GUI terminal display.

[0038] Figure 11D This is a diagram showing an example of a GUI terminal display.

[0039] Figure 12 This is a diagram illustrating an example of a mold-making support method. Detailed Implementation

[0040] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that in this specification and the various drawings, sometimes the same reference numerals are used to denote the same elements as described above in the figures shown, and detailed descriptions are appropriately omitted.

[0041] The functions of the elements disclosed in this specification can be executed using circuitry or processing circuitry, including general-purpose processors, special-purpose processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. Processors include transistors and other circuitry, and are therefore considered as processing circuitry or circuitry. In this disclosure, a circuit, unit, or mechanism is hardware that performs the exemplified functions, or it may be hardware programmed to perform the exemplified functions. The hardware may also be the hardware disclosed in this specification, or it may be other known hardware programmed or configured to perform the exemplified functions. Where the hardware is a processor considered a type of circuit, the circuit, mechanism, or unit is a combination of hardware and software, with the software used in the structure of the hardware and / or processor.

[0042] [First Implementation]

[0043] Figure 1 This is a diagram illustrating a structural example of the mold making support system 100A according to the first embodiment. The mold making support system 100A is a system for supporting the making of molds used in stamping.

[0044] The mold making support system 100A includes support device 1, GUI (Graphical User Interface) terminal 2, CAM / machine tool 3, stamping machine 4, shape measuring instrument 5, and forming analysis DB (database) 6.

[0045] Support device 1 is a computer including a CPU, RAM, ROM, non-volatile memory, and input / output interfaces. GUI terminal 2 is a terminal that provides a GUI to user U, and like support device 1, it is equipped with a computer.

[0046] The CPU executes information processing according to programs loaded from ROM or non-volatile memory into RAM. Programs can be provided via information storage media or via communication networks.

[0047] In this embodiment, the support device 1 and the GUI terminal 2 are in a server-client relationship. However, this is not a limitation; the support device 1 and the GUI terminal 2 can also be integrated. That is, the functions of the support device 1 and the GUI terminal 2 can be implemented by a single device.

[0048] CAM / Machine Tool 3 includes a CAM unit that generates NC data and a machine tool that makes molds based on the NC data.

[0049] The stamping machine 4 produces shaped products by using a die for stamping.

[0050] The shape measuring device 5 measures the shape of the molded article being produced. The shape measuring device 5 is, for example, a 3D scanner, or a contact or non-contact displacement sensor.

[0051] The DB6 forming analysis database stores data on molds and molded products manufactured in the past. This data includes not only actual shape data obtained through testing and measurement, but also data obtained through CAE (Computer-Aided Engineering).

[0052] The following is an explanation. Figure 1 The processes shown are (a) to (j).

[0053] (a) User U inputs raw data representing the actual shape of the mold, raw data representing the actual shape of the molded product formed by the mold, and raw data representing the target shape of the molded product into GUI terminal 2. The raw data may be, for example, CAD data or STL data.

[0054] (b) The GUI terminal 2 generates data for input to the learned model based on the raw data and outputs it to the support device 1. Specifically, the GUI terminal 2 transforms the raw data representing the actual shape of the mold into actual mold shape data for input to the learned model. In addition, the GUI terminal 2 calculates molded article difference data representing the difference between the actual shape and the target shape of the molded article.

[0055] (c) The support device 1 uses the learned model to estimate the target shape data of the mold, representing the target shape of the mold, based on the actual shape data of the mold and the difference data of the molded product obtained from the GUI terminal 2 (see reference). Figure 2 ), and output it to GUI terminal 2.

[0056] The learned model is a pre-generated model that uses machine learning to process data from the mold shape data and the difference data of the formed product as input data, and the target shape data of the mold as teaching data. The generation of the learning data will be discussed later. The learned model can be, for example, a neural network or a Gaussian process regression model.

[0057] (d) The GUI terminal 2 displays the mold target shape data output from the support device 1. The user U confirms the mold target shape data displayed on the GUI terminal 2 and determines whether to adopt it. (e) When the GUI terminal 2 receives the mold target shape decision made by the user, (f) it outputs the mold target shape data.

[0058] (g) User U inputs the target shape data of the mold to CAM / Machine Tool 3. (h) CAM / Machine Tool 3 transforms the target shape data of the mold into NC data and corrects the actual shape of the mold based on the transformed NC data. (i) The stamping machine 4 forms the molded product using the corrected mold.

[0059] (j) The shape measuring device 5 measures the actual shape of the molded article formed by the corrected mold. Then, the target shape data of the mold is set as the new actual shape data of the mold, and the corrected difference data, which represents the difference between the measured actual shape of the molded article and the target shape, is set as the new difference data of the molded article. Steps (a) to (j) are then repeated. It should be noted that the data obtained by measuring the actual mold can also be used for the actual shape data of the mold.

[0060] Repeat steps (a) to (j) as described above until there is no longer a difference between the actual shape and the target shape of the molded product.

[0061] Figure 3This is a block diagram illustrating an example of the structure of the support device 1 and the GUI terminal 2. The support device 1 includes an acquisition unit 11, an estimation unit 12, and a learning unit 13. These functional units are implemented by the CPU of the support device 1 performing information processing according to a program loaded from ROM or non-volatile memory into RAM.

[0062] The GUI terminal 2 includes a transformation unit 21 and a subtraction calculation unit 22. These functional units are implemented by the CPU of the GUI terminal 2 executing information processing according to a program loaded from ROM or non-volatile memory into RAM.

[0063] In addition, the GUI terminal 2 includes a receiving unit 23 and a display unit 24. The receiving unit 23 is, for example, a keyboard or mouse, and accepts operations from the user U. The display unit 24 is, for example, a liquid crystal display.

[0064] Figure 4 This is a flowchart illustrating an example of the steps of a mold making support method implemented in a mold making support system 100A. The support device 1 and the GUI terminal 2 each execute the information processing shown in the diagram according to their respective programs.

[0065] First, the GUI terminal 2 obtains raw data representing the actual shape of the mold, raw data representing the actual shape of the molded article formed by the mold, and raw data representing the target shape of the molded article (S21, equivalent to step (a) above).

[0066] Next, the GUI terminal 2 transforms the raw data representing the actual shape of the mold into actual shape data of the mold for input into the learned model (S22, as a process of the transformation unit 21, equivalent to the step (b) above). The actual shape data of the mold is represented by point cloud data such as CAD data or STL data. When the number of measurement points in the STL data is large, dimensionality reduction based on feature extraction can also be performed.

[0067] Next, the GUI terminal 2 calculates the molded product difference data, which represents the difference between the actual shape and the target shape of the molded product (S23, as a process of the subtraction calculation unit 22, equivalent to the process in (b) above).

[0068] The support device 1 obtains the actual shape data of the mold and the difference data of the molded product from the GUI terminal 2 (S11, as processing of the acquisition unit 11).

[0069] Next, the support device 1 uses the learned model to estimate the target shape data of the mold based on the actual shape data of the mold and the difference data of the molded product (S12, as the processing of the estimation unit 12, is equivalent to the process in (c) above).

[0070] The GUI terminal 2 displays the target shape of the mold, which is estimated by learning the model in the support device 1, on the display unit 24 (S24, equivalent to the process in (d) above).

[0071] Next, when the receiving unit 23 receives the decision of the mold target shape made by the user (S25: Yes), the GUI terminal 2 outputs the mold target shape data (S26, equivalent to the process in (e) and (f) above).

[0072] The target shape data of the mold is used to correct the actual shape of the mold by the CAM / machine tool 3 as described above. The shape of the molded article formed by the corrected mold is measured by the shape measuring device 5 (equivalent to the processes (g) to (j) above).

[0073] The GUI terminal 2 calculates corrected difference data representing the difference between the actual shape of the molded product measured by the shape measuring device 5 and the target shape (as processed by the subtraction calculation unit 22).

[0074] The support device 1 sets the mold target shape data to the new actual mold shape data from the GUI terminal 2 and sets the corrected difference data to the new molded product difference data to estimate the new mold target shape data (as a process of the estimation unit 12).

[0075] According to this embodiment, instead of relying on the experience, know-how, intuition, etc. of skilled personnel, the appropriate target shape of the mold can be quickly obtained by learning the model, thus reducing the number of mold shape corrections.

[0076] Figure 5 and Figure 6 This diagram illustrates the creation of the learning dataset. The creation of the learning dataset and the creation of the learned model using the learning dataset are performed by the learning unit 13 of the support device 1.

[0077] like Figure 5 As shown, the forming analysis DB6 prepares multiple mold shape data and multiple corresponding molded product shape data. That is, it prepares multiple sets of mold shape data representing the shape of the mold and molded product shape data representing the shape of the molded product formed by the mold. Conditions A to E refer to the shape of the mold.

[0078] like Figure 6As shown, the first mold shape data and the second mold shape data, as well as the corresponding first molded product shape data and the second molded product shape data, are extracted from the forming analysis DB6. The first mold shape data is set as the learning mold shape data, the difference between the first molded product shape data and the second molded product shape data is set as the learning molded product difference data, and the second mold shape data is set as the learning mold target shape data to create a learning dataset.

[0079] For example, after extracting the mold shape data and molded product shape data of condition A, as well as the mold shape data and molded product shape data of condition B, the mold shape data of condition A is set as the learning mold shape data, the difference between the molded product shape data of condition A and the molded product shape data of condition B is set as the learning molded product difference data, and the mold shape data of condition B is set as the learning mold target shape data to create a learning dataset based on conditions A and B.

[0080] Similarly, it is possible to create learning datasets based on conditions AC, AD, and AE, and further, to create learning datasets based on conditions BA, BC, ..., CA, CB, ..., DA, DB, ..., EA, EB, ... By changing the combination of the two extracted conditions in this way, multiple learning datasets can be created.

[0081] [Second Implementation]

[0082] Figure 7 This is a diagram illustrating an example of the mold-making support system 100B according to the second embodiment. In this embodiment, (e) when the GUI terminal 2 receives a decision on the target shape of the mold, (k) the target shape data of the mold is directly output from the GUI terminal 2 to the CAM / machine tool 3. This enables automation of mold correction, reducing the burden on the user U.

[0083] [Third Implementation]

[0084] Figure 8 This is a diagram illustrating an example of the mold making support system 100C according to the third embodiment. In this embodiment, (j) when the shape measuring device 5 measures the actual shape of the molded article, (m) the mold shape data and the molded article shape data are registered with the molding analysis DB6, and the learning unit 13 of the support device 1 performs relearning of the learned model. As a result, the estimation accuracy can be improved.

[0085] It should be noted that this embodiment is an embodiment obtained by adding (m) to the first embodiment described above, but it is not limited to this, and the second embodiment described above may also be supplemented with (m) steps.

[0086] [Fourth Implementation]

[0087] Figure 9 This is a diagram illustrating an example of the mold making support system 100D according to the fourth embodiment. In this embodiment, a learned model (see [reference]) is used, which outputs predicted shape data of the molded article representing the predicted shape of the molded article when mold shape data representing the shape of the mold is input. Figure 10 This allows for the determination of mold shape through dialogue (also known as the "surrogate model").

[0088] The following is an explanation. Figure 9 The process shown is (n) ~ (w).

[0089] (n) User U inputs raw data representing the shape of the mold and raw data representing the target shape of the molded product into GUI terminal 2. The raw data may be, for example, CAD data or STL data.

[0090] (o) The GUI terminal 2 generates data for input to the learned model based on the raw data and outputs it to the support device 1. Specifically, the GUI terminal 2 transforms the raw data representing the shape of the mold into mold shape data for input to the learned model.

[0091] (p) The support device 1 uses the learned model to estimate the predicted shape data of the molded article, representing the predicted shape of the molded article, based on the mold shape data obtained from the GUI terminal 2 (see reference). Figure 10 ), and output it to GUI terminal 2.

[0092] (q) The GUI terminal 2 displays the predicted shape data of the molded article output from the support device 1. Here, the mold shape data and the predicted shape data of the molded article are displayed in an alternating manner. In addition, the predicted shape data of the molded article is displayed overlapping with the target shape data of the molded article, which represents the target shape of the molded article.

[0093] (r) While confirming the mold shape data, predicted molded product shape data, and target molded product shape data displayed on the GUI terminal 2, user U inputs corrections to the mold shape data. The GUI terminal 2 accepts the corrections to the mold shape data made by the user.

[0094] (s) The GUI terminal 2 outputs the corrected mold shape data to the support device 1. (t) The support device 1 uses the learned model to estimate new molded product prediction shape data based on the corrected mold shape data and outputs it to the GUI terminal 2.

[0095] (u) The GUI terminal 2 displays the new predicted shape data of the molded article output from the support device 1. Here, the GUI terminal 2 also displays the mold shape data, the predicted shape data of the molded article, and the target shape data of the molded article in the same way as in step (q) above.

[0096] User U repeatedly corrects the mold shape data and accordingly repeats the above (r) to (u) steps until there is no longer a difference between the predicted shape of the molded product and the target shape of the molded product, or until the error is small enough to be acceptable.

[0097] (v) When there is no longer a difference between the predicted shape and the target shape of the molded product, user U inputs the mold shape data. (w) When GUI terminal 2 receives the mold shape data decision made by the user, it outputs the mold shape data.

[0098] The output mold shape data is used for mold making by CAM / machine tool 3, and the mold made is used for stamping by stamping machine 4 (see reference). Figure 1 wait).

[0099] Figures 11A-11D This is a diagram showing an example of the display of GUI terminal 2.

[0100] like Figure 11A As shown, the GUI terminal 2 displays a mold shape object DS based on mold shape data and a molded product predicted shape object ES based on molded product predicted shape data (corresponding to the process in (q) above). Furthermore, the molded product predicted shape object ES and the molded product target shape object GS based on molded product target shape data are displayed overlappingly.

[0101] By arranging and displaying the mold shape object DS and the molded product predicted shape object ES, one can visually grasp the shape of the molded product predicted by the mold. In addition, by overlaying and displaying the molded product predicted shape object ES and the molded product target shape object GS, one can visually grasp the difference between the predicted shape and the target shape of the molded product.

[0102] like Figure 11B As shown, while observing the screen MG, the user U modifies the shape of the mold shape object DS (equivalent to the process described in (r) above). The shape of the mold shape object DS is modified, for example, by dragging the outline using the pointer PT. By modifying the shape of the mold shape object DS, the mold shape data that forms the basis of the mold shape object DS is modified.

[0103] like Figure 11CAs shown, when the shape of the mold shape object DS is modified, the shape of the molded article prediction shape object ES changes accordingly (equivalent to the steps (t) and (u) above). That is, when the mold shape data is modified, new molded article prediction shape data is deduced from the modified mold shape data, thereby changing the shape of the molded article prediction shape object ES.

[0104] like Figure 11D As shown, while observing the screen MG, the user U repeatedly corrects the shape of the mold shape object DS until there is no longer a difference between the predicted shape object ES and the target shape object GS of the molded product, or until the error is small enough to be acceptable. Finally, when the shape of the mold shape object DS is determined, the mold shape data corresponding to the mold shape object DS is output.

[0105] Figure 12 This is a flowchart illustrating an example of the steps of a mold making support method implemented in a mold making support system 100D. The support device 1 and the GUI terminal 2 each execute the information processing shown in the diagram according to their respective programs.

[0106] First, the GUI terminal 2 obtains the original data representing the shape of the mold (S41, equivalent to the process in (n) above), and transforms it into mold shape data for input to the learned model (S42, as a process of the transformation unit 21, equivalent to the process in (o) above).

[0107] The support device 1 obtains mold shape data from the GUI terminal 2 (S31, as a process of the acquisition unit 11), and uses the learned model to estimate the predicted shape data of the molded product based on the mold shape data (S32, as a process of the estimation unit 12, equivalent to the process in (p) above).

[0108] The GUI terminal 2 displays the mold shape data, the predicted shape data of the molded product estimated by the model learned in the support device 1, and the target shape data of the molded product on the display unit 24 (S43, equivalent to the process in (q) above, referring to...). Figure 11A ).

[0109] When the GUI terminal 2 receives a correction of the mold shape data made by the user in the receiving unit 23 (S44: Yes), it outputs the corrected mold shape data to the support device 1 (equivalent to the processes (r) and (s) above, see refer to...). Figure 11B ).

[0110] When the support device 1 obtains the corrected mold shape data from the GUI terminal 2 (S31), it uses the learned model to estimate the new molded product prediction shape data based on the corrected mold shape data (S32, equivalent to the process in (t) above).

[0111] The GUI terminal 2 replaces the previously displayed target shape data of the molded article with the new predicted shape data of the molded article derived from the learned model in the support device 1 (S43, equivalent to the process described in (u) above, refer to...). Figure 11C ).

[0112] Repeat the above process until the mold shape data correction performed by user U is completed.

[0113] Afterwards, when the GUI terminal 2 receives the mold shape decision made by the user in the receiving unit 23 (S45: Yes), it outputs the mold shape data (S46, equivalent to the process (v) and (w) above).

[0114] According to this embodiment, instead of relying on the experience, know-how, intuition, etc. of skilled personnel, the mold shape used to achieve the target shape of the molded product can be determined through dialogue, thus reducing the number of mold shape corrections.

[0115] The above describes the embodiments of the present invention, but the present invention is not limited to the embodiments described above, and various modifications can be made by those skilled in the art.

[0116] This disclosure may include the following solutions.

[0117] (Option 1)

[0118] A mold making support system, wherein,

[0119] The mold making support system has the following features:

[0120] The acquisition unit acquires mold actual shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and the target shape of the molded article formed by the mold; and

[0121] The estimation unit uses a pre-generated learning model generated through machine learning, which takes learning mold shape data and learning molded product difference data as input data and learning mold target shape data as teaching data, to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded product difference data.

[0122] (Option 2)

[0123] According to the mold manufacturing support system described in Scheme 1, wherein,

[0124] The mold making support system also includes a transformation unit that transforms the original data representing the actual shape of the mold into actual shape data of the mold for input to the learned model.

[0125] (Option 3)

[0126] According to the mold making support system described in Scheme 1 or 2, wherein,

[0127] The mold making support system also includes a CAM unit that transforms the target shape data of the mold into NC data, and a machine tool that corrects the actual shape of the mold based on the NC data.

[0128] (Option 4)

[0129] According to the mold manufacturing support system described in Scheme 3, wherein...

[0130] The mold making support system also includes a shape measuring device that measures the actual shape of the molded product formed by the mold after being corrected by the machine tool.

[0131] (Option 5)

[0132] According to the mold manufacturing support system described in Scheme 4, wherein...

[0133] The mold making support system also includes a subtraction calculation unit that calculates corrected difference data representing the difference between the actual shape of the molded article as measured by the shape measuring instrument and the target shape.

[0134] (Option 6)

[0135] According to the mold manufacturing support system described in Scheme 5, wherein...

[0136] The estimation unit uses the target shape data of the mold as the new actual shape data of the mold and the corrected difference data as the new difference data of the molded product to estimate the new target shape data of the mold.

[0137] (Option 7)

[0138] According to any one of Schemes 1 to 6, a mold manufacturing support system is provided, wherein,

[0139] The mold making support system further includes: a display unit that displays the target shape data of the mold estimated by the learned model; and a receiving unit that accepts the user's decision on the target shape of the mold.

[0140] (Option 8)

[0141] According to the mold manufacturing support system described in Scheme 5, wherein...

[0142] The mold making support system also includes a learning unit that uses the target shape data of the mold derived from the learned model and the actual shape data of the molded article representing the actual shape of the molded article as measured by the shape measuring device to relearn the learned model.

[0143] (Option 9)

[0144] A mold making support method, wherein,

[0145] The mold manufacturing support method is processed as follows:

[0146] Obtain mold actual shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and the target shape of the molded article formed by the mold; and

[0147] Using pre-generated, learned models generated through machine learning, with learning mold shape data and learning molded product difference data as input data and learning mold target shape data as teaching data, are used to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded product difference data.

[0148] (Option 10)

[0149] A method for generating a learning dataset, wherein,

[0150] The method for generating the learning dataset involves the following processing:

[0151] Prepare multiple mold shape data and multiple molded product shape data corresponding to the multiple mold shape data;

[0152] Extract the first mold shape data and the second mold shape data from the plurality of mold shape data;

[0153] Extract first molded product shape data and second molded product shape data corresponding to the first mold shape data and the second mold shape data from the plurality of molded product shape data; and

[0154] Set the first mold shape data as the learning mold shape data, set the difference between the first molded product shape data and the second molded product shape data as the learning molded product difference data, and set the second mold shape data as the learning mold target shape data.

[0155] (Option 11)

[0156] A mold making support system, wherein,

[0157] The mold making support system has the following features:

[0158] The acquisition unit acquires mold shape data representing the shape of the mold; and

[0159] The estimation unit uses a pre-generated, learned model obtained through machine learning, which takes the mold shape data as input data and the molded article shape data as teaching data, to estimate the predicted shape data of the molded article based on the mold shape data.

[0160] (Option 12)

[0161] According to the mold manufacturing support system described in Scheme 11, wherein,

[0162] The mold making support system also includes a display unit that shows the mold shape data and the predicted shape data of the molded product.

[0163] (Option 13)

[0164] According to the mold manufacturing support system described in Scheme 12, wherein,

[0165] The display unit also displays molded article target shape data representing the target shape of the molded article.

[0166] (Option 14)

[0167] According to the mold making support system described in scheme 12 or 13, wherein,

[0168] The mold-making support system also includes a receiving unit that accepts corrections to the mold shape data made by the user.

[0169] The estimation unit estimates new predicted shape data for the molded product based on the corrected mold shape data.

[0170] The display unit shows the new predicted shape data of the molded article.

[0171] (Option 15)

[0172] A mold making support method, wherein,

[0173] The mold manufacturing support method is processed as follows:

[0174] Obtain mold shape data representing the shape of the mold; and

[0175] Using a pre-generated, learned model generated through machine learning, with learning mold shape data as input data and learning molded product shape data as teaching data, the predicted shape data of the molded product is estimated based on the mold shape data.

[0176] This application claims priority to Japanese Patent Application No. 2023-054608, filed on March 30, 2023, which is incorporated herein by reference.

[0177] Explanation of reference numerals in the attached figures

[0178] 1 Support device, 2 GUI terminal, 3 CAM / machine tool, 4 stamping machine, 5 shape measuring instrument, 6 forming analysis DB, 11 acquisition unit, 12 estimation unit, 13 learning unit, 21 transformation unit, 22 subtraction calculation unit, 23 receiving unit, 24 display unit, 100 mold making support system.

Claims

1. A mold making support system, wherein, The mold making support system has the following features: The acquisition unit acquires mold actual shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and the target shape of the molded article formed by the mold; as well as The estimation unit uses a pre-generated learning model generated through machine learning, which takes learning mold shape data and learning molded product difference data as input data and learning mold target shape data as teaching data, to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded product difference data.

2. The mold manufacturing support system according to claim 1, wherein, The mold making support system also includes a transformation unit that transforms the original data representing the actual shape of the mold into actual shape data of the mold for input to the learned model.

3. The mold making support system according to claim 1, wherein, The mold making support system also includes a CAM unit that transforms the target shape data of the mold into NC data, and a machine tool that corrects the actual shape of the mold based on the NC data.

4. The mold making support system according to claim 3, wherein, The mold making support system also includes a shape measuring device that measures the actual shape of the molded product formed by the mold after being corrected by the machine tool.

5. The mold making support system according to claim 4, wherein, The mold making support system also includes a subtraction calculation unit that calculates corrected difference data representing the difference between the actual shape of the molded article as measured by the shape measuring instrument and the target shape.

6. The mold making support system according to claim 5, wherein, The estimation unit uses the target shape data of the mold as the new actual shape data of the mold and the corrected difference data as the new difference data of the molded product to estimate the new target shape data of the mold.

7. The mold making support system according to claim 1, wherein, The mold making support system further includes: a display unit that displays the target shape data of the mold estimated by the learned model; and a receiving unit that accepts the user's decision on the target shape of the mold.

8. The mold making support system according to claim 5, wherein, The mold making support system also includes a learning unit that uses the target shape data of the mold derived from the learned model and the actual shape data of the molded article representing the actual shape of the molded article as measured by the shape measuring device to relearn the learned model.

9. A mold manufacturing support method, wherein, The mold manufacturing support method is processed as follows: Obtain actual mold shape data representing the actual shape of the mold, and molded article difference data representing the difference between the actual shape and the target shape of the molded article formed by the mold; as well as Using pre-generated, learned models generated through machine learning, with learning mold shape data and learning molded product difference data as input data and learning mold target shape data as teaching data, are used to estimate mold target shape data representing the target shape of the mold based on the actual mold shape data and the molded product difference data.

10. A method for generating a learning dataset, wherein, The method for generating the learning dataset involves the following processing: Prepare multiple mold shape data and multiple molded product shape data corresponding to the multiple mold shape data; Extract the first mold shape data and the second mold shape data from the plurality of mold shape data; Extract the first molded product shape data and the second molded product shape data corresponding to the first mold shape data and the second mold shape data from the plurality of molded product shape data; as well as Set the first mold shape data as the learning mold shape data, set the difference between the first molded product shape data and the second molded product shape data as the learning molded product difference data, and set the second mold shape data as the learning mold target shape data.

11. A mold making support system, wherein, The mold making support system has the following features: The acquisition unit acquires mold shape data representing the shape of the mold; and The estimation unit uses a pre-generated, learned model obtained through machine learning, which takes the mold shape data as input data and the molded article shape data as teaching data, to estimate the predicted shape data of the molded article based on the mold shape data.

12. The mold making support system according to claim 11, wherein, The mold making support system also includes a display unit that shows the mold shape data and the predicted shape data of the molded product.

13. The mold making support system according to claim 12, wherein, The display unit also displays molded article target shape data representing the target shape of the molded article.

14. The mold making support system according to claim 12, wherein, The mold-making support system also includes a receiving unit that accepts corrections to the mold shape data made by the user. The estimation unit estimates new predicted shape data for the molded product based on the corrected mold shape data. The display unit shows the new predicted shape data of the molded article.

15. A mold manufacturing support method, wherein, The mold manufacturing support method is processed as follows: Obtain mold shape data representing the shape of the mold; and Using a pre-generated, learned model generated through machine learning, with learning mold shape data as input data and learning molded product shape data as teaching data, the predicted shape data of the molded product is estimated based on the mold shape data.

Images