Mold manufacturing support system, mold manufacturing support method and program

Abstract

To provide a mold manufacturing support system capable of reducing the number of mold shape modifications.SOLUTION: The mold manufacturing support system includes: an acquisition unit that acquires an actual shape of a molded product formed by using an actual shape of the mold, a transformed shape of the molded product converted from the actual shape of the mold by the mold-molded product transformation by simulation, and a target shape of the molded product; and an estimation unit that estimates the target shape of the mold from the input data, in which an error between the transformed shape of the molded product and the actual shape of the molded product is added to the target shape of the molded product, by using an inverse transformation model for inverse transformation of mold-to-molded product transformation by simulation.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a mold manufacturing support system, a mold manufacturing support method, and a program. [Background technology]

[0002] Patent Document 1 discloses a technique for obtaining expected shape data of a mold by so-called CAE (Computer Aided Engineering). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2012-119010 Summary of the Invention [Problem to be solved by the invention]

[0004] In the conventional techniques described above, the number of times the die shape needs to be revised can increase due to insufficient forecast accuracy, etc. In particular, in recent years, in addition to problems of cracking and wrinkles due to the increased strength of materials, it has become difficult to satisfy the dimensional accuracy of the formed product due to springback after forming, and the number of times the die shape needs to be revised is on the rise.

[0005] The present invention has been made in consideration of the above-mentioned problems, and its main purpose is to provide a mold manufacturing support system, a mold manufacturing support method, and a program that can reduce the number of times the mold shape needs to be modified. [Means for solving the problem]

[0006] In order to solve the above problems, one aspect of the present invention provides a mold production support system that includes an acquisition unit that acquires the actual shape of a molded product molded from the actual shape of the mold, the transformed shape of the molded product converted from the actual shape of the mold through simulated mold-to-molded product conversion, and a target shape of the molded product, and an estimation unit that uses an inverse conversion model that performs an inverse conversion of the simulated mold-to-molded product conversion to estimate the target shape of the mold from input data that adds an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product. This makes it possible to reduce the number of times the mold shape needs to be modified.

[0007] In the above aspect, the inverse conversion model may be a trained model generated by machine learning using, as input data, a molded product shape converted from a mold shape by the mold-molded product conversion by the simulation, and the mold shape as training data. This makes it possible to obtain an inverse conversion model by machine learning.

[0008] In the above aspect, the estimation unit may repeatedly estimate the target shape of the mold using an actual shape of a new molded product that is molded using an actual shape of the mold that has been corrected based on the previously estimated target shape of the mold. This makes it possible to repeatedly estimate the target shape of the mold.

[0009] In the above aspect, the estimation unit may use the input data used in the previous estimation of the target shape of the mold as the transformed shape of the new molded product, thereby making it possible to omit calculation of the transformed shape of the new molded product.

[0010] In the above aspect, the estimation unit may use, as the input data, a result of adding a sum of errors between the target shape of the molded product and the actual shape of the molded product in the past to the target shape of the molded product, thereby making it possible to omit calculation of the transformed shape of the new molded product.

[0011] In the above aspect, a conversion unit may be further provided that converts the actual shape of the mold into a transformed shape of the molded product by the mold-molded product conversion using the simulation. This makes it possible to convert the actual shape of the mold into the transformed shape of the molded product.

[0012] In the above aspect, the apparatus may further include a calculation unit that calculates the input data by adding an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product. This makes it possible to calculate the input data.

[0013] In the above aspect, the present invention may further include a CAM unit that converts the target shape of the die into NC data, and a machine tool that corrects the actual shape of the die based on the NC data, thereby making it possible to correct the actual shape of the die based on the target shape of the die.

[0014] In the above aspect, the machine tool may further include a shape measuring device that measures the actual shape of a new molded product molded using the actual shape of the mold corrected by the machine tool. This makes it possible to measure the actual shape of the molded product molded using the corrected mold.

[0015] In the above aspect, the apparatus may further include a display unit that displays the estimated target shape of the mold, and a reception unit that receives a user's decision on the target shape of the mold, thereby allowing the user to see and decide on the target shape of the mold.

[0016] Another aspect of the mold production support method of the present invention acquires the actual shape of a molded product molded from the actual shape of the mold, the transformed shape of the molded product converted from the actual shape of the mold by mold-to-molded product conversion through simulation, and a target shape of the molded product, and estimates the target shape of the mold from input data that adds an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product using an inverse transformation model that performs an inverse transformation of the mold-to-molded product conversion through simulation. This makes it possible to reduce the number of times the mold shape needs to be modified.

[0017] In another aspect of the present invention, a program executes the following steps on a computer: acquiring an actual shape of a molded product molded from the actual shape of a mold, a transformed shape of the molded product converted from the actual shape of the mold by simulated mold-to-molded product conversion, and a target shape of the molded product; and using an inverse conversion model that performs an inverse conversion of the simulated mold-to-molded product conversion, estimating the target shape of the mold from input data that adds an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product. This makes it possible to reduce the number of times the mold shape needs to be modified. [Effects of the Invention]

[0018] According to the present invention, it is possible to reduce the number of times the mold shape needs to be modified. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 10 is a diagram for explaining a reference example. [Figure 2] FIG. 10 is a diagram for explaining a reference example. [Figure 3] FIG. 10 is a diagram for explaining estimation of mold target shape data. [Figure 4] FIG. 10 is a diagram illustrating a learning dataset. [Figure 5] FIG. 1 is a diagram illustrating an example of a mold production support system according to a first example. [Figure 6] FIG. 2 illustrates an example of a support device and a GUI terminal. [Figure 7] FIG. 10 is a diagram illustrating an example of a mold manufacturing support method. [Figure 8] FIG. 10 is a diagram illustrating an example of a mold production support system according to a second example. [Figure 9] FIG. 10 is a diagram illustrating an example of a mold production support system according to a third example. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this specification and the drawings, elements similar to those described above with reference to the previous drawings will be designated by the same reference numerals, and detailed descriptions thereof may be omitted as appropriate.

[0021] [Reference example] Before describing the embodiments of the present invention, a related reference example will be described. As shown in Fig. 1, in the reference example, mold actual shape data representing the actual shape of a mold and molded product difference data representing the difference between the actual shape of a molded product molded by the mold and a target shape are input to a trained AI model to estimate mold target shape data representing the target shape of the mold.

[0022] 2, in the reference example, a training data set is created by combining pairs of mold shapes and molded product shapes. The molded product shapes are, for example, molded product shapes converted from the mold shapes by a simulation using CAE's FEM (finite element method). For example, a mold shape and molded product shape under condition A and a mold shape and molded product shape under condition B are prepared, with the mold shape under condition A used as training mold shape data, the difference between the molded product shapes under conditions A and B used as training molded product difference data, and the mold shape under condition B used as training mold target shape data.

[0023] In this reference example, the input dimensions are "mold dimensions + molded product dimensions," which is a large number, resulting in the following problems: - The data volume is large and difficult to handle. - Multidimensional learning requires a large number of training datasets. - While it is possible to increase the number of training datasets by combining them, this further increases the data volume and makes it difficult to handle. - Training takes a long time, reducing development efficiency. - Inference takes a long time, resulting in long wait times during interactive operations.

[0024] Therefore, in the embodiment of the present invention, the above problem is solved by improving the estimation method of the above-mentioned reference example as described below.

[0025] [Embodiment] The estimation method of the above reference example is expressed by the following formula 1. That is, the previously proposed mold shape X k and the target shape of the molded product Y aim and the molded product shape J(X k ) difference Y aim -J(X k ) and input them into the AI ​​model to determine the next mold shape X k+1 is proposed.

[0026] X k represents the k-th proposed mold shape. The dimension of the mold shape is R n Y represents the shape of the molded product. aim is the target shape of the part. The dimension of the part shape is R m J represents the mold-molded product transformation during actual molding. J gives the dimension R n From R m AI represents the estimation by the AI ​​model of the reference example. The input dimension of AI is [R n ,R m ] and the output dimension is R n is.

[0027]

number

[0028] Mold shape k+1 is the target shape of the molded product Y obtained by the mold-molded product transformation J in the actual molding. aim The mold shape that realizes aim =J(X k+1 ) is desirable.

[0029] When creating an AI model, it is not realistic to prepare a large number of molded product shapes obtained through actual molding, so molded product shapes converted from the mold shape through simulation are used instead.

[0030] Specifically, the set of the mold shape and the molded product shape converted by the simulation (X a ,F(Xa )), (X b ,F(X b )) learning is performed so that the following formula 2 holds. In other words, by setting a = k and b = k + 1, formula 2 can be obtained from formula 1.

[0031] F represents the mold-to-part transformation by simulation. F reduces the dimension to R n From R m The mold-molded product conversion by simulation is, for example, a simulation using FEM (finite element method) in CAE.

[0032]

number

[0033] If the learning is complete, then Equation 2 will hold for all a and b. b ,F(X b ))=(F -1 (Y b ), Y b ) and we get the following formula 3. -1 represents the inverse transformation of the simulated mold-to-part transformation.

[0034]

number

[0035] Next, add J(X a )-J(X a ) to obtain the following formula 4.

[0036]

number

[0037] Next, the second term of AI in Equation 4 is replaced to obtain Equation 5 below.

[0038]

number

[0039] Next, Y l =Y b -(F(X a )-J(X a )) into Equation 5 to obtain Equation 6 below.

[0040]

number

[0041] Next, a=k, Y l =Y aim As such, the following Equation 7 is obtained from Equation 1 and Equation 6.

[0042]

number

[0043] As a result of the above transformation, the next mold shape X k+1 can be determined by the following formula 8, making it possible to make a proposal equivalent to the above reference example.

[0044]

number

[0045] That is, as shown in FIG. 3, in this embodiment, an inverse conversion model F that performs the inverse conversion of the mold-molded product conversion by simulation is used. -1 Using this, the target shape of the molded product Y aim Molded product transformation shape F(X k ) and the actual shape of the molded product J(X k ) from the input data with the error of k+1 Estimate.

[0046] According to this, in this embodiment, the input dimension is the "molded product dimension" (number of dimensions m), so the number of dimensions can be reduced compared to the input dimension of the above reference example, "mold dimension + molded product dimension" (number of dimensions n + m), and it becomes possible to solve the problems of the above reference example.

[0047] In this embodiment, an inverse conversion model F performs the inverse conversion of the mold-molded product conversion by simulation. -1 may be configured using a trained model.

[0048] That is, as shown in Figure 4, a learning data set containing the mold shape and the molded product shape converted from the mold shape by the mold-molded product conversion by simulation was prepared for each condition, and machine learning was performed using the molded product shape as input data and the mold shape as training data to generate the inverse conversion model F. -1 This can be achieved using a trained model.

[0049] Not only this, but also the reverse conversion model F -1 may be realized by simulation, similar to the simulated mold-to-molded part transformation.

[0050] [Variations] In the above embodiment, the molded product target shape Y aim Molded product transformation shape F(X k ) and the actual shape of the molded product J(X k ) error is added to the input data [Y aim +F(X k )-J(X k )] is calculated, and the molded product transformation shape F(X k ) has no data, so the mold shape X is calculated by converting the mold to the molded product through simulation. k It needs to be calculated from

[0051] However, the molded product transformation shape F(X k ) requires time and effort for simulation setup and preprocessing, which may be a bottleneck in shortening the production period.

[0052] Therefore, as shown in the following formula 9, the mold shape X k The previous mold shape X k-1 From the reverse conversion model F -1 It is noted that the formula (9) is calculated using the formula (10) below. That is, the molded product transformation shape F(X k ) is the previous mold shape X k When estimating the inverse transformation model F -1 The input data [Y aim +F(X k-1 )-J(X k-1 )].

[0053] By utilizing this, in the first modified example, the previous (kth) die target shape X k The input data [Y aim +F(X k-1 )-J(X k-1 )] to the new molded part transformation shape F(X k ) and the next (k+1) mold target shape X k+1 This allows the molded part transformation shape F(X k ) can be saved.

[0054]

number

[0055]

number

[0056] Furthermore, as shown in the following formula 11, the inverse transformation model F -1 The molded part transformation shape F(X k ) to the previous input data [Y aim +F(X k-1 )-J(X k-1 )] and repeat this recursively, and then F(X1)=Y aim (The initial molded product transformation shape F(X1) is the molded product target shape Y aim), then the inverse transformation model F -1 The input data for Y aim +Σ(Y aim -J(X n )) can be transformed into

[0057] Taking advantage of this, in the second modified example, the molded product target shape Y aim Past molded product target shape Y aim and the actual shape of the molded product J(X n ) and the result is input data [Y aim -Σ(Y aim -J(X n ))] to obtain the next (k+1) mold target shape X k+1 This also estimates the molded part transformation shape F(X k ) can be saved.

[0058] Note that F(X1)=Y aim Instead of replacing it with, as shown in the second line from the bottom of Equation 11, the inverse transformation model F -1 Input data of kY aim +F(X1)-ΣJ(X n ) can also be used.

[0059]

number

[0060] [Example 1] 5 is a diagram showing an example of the configuration of a die production support system 100A according to Example 1. The die production support system 100A is a system for supporting the production of dies used in press molding.

[0061] The die production support system 100A includes a support device 1, a GUI (Graphical User Interface) terminal 2, a CAM / machine tool 3, a press machine 4, a shape measuring instrument 5, and a molding DB (database) 6.

[0062] The support device 1 includes a computer including a CPU, RAM, ROM, nonvolatile memory, an input / output interface, etc. The GUI terminal 2 is a terminal that provides a GUI to the user U, and like the support device 1, includes a computer.

[0063] The CPU executes information processing in accordance with a program loaded from the ROM or non-volatile memory into the RAM. The program may be supplied via an information storage medium or a communication network.

[0064] In this example, the support device 1 and the GUI terminal 2 are in a server-client relationship. However, the support device 1 and the GUI terminal 2 may be integrated. In other words, the functions of the support device 1 and the GUI terminal 2 may be realized by a single device.

[0065] The CAM / machine tool 3 includes a CAM unit that generates NC data and a machine tool that manufactures a mold based on the NC data.

[0066] The press machine 4 produces a molded product by press molding using a mold.

[0067] The shape measuring instrument 5 measures the shape of the produced molded product. The shape measuring instrument 5 is, for example, a 3D scanner or a contact or non-contact displacement sensor.

[0068] The molding DB6 is a database that stores CAE mold shape data and molded product shape data. The molding DB6 may also store actual shape data obtained through testing and measurement of molds and molded products that have been manufactured in the past.

[0069] The steps (a) to (j) shown in FIG. 5 will be explained below.

[0070] (a) A user U inputs mold actual shape data representing the actual shape of a mold, molded product actual shape data representing the actual shape of a molded product molded by the mold, and molded product target shape data representing the target shape of the molded product into the GUI terminal 2. Each piece of data is, for example, CAE data, CAD data, STL data, or the like.

[0071] (b) The GUI terminal 2 generates data to be used as input for the inverse conversion model and outputs it to the support device 1. Specifically, the GUI terminal 2 converts the actual mold shape data into converted molded product shape data by a mold-molded product conversion using simulation, and outputs the actual molded product shape data, the converted molded product shape data, and the target molded product shape data to the support device 1.

[0072] (c) The support device 1 uses an inverse conversion model that performs the inverse conversion of the mold-molded product conversion by simulation to estimate mold target shape data that represents the target shape of the mold from input data obtained from the GUI terminal 2 by adding errors between the molded product conversion shape and the molded product actual shape (see Figure 3), and outputs the data to the GUI terminal 2.

[0073] (d) The GUI terminal 2 displays the die target shape data output from the support device 1. The user U checks the die target shape displayed on the GUI terminal 2 and determines whether to accept it. (e) When the GUI terminal 2 receives the die target shape decision made by the user, (f) the GUI terminal 2 outputs the die target shape data.

[0074] (g) User U inputs die target shape data into CAM / machine tool 3. (h) CAM / machine tool 3 converts the die target shape data into NC data and modifies the actual shape of the die based on the converted NC data. (i) Press 4 forms a molded product using the modified die.

[0075] (j) The shape measuring instrument 5 measures the actual shape of a new molded product molded by the mold modified by the machine tool. Then, steps (a) to (j) are performed again using the mold target shape data of the modified mold and the molded product actual shape data of the new molded product.

[0076] The above-described steps (a) to (j) are repeated until there is no difference between the target shape and the actual shape of the molded product.

[0077] 6 is a block diagram showing an example of the configuration of the support device 1 and the GUI terminal 2. The support device 1 includes an acquisition unit 11, a calculation unit 12, and an estimation unit 13. These functional units are realized by the CPU of the support device 1 executing information processing in accordance with a program loaded from the ROM or non-volatile memory to the RAM.

[0078] The GUI terminal 2 includes a conversion unit 21. This functional unit is realized by the CPU of the GUI terminal 2 executing information processing in accordance with a program loaded from the ROM or nonvolatile memory to the RAM.

[0079] The GUI terminal 2 also includes a reception unit 22 and a display unit 23. The reception unit 22 is, for example, a keyboard or a mouse, and receives operations from the user U. The display unit 23 is, for example, a liquid crystal display.

[0080] 7 is a flow chart showing an example of the procedure of the mold production support method realized in the mold production support system 100 A. Each of the support device 1 and the GUI terminal 2 executes the information processing shown in the drawing in accordance with a program.

[0081] First, the GUI terminal 2 acquires mold actual shape data representing the actual shape of the mold, molded product actual shape data representing the actual shape of the molded product using the mold, and molded product target shape data representing the target shape of the molded product (S21, corresponding to step (a) above).

[0082] The mold actual shape data, molded product actual shape data, and molded product target shape data are represented by point cloud data such as CAE data, CAD data, STL data, etc. When the number of measurement points is large, dimension reduction may be performed by feature extraction.

[0083] Next, the GUI terminal 2 converts the actual mold shape data into converted molded product shape data by a mold-molded product conversion using simulation (S22, processing as the conversion unit 21, corresponding to the above step (b)).

[0084] The support device 1 acquires the molded product actual shape data, the molded product converted shape data, and the molded product target shape data from the GUI terminal 2 (S11, processing as the acquisition unit 11).

[0085] Next, the support device 1 calculates input data by adding an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product (S12, processing as the calculation unit 12).

[0086] Next, the support device 1 estimates the die target shape data from the input data using an inverse transformation model that performs the inverse transformation of the die-molded product transformation by simulation (S13, processing as the estimation unit 13, corresponding to the above step (c)).

[0087] The GUI terminal 2 displays the die target shape data estimated by the support device 1 on the display unit 23 (S23, corresponding to the above step (d)).

[0088] Next, when the GUI terminal 2 receives the determination of the die target shape by the user at the receiving unit 22 (S24: YES), it outputs die target shape data (S25, corresponding to the above steps (e) and (f)). Note that the GUI terminal 2 may receive a correction of the die target shape by the user and cause the support device 1 to execute the estimation process again using the corrected die target shape.

[0089] The mold target shape data is used to modify the actual shape of the mold using CAM / machine tool 3, and the shape of the new molded product formed using the modified mold is measured by shape measuring instrument 5 (corresponding to steps (g) to (j) above).

[0090] The support device 1 repeatedly estimates the mold target shape data using actual product shape data of a new molded product molded with the actual shape of the mold corrected based on the previously estimated mold target shape data.

[0091] As in the above-described modified example, when estimating the mold target shape data, the input data used in the previous estimation of the mold target shape data may be used as new molded product converted shape data, or the result of adding the sum of errors between the previous molded product target shape and the molded product actual shape to the molded product target shape may be used as input data.

[0092] According to this example, an appropriate target mold shape can be quickly obtained using a trained model without relying on the experience, know-how, or intuition of an expert, thereby reducing the number of times the mold shape needs to be modified.

[0093] [Example 2] 8 is a diagram showing an example of the configuration of a mold production support system 100B according to the second example. In this example, (e) when the GUI terminal 2 accepts the determination of the mold target shape, (k) the mold target shape data is directly output from the GUI terminal 2 to the CAM / machine tool 3. This makes it possible to realize automation of mold correction and reduce the burden on the user U.

[0094] [Example 3] 9 is a diagram showing an example of the configuration of a mold production support system 100C according to a third example. In this example, when (j) the shape measuring instrument 5 measures the actual shape of a molded product, (m) the mold shape data and molded product shape data are registered in the molding DB 6, and are used for re-learning the trained model. This makes it possible to improve estimation accuracy.

[0095] Although this example is obtained by adding step (m) to the first example, the present invention is not limited to this, and step (m) may be added to the second example.

[0096] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made by those skilled in the art. [Explanation of symbols]

[0097] 1 Support device, 2 GUI terminal, 3 CAM / machine tool, 4 Press machine, 5 Shape measuring instrument, 6 Molding DB, 11 Acquisition unit, 12 Calculation unit, 13 Estimation unit, 21 Conversion unit, 22 Reception unit, 23 Display unit, 100 Mold production support system

Claims

1. an acquisition unit that acquires an actual shape of a molded product molded from the actual shape of a mold, a converted shape of the molded product converted from the actual shape of the mold by mold-molded product conversion through simulation, and a target shape of the molded product; an estimation unit that estimates a target shape of the mold from input data that adds an error between the converted shape of the molded product and the actual shape of the molded product to the target shape of the molded product using an inverse conversion model that performs an inverse conversion of the mold-molded product conversion by the simulation; A mold manufacturing support system equipped with the above.

2. The inverse conversion model is a trained model generated by machine learning using the molded product shape converted from the mold shape by the mold-molded product conversion by the simulation as input data and the mold shape as training data. The mold manufacturing support system according to claim 1 .

3. the estimation unit repeatedly estimates the target shape of the die using an actual shape of a new molded product molded by an actual shape of the die corrected based on the previously estimated target shape of the die. The mold manufacturing support system according to claim 1 .

4. The estimation unit uses the input data used in the previous estimation of the target shape of the mold as the transformed shape of the new molded product. The mold manufacturing support system according to claim 3.

5. the estimation unit uses, as the input data, a result of adding a sum of errors between the target shape of the molded product and the actual shape of the molded product in the past to the target shape of the molded product. The mold manufacturing support system according to claim 3.

6. Further provided is a conversion unit that converts the actual shape of the mold into a converted shape of the molded product by the mold-molded product conversion by the simulation. The mold manufacturing support system according to claim 1 .

7. a calculation unit that calculates the input data by adding an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product, The mold manufacturing support system according to claim 1 .

8. The system further includes a CAM unit that converts the target shape of the die into NC data, and a machine tool that corrects the actual shape of the die based on the NC data. The mold manufacturing support system according to claim 1 .

9. The machine tool further includes a shape measuring device for measuring the actual shape of a new molded product molded using the actual shape of the mold corrected by the machine tool. The mold manufacturing support system according to claim 8.

10. The apparatus further includes a display unit that displays the estimated target shape of the die, and a reception unit that receives a user's determination of the target shape of the die. The mold manufacturing support system according to claim 1 .

11. Acquire an actual shape of a molded product molded from the actual shape of the mold, a converted shape of the molded product converted from the actual shape of the mold by mold-molded product conversion through simulation, and a target shape of the molded product; Using an inverse transformation model that performs an inverse transformation of the mold-molded product transformation by the simulation, a target shape of the mold is estimated from input data that adds an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product. Mold making support method.

12. Obtaining an actual shape of a molded product molded by the actual shape of the mold, a transformed shape of the molded product transformed from the actual shape of the mold by mold-molded product transformation by simulation, and a target shape of the molded product; and Using an inverse transformation model that performs an inverse transformation of the mold-molded product transformation by the simulation, estimating the target shape of the mold from input data that adds an error between the transformed shape of the molded product and the actual shape of the molded product to the target shape of the molded product; A program for running on a computer.

Images