Machine learning device and design assistance device

Abstract

The present invention provides a machine learning device and a design support device, which can predict the molding shrinkage rate corresponding to the pre-set molding conditions in the mold design. The machine learning device has: an input data acquisition unit, which acquires input data containing arbitrary molding conditions, wherein the molding conditions at least include the type of resin, the type of additive, the mixing ratio of the additive, the surface temperature of the mold, and the product of the holding pressure and the holding pressure time in the molding of an arbitrary molded product by an arbitrary injection molding machine; a label acquisition unit, which acquires label data, wherein the label data represents the molding shrinkage rate of the resin in the flow direction measured in the molded product molded under the molding conditions included in the input data, and the molding shrinkage rate in the vertical direction perpendicular to the flow direction; and a learning unit, which uses the input data acquired by the input data acquisition unit and the label data acquired by the label acquisition unit to perform supervised learning and generate a learning completion model.

Description

Technical Field

[0001] The present invention relates to a machine learning device and a design assistance device. Background Art

[0002] Resin products such as plastics (hereinafter also referred to as "molded products") are molded by, for example, injecting resin into a mold using an injection molding machine. In other words, molded products of various desired shapes can be molded by designing the mold.

[0003] In the design of the mold, after the melted resin flows into the mold (cavity) and cools, it is taken out of the mold as a molded product. However, considering the molding shrinkage that reduces the volume of the resin due to the cooling, the size of the mold (cavity) needs to be designed to be larger than the molded product. For example, the molding shrinkage rate of ABS (acrylonitrile butadiene styrene) is 4 / 1000 to 9 / 1000, so the total length of 1000mm ABS shrinks by 4mm to 9mm.

[0004] In addition, the molding shrinkage rate is different in the flow direction of the melted resin in the mold and in the vertical direction perpendicular to the flow direction. That is, the molding shrinkage rate in the flow direction is small, and the molding shrinkage rate in the vertical direction is large.

[0005] In addition, the molding shrinkage rate varies greatly depending on the additives such as glass fiber added to the resin. For example, when 30% of glass fiber is mixed with polyphenylene sulfide (PPS), the molding shrinkage rate in the flow direction is 0.3% to 0.4%, and the molding shrinkage rate in the perpendicular direction is 0.6% to 0.7%. In addition, when 40% of glass fiber is mixed with liquid crystal polymer (LCP), the molding shrinkage rate in the flow direction is 0.1% to 0.15%, and the molding shrinkage rate in the perpendicular direction is 0.3% to 0.5%. In other words, the molding shrinkage rate in the perpendicular direction is about 2 to 3 times larger than the molding shrinkage rate in the flow direction.

[0006] In addition, in resin mixed with glass fibers, the molding shrinkage is not uniform in all directions, but depends largely on the flow direction. The reason is believed to be that when flowing into the mold, orientation occurs, and the resin molecules and glass fibers are aligned in a certain direction.

[0007] Therefore, the following technology is known: the volume shrinkage distribution of the molded product is obtained by flow analysis, the birefringence distribution is obtained by residual stress analysis, and the relationship between the birefringence and the anisotropy of the molding shrinkage is measured, and the volume shrinkage rate is given to the molding shrinkage rates in the flow direction and the perpendicular direction according to the shrinkage anisotropy measured according to the birefringence distribution, so as to predict the distribution of the molding shrinkage amount and the shrinkage direction of the molded product. For example, refer to Patent Document 1.

[0008] Prior art literature

[0009] Patent Document 1: Japanese Patent Application Publication No. 2007-83602

[0010] However, in the prior art, factors such as the type of resin, the type of additives mixed into the resin, and the mixing ratio thereof are not taken into consideration, and it is sometimes difficult to calculate a strict molding shrinkage rate. In addition, in mold design, it is difficult to pre-set the molding conditions of the molded product and the molding shrinkage rates in the flow direction and the vertical direction, and the operator needs corresponding experience. Therefore, there is a problem that in mold design, the process of repeating the trial production of the mold, the trial injection of the molded product using the trial production, the measurement of the dimensions of the molded product after the trial injection, and then designing the mold (or adjusting the molding conditions) increases the number of trial production of the mold and increases the cost. Summary of the invention

[0011] Therefore, it is desired to predict the molding shrinkage rate corresponding to the preset molding conditions in the design of the mold.

[0012] (1) One method of the machine learning device disclosed in the present invention comprises: an input data acquisition unit, which acquires input data including arbitrary molding conditions, wherein the molding conditions at least include the type of resin, the type of additive, the mixing ratio of the additive, the surface temperature of the mold, and the product of the holding pressure and the holding time in the molding of an arbitrary molded product by an arbitrary injection molding machine; a label acquisition unit, which acquires label data, wherein the label data represents the molding shrinkage rate of the resin in the flow direction and the molding shrinkage rate in the direction perpendicular to the flow direction measured in the molded product molded under the molding conditions included in the input data; and a learning unit, which uses the input data acquired by the input data acquisition unit and the label data acquired by the label acquisition unit to perform supervised learning and generate a learning completion model.

[0013] (2) One mode of the design assistance device disclosed in the present invention comprises: a learning completion model, which is generated by the machine learning device of (1); an input unit, which inputs pre-set forming conditions; and a prediction unit, which inputs the forming conditions input by the input unit into the learning completion model, and predicts the forming shrinkage rate in the flow direction and the forming shrinkage rate in the vertical direction under the forming conditions.

[0014] Effects of the Invention

[0015] According to one embodiment, the molding shrinkage rate corresponding to the preset molding conditions can be predicted in the design of the mold. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a functional block diagram showing an example of a functional configuration of a design support system according to one embodiment.

[0017] Figure 2 It means providing to Figure 1 A diagram of an example of a learning completion model of a design support device.

[0018] Figure 3 This is a flowchart for explaining the prediction processing of the design support device in the application phase.

[0019] Figure 4 This is a diagram showing an example of the configuration of a design support system.

[0020] Figure 5 This is a diagram showing an example of the configuration of a design support system.

[0021] Explanation of symbols

[0022] 10 CAD device

[0023] 20 Design aids

[0024] 201 Input

[0025] 202 Forecasting Department

[0026] 250 Learning Complete Model

[0027] 30 Machine Learning Devices

[0028] 301 Input data acquisition unit

[0029] 302 Label acquisition unit

[0030] 303 Learning Department

[0031] 50 servers DETAILED DESCRIPTION

[0032] Hereinafter, one embodiment of the present disclosure will be described using the drawings.

[0033] <One embodiment>

[0034] Figure 1 1 is a functional block diagram showing an example of a functional structure of a design support system according to one embodiment. Figure 1 As shown, the design support system includes a CAD (Computer-Aided Design) device 10 , a design support device 20 , and a machine learning device 30 .

[0035] The CAD device 10, the design assisting device 20, and the machine learning device 30 can be directly connected to each other via a connection interface not shown in the figure. In addition, the CAD device 10, the design assisting device 20, and the machine learning device 30 can also be connected to each other via a network not shown in the figure, such as a LAN (Local Area Network) or the Internet. In this case, the CAD device 10, the design assisting device 20, and the machine learning device 30 have a communication unit not shown in the figure that communicates with each other through such a connection. In addition, as described later, the CAD device 10 may include a design assisting device 20 and a machine learning device 30.

[0036] The CAD device 10 uses the CPU (Central Processing Unit) to operate the CAD software for drawing on the computer screen. In addition, the CAD device 10 draws (designs) the mold for forming the molded product based on the CAD data of the pre-designed molded product, the pre-set molding conditions, and the flow direction of the resin and the molding shrinkage in the vertical direction predicted by the design auxiliary device 20 described later using the molding conditions. Mold drawing is performed using two-dimensional CAD or three-dimensional CAD. When using two-dimensional CAD, a front view, a top view, a side view, etc. of the mold are made on the X and Y planes. When using three-dimensional CAD, a three-dimensional image of the mold is made in the three-dimensional space of X, Y and Z.

[0037] The design support device 20 can obtain the pre-set molding conditions from the CAD device 10 during the mold design performed by the CAD device 10 in the application stage. The design support device 20 inputs the obtained molding conditions into the learning model provided by the machine learning device 30 described later, thereby predicting the molding shrinkage in the flow direction and the perpendicular direction.

[0038] In addition, the molding conditions include at least the type of resin (material) (thermosetting resins such as epoxy resins or polyurethane resins, thermoplastic resins such as polyesters or polyvinyl chloride), the type of additives (glass fiber, stabilizer, colorant, etc.), the mixing ratio of the additives, the surface temperature of the mold, and the product of the holding pressure and the holding time, which have a great influence on the molding shrinkage rate.

[0039] Hereinafter, the relationship between the type of resin, the type of additive, the mixing ratio of the additive, the surface temperature of the mold, and the product of the holding pressure and the holding pressure time included in the molding conditions and the molding shrinkage rate will be briefly described.

[0040] (1) Resin (material) type: The basic shrinkage range is determined by the type of resin material. However, there are subtle differences in shrinkage due to the resin manufacturer or resin grade.

[0041] (2) Types and mixing ratios of additives: In resins mixed with glass fibers, the molding shrinkage tends to be smaller. In addition, as described above, in resins mixed with glass fibers, the molding shrinkage in the flow direction is smaller than the molding shrinkage in the perpendicular direction.

[0042] (3) Mold surface temperature: The surface temperature of the mold cavity during molding changes the molding shrinkage of the resin. Generally speaking, the higher the surface temperature, the greater the molding shrinkage.

[0043] (4) Product of holding pressure and holding time: The molding shrinkage rate also changes due to the product of the holding height and the holding time after resin filling. Generally speaking, when the holding pressure is high and the holding time is long, the molding shrinkage rate tends to decrease.

[0044] Next, before describing the design support device 20 , machine learning for generating a learned model will be described.

[0045] <Machine learning device 30>

[0046] The machine learning device 30 acquires, for example, in advance any molding conditions for molding any molded product by any injection molding machine as input data.

[0047] Furthermore, the machine learning device 30 acquires, as labels (correct solutions), the mold shrinkage rate in the resin flow direction and the mold shrinkage rate in the perpendicular direction measured in the molded product molded under the molding conditions in the acquired input data.

[0048] The machine learning device 30 performs supervised learning using the acquired training data of the input data and label combinations, and constructs a learning completion model described later.

[0049] In this way, the machine learning device 30 can provide the constructed learning completion model to the design support device 20.

[0050] The machine learning device 30 will be described in detail.

[0051] like Figure 1 As shown, the machine learning device 30 includes an input data acquisition unit 301 , a label acquisition unit 302 , a learning unit 303 , and a storage unit 304 .

[0052] The input data acquisition unit 301 acquires any molding condition of any molded product of any injection molding machine as input data from an injection molding machine (not shown) or a control device (not shown) that controls the injection molding machine, etc., via a communication unit (not shown) during the learning phase. In addition, when the mold prototype designed by the CAD device 10 is used to test the injection molded product under the pre-set molding conditions, the input data acquisition unit 301 can acquire the molding conditions as input data. The input data acquisition unit 301 outputs the acquired input data to the storage unit 304.

[0053] The label acquisition unit 302 acquires data indicating the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction of the resin as label data (correct solution data), wherein the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction are shrinkage rates measured during the dimensional inspection of the molded product obtained by molding under the molding conditions of the acquired input data when the molded product is taken out of the mold. In addition, when the label acquisition unit 302 uses the mold trial product with the pre-set molding conditions to test the injection molded product, the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction measured during the dimensional inspection of the molded product can be acquired as label data (correct solution data). The label acquisition unit 302 outputs the acquired label data to the storage unit 304.

[0054] The molding shrinkage rate in the flow direction (or perpendicular direction) can be calculated by comparing the dimensions of each portion of the molded product along the flow direction (or perpendicular direction) with the dimensions of each portion of the mold.

[0055] The learning unit 303 accepts the above-mentioned input data and label combination as training data, and uses the accepted training data to perform supervised learning, thereby constructing a learning completion model 250 that predicts the molding shrinkage rate in the flow direction of the resin and the molding shrinkage rate in the perpendicular direction according to the preset molding conditions.

[0056] Then, the learning unit 303 provides the constructed learning model 250 to the design support device 20 .

[0057] In addition, it is desirable to prepare a plurality of training data for supervised learning. For example, the training data may be obtained from injection molding machines actually operated at various locations such as a customer's factory or from control devices of the injection molding machines.

[0058] Figure 2 It means providing to Figure 1 FIG. 2 is a diagram showing an example of a learning completed model 250 of the design support device 20. Figure 2As shown, the learning completion model 250 is illustrated as a multi-layer neural network with the type of resin, the type of additive, the mixing ratio of the additive, the surface temperature of the mold, and the molding conditions of the product of the holding pressure and the holding time as the input layer, and the data representing the "molding shrinkage rate" in the flow direction and the "molding shrinkage rate" in the vertical direction as the output layer.

[0059] In addition, after constructing the learned model 250, the learning unit 303 further performs supervised learning on the learned model 250 when new training data is obtained, thereby updating the previously constructed learned model 250.

[0060] The above-mentioned supervised learning can be performed through online learning. In addition, supervised learning can also be performed through batch learning. In addition, supervised learning can also be performed through small batch learning.

[0061] Online learning is a learning method that performs supervised learning every time when training data is produced by injection molding of an injection molding machine (not shown). Batch learning is a learning method that collects a plurality of training data corresponding to the repetitions while producing training data by injection molding of an injection molding machine (not shown), and performs supervised learning using all the collected training data. Small batch learning is a learning method that is between online learning and batch learning, and performs supervised learning every time a certain amount of training data is accumulated.

[0062] The storage unit 304 is a RAM (Random Access Memory) or the like, and stores input data acquired by the input data acquisition unit 301 , label data acquired by the label acquisition unit 302 , and the learned model 250 constructed by the learning unit 303 .

[0063] In the above, the machine learning for generating the learned model 250 included in the design supporting device 20 has been described.

[0064] Next, the design support device 20 in the application stage will be described.

[0065] <Design support device in the application stage 20>

[0066] like Figure 1 As shown, the design support device 20 in the application stage is configured to include an input unit 201 , a prediction unit 202 , an output unit 203 , and a storage unit 204 .

[0067] In addition, the design support device 20 is designed to achieve Figure 1The design support device 20 has an unillustrated operation processing device such as a CPU to control the operation of the functional blocks. In addition, the design support device 20 has an unillustrated auxiliary storage device such as a ROM (Read Only Memory) or HDD that stores various control programs, and an unillustrated main storage device such as a RAM that stores data temporarily required after the operation processing device executes the program.

[0068] Furthermore, in the design support device 20, the operation processing device reads the OS or application software from the auxiliary storage device, expands the read OS or application software in the main storage device, and performs operation processing based on the OS or application software. The design support device 20 controls each hardware according to the operation result. Figure 1 That is, the design support device 20 can be implemented by the cooperation of hardware and software.

[0069] When the CAD device 10 designs (or redesigns) a mold, the input unit 201 inputs, for example, a molding condition set in advance from the CAD device 10. The input unit 201 outputs the input molding condition to the prediction unit 202.

[0070] The prediction unit 202 inputs the molding conditions obtained from the input unit 201 into the Figure 2 The learned model 250 predicts the molding shrinkage rate in the resin flow direction and the molding shrinkage rate in the perpendicular direction of the pre-designed molded product.

[0071] The output unit 203 can output the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction predicted by the prediction unit 202 to an output device (not shown) such as a liquid crystal display included in the CAD device 10. In addition, the output unit 203 can output by sound via a speaker (not shown).

[0072] In this way, the CAD device 10 can calculate the size of the mold (cavity) to design the mold based on the CAD data of the pre-designed molded product, the pre-set molding conditions, and the predicted molding shrinkage rates in the flow direction and the vertical direction. Afterwards, when trial-making the designed mold, a trial injection is performed using the trial-made mold, and the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction are measured when the size of the molded product after the trial injection is checked. In this case, in the machine learning device 30, the input data acquisition unit 301 acquires the pre-set molding conditions as input data, and the label acquisition unit 302 acquires the measured molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction as labels. In addition, the learning unit 303 accepts the combination of input data and labels as new training data, and updates the learning completion model 250 by performing supervised learning.

[0073] In addition, when the pre-set molding conditions are adjusted and the same trial mold is used to test the injection molded product, the input data acquisition unit 301 can acquire the adjusted molding conditions as input data, and the label acquisition unit 302 can acquire the measured molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction as labels. The learning unit 303 accepts the combination of the input data and the label as new training data, and updates the learning completion model 250 by performing supervised learning.

[0074] When the mold is redesigned by the CAD device 10, the input unit 201 inputs the preset molding conditions from the CAD device 10. The prediction unit 202 inputs the input molding conditions into the updated learning model 250, thereby predicting the molding shrinkage rate in the new flow direction and the molding shrinkage rate in the perpendicular direction.

[0075] That is, the mold is designed by the CAD device 10, the designed mold is trial-produced, the trial-produced mold is used to test-inject the molded product, and the molding shrinkage rates in the flow direction and the vertical direction are measured when the dimensions of the molded product after the trial injection are checked. And, when this series of processes are repeated, the learning model 250 is updated. As a result, the accuracy of the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction predicted by the design support device 20 using the learning model 250 is improved, so that the final mold shape and the final molding conditions can be determined with a small number of mold trial productions.

[0076] The storage unit 204 is a ROM, a HDD, or the like, and can store various control programs and the learned model 250 .

[0077] <Prediction Processing of the Design Support Device 20 in the Application Phase>

[0078] Next, the operation of the prediction process of the design support device 20 according to the present embodiment will be described.

[0079] Figure 3 This is a flowchart for explaining the prediction processing of the design support device 20 in the application phase.

[0080] In step S11 , the input unit 201 inputs the molding conditions set in advance from the CAD device 10 when the CAD device 10 designs (or redesigns) the mold.

[0081] In step S12 , the prediction unit 202 inputs the molding conditions input in step S11 to the learned model 250 , and predicts the molding shrinkage rate in the resin flow direction and the molding shrinkage rate in the perpendicular direction in the molded product designed in advance.

[0082] In step S13 , the output unit 203 outputs the molding shrinkage rates in the flow direction and the perpendicular direction predicted in step S12 to the CAD device 10 .

[0083] As described above, the design support device 20 according to one embodiment inputs the preset molding conditions into the learning model 250 when designing (or redesigning) the mold of the pre-designed molded product, thereby predicting the molding shrinkage rate corresponding to the molding conditions.

[0084] In addition, the design assisting device 20 outputs the predicted molding shrinkage rates in the flow direction and the vertical direction to the CAD device 10, whereby the CAD device 10 can design the mold based on the CAD data of the pre-designed molded product, the pre-set molding conditions, and the predicted molding shrinkage rates in the flow direction and the vertical direction. Moreover, whenever the process of trial production of the mold, trial injection of the molded product using the trial-produced mold, dimensional measurement of the molded product after the trial injection, and redesign of the mold is repeated, the learning completion model 250 is updated by the machine learning device 30. As a result, the accuracy of the molding shrinkage rate in the flow direction and the molding shrinkage rate in the vertical direction predicted by the design assisting device 20 using the learning completion model 250 is improved, so the number of trial production of the mold can be reduced, and the cost of the molded product can be reduced. In addition, the time cost of trial injection using the trial-produced mold can also be reduced.

[0085] Furthermore, by using the learned model 250 , the design support device 20 eliminates the need for the operator to predict the molding shrinkage ratio in the resin flow direction and the perpendicular direction in the designed molded product, thereby reducing the burden on the operator.

[0086] Although one embodiment has been described above, the design support device 20 and the machine learning device 30 are not limited to the above-described embodiment, and include modifications and improvements within a range that can achieve the purpose.

[0087] <Variation 1>

[0088] In the above embodiment, the machine learning device 30 is exemplified as a device different from the CAD device 10 and the design support device 20 , but the CAD device 10 or the design support device 20 may have a part or all of the functions of the machine learning device 30 .

[0089] <Variation 2>

[0090] For another example, in the above embodiment, the design support device 20 is illustrated as a device different from the CAD device 10 , but the CAD device 10 may have a part or all of the functions of the design support device 20 .

[0091] Alternatively, for example, the server may include part or all of the input unit 201, prediction unit 202, output unit 203, and storage unit 204 of the design support device 20. Furthermore, each function of the design support device 20 may be implemented using a virtual server function on the cloud.

[0092] Furthermore, the design support apparatus 20 may be a distributed processing system in which the functions of the design support apparatus 20 are appropriately distributed to a plurality of servers.

[0093] <Variation 3>

[0094] For another example, in the above embodiment, the pre-set molding conditions include: the type of resin (material) (thermosetting resins such as epoxy resins, thermoplastic resins such as polyesters), the type of additives (glass fiber, stabilizers, colorants, etc.), the mixing ratio of the additives, the surface temperature of the mold, and the product of the holding pressure and the holding time, but are not limited to these.

[0095] For example, the product of holding pressure and holding time may be set in consideration of the cooling time after resin filling, and the molding conditions may include the product of holding pressure and holding time and the cooling time as independent parameters.

[0096] Furthermore, the surface temperature of the mold may be set in consideration of the cooling rate after resin filling, and the molding conditions may include the surface temperature of the mold and the cooling rate as independent parameters.

[0097] Furthermore, the molding conditions may include wall thickness information of the molded product represented by CAD data of the molded product designed in advance.

[0098] <Variation 4>

[0099] For another example, in the above embodiment, the design support device 20 uses the learning model 250 provided by the machine learning device 30 to predict the molding shrinkage rate in the flow direction and the vertical direction of the resin based on the molding conditions obtained from a CAD device 10, but the present invention is not limited to this. Figure 4 As shown, the server 50 can store the learning model 250 generated by the machine learning device 30, and share the learning model 250 with m design support devices 20A(1)-20A(m) connected to the network 60 (m is an integer greater than or equal to 2). Thus, even if a new CAD device and design support device are configured, the learning model 250 can be applied.

[0100] In addition, each of the design support devices 20A( 1 ) to 20A(m) is connected to each of the CAD devices 10A( 1 ) to 10A(m).

[0101] In addition, each of the CAD devices 10A(1) to 10A(m) is associated with Figure 1Each of the design support devices 20A(1)-20A(m) corresponds to the CAD device 10. Figure 1 The design assisting device 20 corresponds to the embodiment of the present invention.

[0102] Or, if Figure 5 As shown, the server 50 operates as the design support device 20, for example, and predicts the molding shrinkage rate in the flow direction and the perpendicular direction based on the acquired molding conditions for each of the CAD devices 10A(1)-10A(m) connected to the network 60. Thus, even if a new CAD device is installed, the learned model 250 can be applied.

[0103] In addition, each function included in the design support device 20 and the machine learning device 30 in one embodiment can be realized by hardware, software, or a combination thereof. Here, the so-called realization by software means that the computer reads the program and executes it.

[0104] The various structural parts included in the design assistance device 20 and the machine learning device 30 can be implemented by hardware including electronic circuits, software, or a combination thereof. When implemented by software, the program constituting the software is installed on a computer. In addition, these programs can be recorded in a removable medium to be distributed to users, and can also be downloaded to the user's computer via a network for distribution. In addition, when constituted by hardware, for example, integrated circuits (ICs) such as ASIC (Application Specific Integrated Circuit), gate array FPGA (Field Programmable Gate Array), and CPLD (Complex Programmable Logic Device) can be used to constitute part or all of the functions of the various structural parts included in the above-mentioned devices.

[0105] Various types of non-transitory computer readable media (Non-transitory computer readable medium) can be used to store the program and provide it to the computer. Non-transitory computer readable media include various types of tangible recording media (Tangible storage medium). Examples of non-transitory computer readable media include: magnetic storage media (e.g., floppy disks, magnetic tapes, hard disk drives), magneto-optical storage media (e.g., optical magnetic disks), CD-ROM (Read Only Memory), CD-R, CD-R / W, semiconductor memories (e.g., mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM). In addition, the program can also be supplied to the computer through various types of transient computer readable media (Transitory computer readable medium). Examples of transient computer readable media include electrical signals, optical signals, and electromagnetic waves. Transitory computer readable media can supply the program to the computer via wired communication paths such as wires and optical fibers or wireless communication paths.

[0106] Furthermore, the steps describing the program recorded in the recording medium naturally include processing that is performed in time series in the order in question, and also include processing that is not necessarily performed in time series but is executed in parallel or individually.

[0107] In other words, the machine learning device and design support device of the present disclosure can take various embodiments having the following structures.

[0108] (1) The machine learning device 30 disclosed in the present invention comprises: an input data acquisition unit 301, which acquires input data including arbitrary molding conditions, wherein the molding conditions at least include the type of resin, the type of additives, the mixing ratio of the additives, the surface temperature of the mold, and the product of the holding pressure and the holding time in the molding of an arbitrary molded product by an arbitrary injection molding machine; a label acquisition unit 302, which acquires label data, wherein the label data represents the molding shrinkage rate of the resin in the flow direction and the molding shrinkage rate in the direction perpendicular to the flow direction measured in the molded product molded under the molding conditions included in the input data; and a learning unit 303, which uses the input data acquired by the input data acquisition unit 301 and the label data acquired by the label acquisition unit 302 to perform supervised learning and generate a learning completion model 250.

[0109] According to the machine learning device 30, a learning model 250 that predicts the molding shrinkage rate in the flow direction and the molding shrinkage rate in the perpendicular direction in the molded product can be generated.

[0110] (2) The design assistance device 20 disclosed in the present invention comprises: a learning completion model 250, which is generated by the machine learning device 30 of (1); an input unit 201, which inputs a pre-set forming condition; and a prediction unit 202, which inputs the forming condition input by the input unit 201 into the learning completion model 250, and predicts the forming shrinkage rate in the flow direction and the forming shrinkage rate in the vertical direction under the forming condition.

[0111] According to the design support device 20, in the die design, it is possible to predict the molding shrinkage rate of the molded product to be molded by the die to be designed.

[0112] (3) The design support device 20 is connected to the server 50 so as to be accessible via the network, and the server 50 has the learned model 250 .

[0113] In this way, even if a new CAD apparatus 10 and a design support apparatus 20 are installed, the learned model 250 can be applied.

[0114] (4) A machine learning device 30 may be provided.

[0115] In this way, the same effects as any of the above (1) to (3) can be obtained.

Claims

1. A machine learning device, characterized in that: have: an input data acquisition unit that acquires input data including arbitrary molding conditions, the molding conditions including at least the type of resin, the type of additive, the mixing ratio of the additive, the surface temperature of the mold, and the product of the holding pressure and the holding pressure time in molding an arbitrary molded product by an arbitrary injection molding machine; a label acquisition unit that acquires label data indicating a molding shrinkage rate in a resin flow direction and a molding shrinkage rate in a direction perpendicular to the flow direction measured in the molded product molded under the molding conditions included in the input data; as well as A learning unit performs supervised learning using the input data acquired by the input data acquisition unit and the label data acquired by the label acquisition unit to generate a learned model.

2. A design assistance device, characterized in that: have: A learning completion model generated by the machine learning device of claim 1; an input unit that inputs a preset forming condition; and The prediction unit inputs the molding conditions input by the input unit into the learned model, and predicts the molding shrinkage rate in the flow direction and the molding shrinkage rate in the perpendicular direction among the molding conditions.

3. The design assistance device according to claim 2, characterized in that: The design support device is connected to a server so as to be accessible via a network, and the server has the learned model.

4. The design assistance device according to claim 2 or 3, characterized in that: The design support device has the machine learning device according to claim 1.

Images