Information processing device, molding machine, molding condition generation method, and computer program

Abstract

This invention provides an information processing device that enables the smooth generation of molding conditions through interaction with an operator. [Solution] An information processing device comprising a processing unit that generates molding conditions for a molding machine based on molding information relating to a molded product or mold, wherein the processing unit displays the generated molding conditions on a display device in a manner that allows for modification.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus, a molding machine, a molding condition generation method, and a computer program.

Background Art

[0002] There is a molding condition adjustment apparatus capable of performing reinforcement learning on a method for adjusting molding conditions set in a molding machine and adjusting the molding conditions (for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in reinforcement learning, there is a technical problem that various problems occur in the molding machine in the process of searching for optimal molding conditions, or a state is reached where there is no prospect of convergence even after repeated trial molding. Note that this technical problem is not limited to reinforcement learning, and may occur when generating molding conditions based on molding information related to a molded product or a mold.

[0005] An object of the present disclosure is to provide an information processing apparatus, a molding machine, a molding condition generation method, and a computer program that enable smooth generation of molding conditions through interaction with an operator.

Means for Solving the Problems

[0006] The information processing apparatus according to the present disclosure is an information processing apparatus including a processing unit that generates molding conditions for a molding machine based on molding information related to a molded product or a mold, wherein the processing unit displays the generated molding conditions on a display device in a modifiable manner.

[0007] The molding machine according to this disclosure comprises a processing unit that generates molding conditions to be set in the device based on molding information relating to a molded product or mold, a display device that displays the generated molding conditions in a manner that allows for modification, and an operating device for receiving modifications to the displayed molding conditions.

[0008] The molding condition generation method relating to this disclosure is a molding condition generation method that generates molding conditions for a molding machine based on molding information relating to a molded product or mold, and displays the generated molding conditions, accepts modifications to the displayed molding conditions, and regenerates the molding conditions for the molding machine based on the accepted modifications and the molding conditions.

[0009] The computer program relating to this disclosure generates molding conditions for a molding machine based on molding information relating to a molded product or mold, obtains the modified content when the generated molding conditions are modified, and causes the computer to execute a process to regenerate the molding conditions for the molding machine based on the obtained modified content and the molding conditions.

[0010] The computer program relating to this disclosure generates molding conditions for a molding machine by inputting prompts containing molding information relating to a molded product or mold into a language model, and when the generated molding conditions are modified, it obtains the modified content and causes the computer to execute a process to regenerate the molding conditions for the molding machine by re-inputting the obtained modified content and prompts containing the molding conditions into the language model. [Effects of the Invention]

[0011] According to this disclosure, molding conditions can be smoothly generated through dialogue with the operator. [Brief explanation of the drawing]

[0012] [Figure 1] This is a schematic diagram illustrating an example of the configuration of an injection molding machine according to this embodiment 1. [Figure 2] This is a block diagram showing an example configuration of the information processing device according to this embodiment 1. [Figure 3]This flowchart shows the procedure for generating molding conditions according to this embodiment 1. [Figure 4] This flowchart shows the procedure for generating molding conditions according to this embodiment 1. [Figure 5] This is a schematic diagram showing an example of a molding information input screen. [Figure 6] This is a schematic diagram showing an example of a molding condition display screen. [Figure 7] This is a schematic diagram illustrating an example of how to display the reasons for a recommendation. [Figure 8] This is a schematic diagram showing an example of a rejection of amendment. [Figure 9] This is a schematic diagram showing the flow of the process for adjusting molding conditions. [Figure 10] This is a block diagram showing an example configuration of the information processing device according to this second embodiment. [Figure 11] This is a functional block diagram of the information processing device according to this second embodiment. [Modes for carrying out the invention]

[0013] Specific examples of information processing apparatus, molding machine, molding condition generation method, and computer program (program product) according to embodiments of the present invention will be described below with reference to the drawings. At least some of the embodiments described below may be arbitrarily combined. However, the present invention is not limited to these examples and is indicated by the claims, and all modifications within the meaning and scope equivalent to the claims are intended.

[0014] (Embodiment 1) Figure 1 is a schematic diagram illustrating an example of the configuration of an injection molding machine 1 according to this embodiment 1. The injection molding machine 1 according to this embodiment 1 comprises a mold clamping device 2 for clamping a mold 21, an injection device 3 for melting and injecting resin, which is the molding material, a control device 4, an information processing device 5, a plurality of sensors 6, and a molded product inspection device 7. In this embodiment 1, an example in which the molding material is resin is described, but the technology according to this disclosure can also be applied to an injection molding machine 1 using a molding material other than resin, such as magnesium.

[0015] The mold clamping device 2 includes a fixed platen 22 fixed on a bed 20, a mold clamping housing 23 slidably disposed on the bed 20, and a movable platen 24 that similarly slides on the bed 20. The fixed platen 22 and the mold clamping housing 23 are connected by a plurality of, for example, 4 tie bars 25, 25, .... The movable platen 24 is configured to be slidable between the fixed platen 22 and the mold clamping housing 23. A mold clamping mechanism 26 is provided between the mold clamping housing 23 and the movable platen 24. The mold clamping mechanism 26 is constituted by, for example, a toggle mechanism. The mold clamping mechanism 26 may be configured as a direct pressure type or other hydraulic mold clamping mechanism, that is, a mold clamping cylinder. A fixed mold 27 and a movable mold 28 can be attached to the fixed platen 22 and the movable platen 24 respectively, and driving the mold clamping mechanism 26 causes the mold 21 to open and close.

[0016] The plurality of sensors 6 are elements and circuits that detect physical quantities related to the state of the injection molding machine 1 or the quality of molded articles. The sensors 6 include those provided in the injection molding machine 1 as necessary for operation control of the injection molding machine 1, and those separately provided for estimating the state of the injection molding machine 1. The plurality of sensors 6 are connected to the control device 4 and the information processing device 5, and the sensors 6 output signals indicating the detected physical quantities to the control device 4 and the information processing device 5. The information processing device 5 acquires the signals output from the sensors 6 as physical quantity data. The physical quantity data is data of sensor values indicating the detected physical quantities. A part of the plurality of sensors 6 may be connected to the control device 4, and the information processing device 5 may be configured to acquire the physical quantity data from the sensors 6 via the control device 4.

[0017] The physical quantities include temperature, position, speed, acceleration, current, voltage, pressure, time, image data, torque, force, strain, power consumption, weight, and the like. These physical quantities can be measured using a thermometer, position sensor, speed sensor, acceleration sensor, ammeter, voltmeter, pressure gauge, timer, camera, torque sensor, power meter, weight scale, and the like.

[0018] The multiple sensors 6 include, for example, a speed sensor for detecting the speed of the screw 32, a speed sensor for detecting the speed of the movable platen 24, and a speed sensor for detecting the speed of the ejector pins. The multiple sensors 6 include, for example, a torque sensor for detecting the torque of the motor that drives the screw 32, a torque sensor for detecting the torque of the motor that drives the movable platen 24, and a torque sensor for detecting the torque of the motor that drives the ejector pins. The multiple sensors 6 include, for example, a pressure sensor that detects the pressure in the screw 32 or heating cylinder 31, and a pressure sensor that detects the pressure applied to the mold 21. Multiple sensors 6 include load cells that detect pressure applied longitudinally to the screw 32. The multiple sensors 6 include, for example, a position sensor for detecting the position of the screw 32, a position sensor for detecting the position of the movable platen 24, and a position sensor for detecting the position of the ejector pin. The multiple sensors 6 include, for example, a rotational speed sensor for detecting the rotational speed of the motor that drives the screw 32, a rotational speed sensor for detecting the rotational speed of the motor that drives the movable platen 24, and a rotational speed sensor for detecting the rotational speed of the motor that drives the ejector pins. It is common for the position sensor and rotational speed sensor to be combined by a rotary encoder attached to the servo motor. Multiple sensors 6 include load sensors that detect the clamping force by the clamping device 2. The multiple sensors 6 include a temperature sensor for detecting the temperature of the hopper 34, a temperature sensor for detecting the temperature at one or more locations in the heating cylinder 31, and a temperature sensor for detecting the temperature at one or more locations in the nozzle 31a.

[0019] Furthermore, sensor 6 includes any other detectors capable of detecting physical quantities that contribute to estimating the quality of the molded product.

[0020] The molded product inspection device 7 is, for example, an imaging device that images a molded product. The imaging device images the molded product and transmits the image data showing the appearance of the molded product as inspection data. The imaging device is a still camera or a video camera. The imaging device may also be a smartphone, tablet terminal, or laptop PC (Personal Computer) having a light-receiving lens and an image sensor. Furthermore, the molded product inspection device 7 may also be a laser displacement sensor for measuring the amount of deformation of the molded product, an optical measuring instrument for measuring the color of the molded product, a weighing scale for measuring the weight of the molded product, or a strength measuring instrument for measuring the strength of the molded product.

[0021] Returning to Figure 1, the injection device 3 will now be described. The injection device 3 is mounted on a base 30. The injection device 3 comprises a heating cylinder 31 having a nozzle 31a at its tip, and a screw 32 rotatably arranged within the heating cylinder 31 in both the circumferential and axial directions. The screw 32 is driven in both the rotational and axial directions by a drive mechanism 33. The drive mechanism 33 consists of a metering motor that drives the screw 32 in the rotational direction and an injection motor that drives the screw 32 in the axial direction. Note that the internal structure of the drive mechanism 33 shown in Figure 1 is not shown because it is covered by a cover.

[0022] A hopper 34 into which the molding material is fed is provided near the rear end of the heating cylinder 31. The injection molding machine 1 also includes a nozzle touch device 35 that moves the injection device 3 in the front-rear direction (left-right direction in Figure 1). When the nozzle touch device 35 is driven, the injection device 3 moves forward and the nozzle 31a of the heating cylinder 31 touches the fixed mold 27.

[0023] The control device 4 sets parameters that define molding conditions such as injection start time, injection speed, injection acceleration, injection pressure, injection stroke, VP switching position, holding pressure time, holding pressure, cylinder temperature, nozzle temperature, hopper temperature, resin temperature inside the mold, clamping force, mold opening / closing speed, mold opening position, mold cooling time, and mold temperature, and controls the operation of the injection device 3 and the clamping device 2. The various parameters that define the operating conditions of the injection molding machine 1 are called operating data.

[0024] Figure 2 is a block diagram showing an example configuration of the information processing device 5 according to this embodiment 1. The information processing device 5 is a computer and its hardware configuration includes a processing unit 51, a storage unit 52, a display unit 53, an operation unit 54, a first communication unit 55, and a second communication unit 56. The information processing device 5 may also be a server device connected to a network. Furthermore, the information processing device 5 may be configured with multiple computers for distributed processing, or it may be implemented by multiple virtual machines provided within a single server, or it may be implemented using a cloud server.

[0025] The processing unit 51 is a processor that includes arithmetic circuits such as a CPU (Central Processing Unit), multi-core CPU, GPU (Graphics Processing Unit), GPGPU (General-purpose computing on graphics processing units), TPU (Tensor Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), and NPU (Neural Processing Unit), as well as internal storage devices such as ROM (Read Only Memory) and RAM (Random Access Memory), and I / O terminals. The processing unit 51 functions as an information processing device 5 by executing a computer program (program product) P1 stored in the storage unit 52, which will be described later. Each functional part of the information processing device 5 may be implemented in software, or some or all of them may be implemented in hardware.

[0026] The storage unit 52 is a non-volatile memory such as a hard disk, EEPROM (Electrically Erasable Programmable ROM), or flash memory. The storage unit 52 stores a computer program P1 that causes the computer to execute the process of adjusting the molding conditions in injection molding.

[0027] Furthermore, the storage unit 52 includes a database 52a that stores data related to the molding conditions of the injection molding machine 1 (hereinafter referred to as basic data). For example, the database 52a stores information regarding the specifications and characteristics of the injection molding machine 1. The database 52a also stores data that associates standard molding information with molding conditions. In addition, the database 52a may include data that associates molding information, which is past molding results, with molding conditions and molded product inspection results. The information stored in the database 52a is data for RAG (Retrieval-Augmented Generation). The database 52a stores information related to molding information, that is, information useful for identifying optimal molding conditions based on molding information, in a searchable format. For example, the database 52a may store basic data such as the above-mentioned various types of information encoded in vector representations and associated with that information, or it may be configured to allow simple keyword searches.

[0028] The molding information described above includes molded product information, material information, and mold information. Molded product information is input information that includes information such as the total injection weight of the material required for molding the molded product, the average wall thickness, and the total projected area. Molded product information may be 2D or 3D data representing the shape and dimensions of the molded product. Material information includes information such as the name and type of resin that is the material of the molded product. Mold information includes information such as the runner cross-sectional area, runner length, runner shape, gate shape, gate thickness, and whether a hot runner or cold runner is used for the mold 21. Mold information may be 2D or 3D data representing the shape and dimensions of the mold 21.

[0029] The computer program P1 according to this embodiment 1 may be recorded on a recording medium M1 in a manner that is computer-readable. The storage unit 52 stores the computer program P1 read from the recording medium M1 by the reading device. The recording medium M1 is a semiconductor memory such as flash memory. The recording medium M1 may also be an optical disc such as a CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, or BD (Blu-ray® Disc). Furthermore, the recording medium M1 may be a magnetic disc such as a flexible disk or hard disk, or a magneto-optical disc. In addition, the computer program P1 according to this embodiment 1 may be downloaded from an external server connected to a communication network and stored in the storage unit 52.

[0030] The display unit 53 is, for example, a liquid crystal panel, an organic EL display, electronic paper, a plasma display, etc. The display unit 53 displays various information corresponding to the image data provided by the processing unit 51.

[0031] The operation unit 54 has input devices such as a touch panel, soft keys, hard keys, keyboard, and mouse. The display unit 53 and the operation unit 54 may be a touch panel type liquid crystal screen or the like provided on the injection molding machine 1.

[0032] The first communication unit 55 is a communication circuit that sends and receives information according to a predetermined communication protocol. The first communication unit 55 is connected to the control device 4, the sensor 6, and the molded product inspection device 7 via a communication network, and the processing unit 51 sends and receives various data to and from the control device 4, the sensor 6, and the molded product inspection device 7 via the first communication unit 55.

[0033] The second communication unit 56 is connected to the language model server 50 via a communication network (not shown) and transmits and receives information related to the generation of molding conditions.

[0034] The language model server 50 is an LLM server that has a language model, for example, a Large Language Model (LLM). The information processing device 5 according to this embodiment 1 accesses the language model server 50 which provides an API (Application Programming Interface) related to the large language model, and describes an example in which the large language model provided by the LLM server is used through the API. Of course, the information processing device 5 may also be configured to have a large language model.

[0035] A large-scale language model is a machine learning model that uses a mechanism called a Transformer, for example. The large-scale language model according to this embodiment 1 is a model that, when a prompt created based on molding information input to at least the information processing device 5 is input, outputs recommended values ​​for molding conditions for molding a molded product according to the said molding conditions, the reason for the recommendation, the recommended range, and the NG range (prohibited range).

[0036] The recommendation reason is the reason for recommending that the specified value be set. The recommendation range is the numerical range recommended for setting the molding conditions, for example, the upper and lower limits of the recommended value.

[0037] Furthermore, when a prompt created based on molding information and basic data stored in database 52a is input to the large-scale language model, it can output recommended values ​​for molding conditions that take the basic data into account. Furthermore, when a prompt is input that is created based on molding information, basic data stored in database 52a, and the most recent molding results (molding information, molding conditions, modifications to molding conditions, and inspection results of the molded product), the large-scale language model can output recommended values ​​for molding conditions that take the basic data into account.

[0038] For the large-scale language model, a general-purpose LLM may be used, or a finely tuned model may be used to output appropriate molding conditions according to the molding information.

[0039] While a large-scale language model was described as an example of a language model, the type of model is not particularly limited as long as it can perform natural language processing that processes the character data related to the input molding information and outputs the required molding conditions.

[0040] Figures 3 and 4 are flowcharts showing the molding condition generation process procedure according to this embodiment 1. The processing unit 51 of the information processing device 5 acquires molding information from the operation unit 54 (step S111). As an example, the information processing device 5 acquires molding information when various molding information is input by the operator from the setting display screen corresponding to the operation unit 54 of the injection molding machine 1. In this case, the more molding information the operator inputs, the more accurate the generation of recommended values ​​will be. However, inputting a large amount of molding information places a heavy burden on the operator, and the system will ultimately not be used. Therefore, the molding information only needs to include at least the weight (or volume) of the molded product and the type of resin. Alternatively, as another example of inputting molding information, the information processing device 5 acquires molding information when three-dimensional data of the molded product and material data are input from the operation unit 54. Furthermore, molding information is acquired when CAD data of the mold 21 is input from the operation unit 54.

[0041] Figure 5 is a schematic diagram showing an example of the molding information input screen 8. The molding information input screen 8 includes a molded product information input section 81, a material information input section 82, a mold information input section 83, and a molding condition calculation button 84.

[0042] The molded product information input unit 81 includes a text input field that accepts input of the molded product information described above, such as the total injection weight of the material required for molding the molded product, the average wall thickness, and the total projected area. The operator can input the molded product information by operating the operation unit 54. The processing unit 51 acquires the molded product information via the operation unit 54 and the molded product information input unit 81.

[0043] The material information input unit 82 includes a text input field that accepts input of the material information described above, such as the name of the resin that is the material of the molded product and the type of resin. The operator can input material information by operating the operation unit 54. The processing unit 51 acquires the material information via the operation unit 54 and the molded product information input unit 81.

[0044] The mold information input unit 83 includes a text input field that accepts the above-mentioned mold information, such as the gate shape and gate thickness of the mold 21. The operator can input mold information by operating the operation unit 54. The processing unit 51 acquires the mold information via the operation unit 54 and the molded product information input unit 81.

[0045] The molding condition calculation button 84 is used to confirm the entered molding information and to perform the calculation of molding conditions. By operating the molding condition calculation button 84, the operator can instruct the calculation of molding conditions.

[0046] The above-mentioned molded product information, material information, and mold information are examples only and are not limited thereto. Furthermore, the information processing device 5 may be configured to acquire all or part of the molding information from the control device 4 of the injection molding machine 1 if this information is already set in the control device 4. Also, at least part of the molded product information, mold information, and material information may be obtained by scanning a barcode or QR code (registered trademark) attached to the mold 21, or by communication from a storage medium provided in relation to the mold 21.

[0047] Next, the processing unit 51 generates a prompt based on molding information, basic data stored in the database 52a, etc., and inputs the generated prompt into a large-scale language model to generate recommended values ​​for molding conditions, reasons for recommendations, recommended ranges, and NG ranges (step S112). Then, the processing unit 51 displays the generated recommended values ​​for molding conditions in a modifiable manner (step S113). In this embodiment 1, the recommended ranges and NG ranges are not displayed on the screen from the beginning. However, this does not mean that all items that are displayed on the screen from the beginning are excluded.

[0048] Figure 6 is a schematic diagram showing an example of the molding condition display screen 9. The molding condition display screen 9 is the screen displayed when the molding condition calculation button 84 shown in Figure 5 is operated, and includes a recommended value display unit 91. The processing unit 51 displays multiple recommended values ​​that constitute the molding conditions in the recommended value display unit 91. The recommended value display unit 91 is a text input field that allows text input or editing. The recommended value display unit 91 includes text input fields that allow editing of recommended values ​​such as heating cylinder temperature, injection speed, and injection pressure. The operator can change the values ​​of the recommended values ​​by operating the recommended value display unit 91 using the operation unit 54. The molding condition display screen 9 also includes a setting button 92 for confirming the recommended values ​​as molding conditions. That is, the generated recommended values ​​for the molding conditions are displayed on the display device in a modifiable state.

[0049] Next, the processing unit 51 monitors the operating status of the operation unit 54 and determines whether the recommended value display unit 91 has been pressed and held down (step S114). If it is determined that the recommended value display unit 91 has been pressed and held down (step S114: YES), the processing unit 51 displays the recommended range related to the recommended value displayed on the pressed and held recommended value display unit 91 (step S115). This recommended range is generated using a large-scale language model.

[0050] Figure 7 is a schematic diagram showing an example of displaying the reason for the recommendation. The processing unit 51, for example, overlays a speech bubble image 93 on the molding condition display screen 9 and displays the recommended range in the speech bubble image.

[0051] Next, the processing unit 51 monitors the operation status of the operation unit 54 and determines whether or not the recommended value display unit 91 has been tapped (step S116). If it is determined that the recommended value display unit 91 has been tapped (step S116: YES), the processing unit 51 displays the reason why the recommended value displayed on the long-pressed recommended value display unit 91 is recommended (step S117). The reason for the recommendation is generated using a large-scale language model.

[0052] The processing unit 51, for example, overlays a speech bubble image on the molding condition display screen 9 and displays the reason for the recommendation on the speech bubble image. While the display of the recommended range (step S115) and the display of the reason for the recommendation (step S117) are not essential functions in this invention, they are useful functions that serve as indicators for operators with less experience when modifying molding conditions.

[0053] Next, the processing unit 51 monitors the operating status of the operation unit 54, accepts a request to modify the recommended value (step S118), and determines whether or not a modification has been made (step S119). If the recommended value has been modified by inputting into the molding condition display screen 9 by the operator (step S119: YES), the processing unit 51 determines whether or not the modified molding condition value is within the NG range (step S120). If it determines that the modified molding condition value is within the NG range (step S120: YES), the processing unit 51 rejects the modification of the recommended value, which is the molding condition (step S121), and returns the process to step S113. Figure 8 is a schematic diagram showing an example of a display indicating a rejection of a modification. The processing unit 51, for example, overlays a speech bubble image 94 on the molding condition display screen 9 and displays a recommended range in the speech bubble image indicating that the modification of the recommended value is rejected. For example, it displays "Please modify the heating cylinder temperature."

[0054] If the processing unit 51 determines that the values ​​of the molding conditions modified by the operator are not within the NG range (step S120: NO), it performs positive feedback learning by inputting a prompt into the language model that describes that the recommended values ​​of the generated molding conditions were appropriate (step S122). A series of dialogue contents, including prompts based on molding information and basic data, molding information as response statements, and prompts based on the modifications made by the operator, are stored as learned content. The processing unit 51 of the information processing device 5 may store the series of dialogue contents in the language model of the language model server 50, or it may be configured to output the series of dialogue contents to be stored in the storage unit 52. The processing unit 51 may also have the language model output a prompt that expresses substantially the same content as the series of dialogue contents, and store the prompt in the storage unit 52. The processing unit 51 can reproduce the learned state by inputting the prompt into the language model.

[0055] Next, the processing unit 51 generates prompts based on molding information, modifications to recommended values, basic information, etc., and inputs the generated prompts into a large-scale language model to generate recommended values ​​for molding conditions, reasons for recommendations, recommended ranges, and NG ranges (step S123). The generated recommended values ​​for molding conditions are then displayed in a modifiable manner. Furthermore, the processing unit 51 may be configured to generate prompts based on molding information, modifications to recommended values, basic information, and molded product inspection results when actual molding is performed during the molding condition adjustment process, and to input the generated prompts into a large-scale language model to generate recommended values ​​for molding conditions, reasons for recommendations, recommended ranges, and NG ranges. Furthermore, the processing unit 51 is configured to generate prompts based on molding information, modifications to recommended values, basic information, molded product inspection results, and physical quantities obtained from the sensor 6, and to generate recommended values ​​for molding conditions, reasons for recommendations, recommended ranges, and NG ranges by inputting the generated prompts into a large-scale language model.

[0056] If it is determined in step S119 that no correction is needed (step S119: NO), the processing unit 51 determines whether or not the setting button 92 on the molding condition display screen 9 has been operated (step S124). If it is determined that the setting button 92 has not been operated (step S124: NO), the processing unit 51 returns to step S113 and continues the process of checking and correcting the recommended values. In other words, step S119 is repeatable.

[0057] If the setting button 92 is operated (step S124: YES), the processing unit 51 performs positive feedback learning by inputting a prompt to the language model that explains that the recommended values ​​for the generated molding conditions were appropriate (step S125). A series of dialogue contents, including prompts based on molding information and basic data, molding information as response statements, and prompts that explain that the recommended values ​​for the generated molding conditions were appropriate, are retained as learned content. The method for retaining learned content is the same as in step S122. Note that the system may be configured to execute only one of steps S122 or S125.

[0058] Next, the processing unit 51 transmits the confirmed molding conditions to the control device 4 and sets the molding conditions (step S126). The processing unit 51 determines whether the operation unit 54 or the control device 4 has pressed the manual molding start button or the semi-automatic molding start button to initiate the molding operation (step S127). If the molding start operation has been initiated (step S127: YES), the control device 4 performs control related to the trial molding of the molded product (step S128). In the case of the above (step S119: NO), the molding start button may be pressed to initiate the molding operation (step S127). In that case, learning (positive feedback) (step S125), setting of molding conditions (step S126), and saving are performed when the molding start button is pressed.

[0059] Next, a trial molding of the molded product (step S128) is performed using the above operation. In reality, when the molding information entered by the operator is the weight of the molded product and the type of resin material, it is almost impossible to generate accurate molding conditions from the beginning. Also, if the resin material overfills the cavity of the mold 21, causing overpacking, it can lead to damage to the mold 21 and the injection molding machine 1. Therefore, in the trial molding of the molded product (step S128), the processing unit 51 of the information processing device 5 generates and displays short-shot molding conditions, in which the injection amount is injected less than the molding conditions that are considered perfect, in a modifiable manner. Then, based on the molding conditions, short-shot molding is started. After that, the setting is changed by the operator to add the injection amount with each molding round, or the processing unit 51 generates molding conditions with the injection amount added, and this is repeated until the molding conditions are such that the cavity is completely filled with resin material. According to the flowchart in Figure 3, after (step S119=YES) is repeated, (step S119=NO) is reached and the molding conditions are completed. Therefore, when molding using short shots, steps S122 and S123 may be omitted.

[0060] Thus, generating molding conditions using the information processing device 5 of this embodiment 1 minimizes the effort required for the operator to input molding information on the first attempt, and is an effective means of preventing major failures even if the operator makes a mistake or has a low level of knowledge. However, if high-precision or molded product inspection equipment 7 is not introduced and the operator inspects the molded product, the final molding conditions will often be generated by the operator.

[0061] Next, the processing unit 51 of the information processing device 5 performs a molded product inspection of the molded product generated using the molded product inspection device 7 (step S129) and determines whether or not there are any defective elements in the molded product (step S130). If it is determined that there are no defective elements (step S130: NO), the processing unit 51 ends the process. If it is determined that there are defective elements (step S130: YES), the processing unit 51 performs negative feedback learning by inputting a prompt into the language model that describes an explanation that the corrected value or content of the molding conditions was inappropriate (step S131), and returns the process to step S113.

[0062] Furthermore, even when continuous molding is performed using the molding conditions set above, the defect rate of molded products may increase due to factors such as aging deterioration of the mold 21 and injection molding machine 1, changes in resin material, and changes in weather conditions. If the defect rate increases, the information processing device 5 may take in molding information such as aging deterioration of the mold 21 and injection molding machine 1, changes in resin material, and changes in weather conditions, either at the operator's discretion or because it has exceeded a preset defect rate, and regenerate the molding conditions. In this case as well, the operator has the final decision authority to generate the molding conditions through dialogue with the operator following the flowcharts in Figures 3 and 4.

[0063] As shown in Figure 9, the injection molding machine 1 according to this embodiment 1 can smoothly generate molding conditions through dialogue with the operator. For example, when molding information is entered via the molding information input screen 8 and the molding condition calculation button 84 is operated, molding conditions as shown in the upper left of Figure 9 are proposed as recommended values. The operator can arbitrarily modify the presented molding conditions as shown in the upper right figure. In the example in the upper right figure, the heating cylinder temperature is modified from 270.0 to 275.0. When molding conditions are modified, the molding conditions are not finalized as is. Instead, the processing unit 51 uses a language model to calculate and display molding conditions that take into account the operator's modifications. For example, in response to the operator's modifications, the processing unit 51 modifies the injection speed from 17.0 to 16.5, and the re-modified molding conditions are displayed as shown in the lower left figure. The operator can then further modify the molding conditions as shown in the lower right figure. The processing unit 51 can recalculate the molding conditions using the language model, but if there are no further modifications to the molding conditions, the system will confirm the molding conditions made by the operator as shown in the lower right figure. Thus, according to this embodiment 1, molding conditions can be smoothly generated through dialogue with the operator.

[0064] In this embodiment 1, an example was described in which the information processing device 5 is a separate device from the control device 4, but the information processing device 5 may also be integrated into the control device 4. Alternatively, the information processing device 5 may be configured as a cloud server.

[0065] (Embodiment 2) The information processing device 5, injection molding machine 1, molding condition generation method, and computer program according to Embodiment 2 differ from those of Embodiment 1 in that they adjust the molding conditions of the injection molding machine 1 using a reinforcement learning system that employs a deep neural network (DNN) of a non-large-scale language model. The other components of the information processing device 5 and other components are the same as those of Embodiment 1, so the same reference numerals are used for the same parts, and detailed explanations are omitted.

[0066] Figure 10 is a block diagram showing an example configuration of the information processing device 5 according to Embodiment 2. The information processing device 5 according to Embodiment 2 functions as a reinforcement learner having a deep neural network such as DQN (Deep Q-Network) without using a large-scale language model. Alternatively, all or part of the deep neural network of the reinforcement learner may be built on a cloud server, and the reinforcement learner may be configured by communicating with the cloud server via the second communication unit 56, similar to Embodiment 1. In other words, the reinforcement learner of the information processing device 5 may be configured using multiple computers for distributed processing, implemented using multiple virtual machines on a single server, or implemented using a cloud server. For simplicity, the following explanation will describe an example where the processing unit 51 functions as a reinforcement learner.

[0067] Figure 11 is a functional block diagram of the information processing device 5 according to this second embodiment. The processing unit 51 of the information processing device 5 includes an observation unit 51a, a reward calculation unit 51b, an agent 51c, and a molding condition adjustment unit 51d as a reinforcement learning device. Each functional unit of the information processing device 5 may be implemented in software, or some or all of them may be implemented in hardware.

[0068] The observation unit 51a acquires molding information from the control device 4 via the first communication unit 55. The observation unit 51a also acquires molded product inspection results from the molded product inspection device 7 and physical quantities indicating the status of the injection molding machine 1 from the sensor 6. The observation unit 51a may also acquire molding information via the operation unit 54. The observation unit 51a outputs the acquired molding information, molded product inspection results, and physical quantities to the agent 51c. The observation unit 51a also outputs the acquired molded product inspection results to the reward calculation unit 51b.

[0069] The reward calculation unit 51b calculates reward data obtained by evaluating the currently set molding conditions based on the data observed by the observation unit 51a, particularly the molded product inspection results, and outputs the calculated reward data to the agent 51c. For example, the reward calculation unit 51b calculates reward data that is larger when the molded product is good compared to when it is defective. When the molded product is defective, the reward calculation unit 51b calculates reward data according to the degree of defect. For example, when the molded product has burrs or sink marks, the reward calculation unit 51b calculates reward data with a larger negative value the larger the area of ​​the burr or sink mark region is.

[0070] Agent 51c is, for example, a reinforcement learning model with a deep neural network such as DQN (Deep Q-Network), A3C, or D4PG, or a model-based reinforcement learning model such as PlaNet or SLAC. Below, we will describe an example in which Agent 51c is equipped with DQN.

[0071] Agent 51c uses DQN to determine an action a corresponding to the state s of the injection molding machine 1 indicated by the observed data. The state s includes molding condition data, operating status data, and molded product state data.

[0072] DQN is a neural network model that, given a state s represented by observed data as input, outputs the value of each of several actions a. Each of the multiple actions a includes the items to be adjusted and the amount of adjustment. Actions a with high value represent the adjustment items and amounts for setting appropriate conditions. Agent 51c outputs the multiple actions a and the value of each action a to the shaping condition adjustment unit 51d. Agent 51c basically selects action a with high value. However, if another action a is selected by the operator and the molding conditions are determined (or modified), the injection molding machine 1 transitions to another state according to the selected action a. After the state transition, agent 51c receives the reward calculated by the reward calculation unit 51b and is trained to maximize the revenue, i.e., the cumulative reward.

[0073] More specifically, DQN has an input layer, an intermediate layer, and an output layer. The input layer comprises multiple nodes into which a state s, i.e., observed data, is input. The output layer comprises multiple nodes, each corresponding to one of several actions a, and outputting the value Q(s,a) of that action a in the input state s.

[0074] Based on the state s, action a, and the reward r obtained from that action, the DQN of agent 51c can be reinforced by adjusting various weight coefficients that characterize the DQN, using the value Q represented by the following equation (1) as training data for which the correct answer is. Q(s,a)←Q(s,a)+α(r+γmaxQ(snext,anext)-Q(s,a))...(1) however, s: state a:Action α: Learning rate r:Reward γ: discount rate maxQ(snext,anext): The maximum Q-value for the next possible action.

[0075] The reward can be expressed, for example, by the following formula (2). Reward r = Σwi × ri + cycle time reward - Σ constraint violation penalty ... (2)

[0076] However, wi is the weight coefficient and ri is the reward element (i=1,2,3...). The reward element is a function that becomes smaller as the area of ​​burrs and sink marks increases, for example, as shown in the following equation. k1 and k2 are constants. r1 = -k1 × burr area r² = -k² × sink area

[0077] The cycle time reward is a function that increases as the cycle time decreases. The constraint violation panel ty is a function that becomes negative if the set value exceeds the upper limit or falls below the lower limit of the possible values ​​for that set value.

[0078] The molding condition adjustment unit 51d is a functional unit that adjusts molding conditions through interaction with the operator, similar to Embodiment 1. Based on the value Q of each of the multiple actions a calculated by agent 51c, the molding condition adjustment unit 51d identifies the optimal molding conditions, the recommended range of molding conditions, and the NG range of molding conditions.

[0079] Specifically, the processing unit 51, acting as a molding condition adjustment unit 51d, refers to the value Q of each of several possible setting values ​​for the molding conditions and identifies a recommended range based on the setting value, i.e., an action a that is equal to or greater than a first predetermined percentage (e.g., 80% or more) of the maximum value of value Q. Similarly, the processing unit 51 refers to the value Q of each of several possible setting values ​​for the molding conditions and identifies an NG range based on the setting value, i.e., an action a that is less than a second predetermined percentage (e.g., 20% or more) of the maximum value of value Q.

[0080] The first predetermined ratio and the second predetermined ratio may be fixed values ​​or may be configured to be changed as appropriate. Since they may change depending on the operating environment of the injection molding machine 1, it is preferable that the processing unit 51 dynamically changes the first predetermined ratio and the second predetermined ratio. For example, the storage unit 52 stores the recommended range and NG range of standard molding conditions, and the processing unit 51 may calculate the value of the molding conditions (action a) in the standard recommended range using agent 51c, and determine the first predetermined ratio based on the calculated value Q. Similarly, the processing unit 51 may calculate the value of the molding conditions (action a) in the standard NG range using agent 51c, and determine the second predetermined ratio based on the calculated value Q.

[0081] Furthermore, if the molding conditions are modified by the operator, the molding condition adjustment unit 51d executes a process to recalculate the molding conditions based on the modification of the molding conditions received from the operator and the value Q of action a recalculated by agent 51c. For example, the setting value modified by the operator may be set as a fixed value, and the processing unit 51 may be configured to calculate the optimal molding conditions, recommended range, and NG range by having agent 51c calculate the value for other combinations of setting values ​​for the molding conditions. The processing unit 51 displays the recalculated molding conditions on the display unit 53.

[0082] Furthermore, the processing unit 51, acting as a reinforcement learner, is configured to perform reinforcement learning on the agent 51c's DQN (by changing the coefficients of the DQN through processes such as backpropagation) based on the modifications to the molding conditions received from the operator at appropriate timings, and to calculate the optimal molding conditions, recommended range, and NG range using the reinforcement-learned agent 51c. The processing unit 51 displays the recalculated molding conditions on the display unit 53.

[0083] With the injection molding machine 1 according to this embodiment 2 configured in this way, molding conditions can be smoothly generated through dialogue with the operator, similar to embodiment 1.

[0084] The business model using the present invention may involve selling the molding machine and information processing device to a molding manufacturer where the molding machine is installed and molding is performed, or it may be in the form of a subscription where at least one of the molding machine and the information processing device is leased and a usage fee is paid at predetermined intervals. In the case of a subscription business model where at least a part of the information processing device is installed by a party other than the molding manufacturer, it becomes possible to collect information from multiple molding machines, and it often becomes easier to update the learning model of the information processing device, including reinforcement learning, and to provide the molding manufacturer with the information processing device that incorporates the latest update information.

[0085] It should be noted that the present invention is not limited to those described above, and is also applicable to modifications made by those skilled in the art based on the spirit of the invention, although these will not be listed individually. The multiple examples described above can also be combined as appropriate. [Explanation of Symbols]

[0086] 1: Injection molding machine 2: Mold clamping device 3: Injection device 4: Control device 5: Information Processing Device 6: Sensor 7: Molded product inspection device 20: Bed 21: Mold 22:Fixed plate 23: Clamping Housing 24: Movable plate 25: Tie bar 26: Mold clamping mechanism 27: Fixed mold 28: Movable mold 30: Base 31: Heating cylinder 31a: Nozzle 32: Screw 33: Drive mechanism 34: Hoppa 35: Nozzle touch device 50: Language Model Server 51: Processing Unit 52: Storage section 52a: Database 53:Display section 54:Operation unit 55: First Communications Department 56: Second Communications Department M1: Recording medium P1: Computer program

Claims

1. An information processing device comprising a processing unit that generates molding conditions for a molding machine based on molding information relating to a molded product or mold, The aforementioned processing unit, The generated molding conditions are displayed on the display device in a manner that allows for modification. Information processing device.

2. The aforementioned processing unit, We accept modifications to the displayed molding conditions. Based on the received modifications and the molding information, the molding conditions for the molding machine are regenerated. The information processing apparatus according to claim 1.

3. The aforementioned processing unit, The regenerated molding conditions are displayed on the display device. The molding conditions can be repeatedly modified and regenerated. The information processing apparatus according to claim 2.

4. The aforementioned processing unit, Based on molding information relating to the molded product or mold, the prohibited range of molding conditions is identified. The modification of the molding conditions that fall within the specified prohibited range will be refused. The information processing apparatus according to claim 2 or claim 3.

5. The aforementioned processing unit, Based on molding information relating to the molded product or mold, the recommended range of molding conditions is identified. Display the specified recommended range. The information processing apparatus according to any one of claims 1 to 3.

6. The aforementioned processing unit, The molding conditions for the molding machine are generated by inputting prompts containing molding information related to the molded product or mold into the language model. The information processing apparatus according to any one of claims 1 to 3.

7. A processing unit that generates molding conditions to be set in the device based on molding information relating to the molded product or mold, A display device that displays the generated molding conditions in a manner that allows for modification, An operating device for accepting modifications to the displayed molding conditions and A molding machine equipped with the following features.

8. A method for generating molding conditions for a molding machine based on molding information relating to a molded product or mold, Display the generated molding conditions. We accept modifications to the displayed molding conditions. Based on the received modifications and the molding conditions, the molding conditions of the molding machine are regenerated. How to generate molding conditions.

9. Based on molding information relating to the molded product or mold, the molding conditions for the molding machine are generated. When the generated molding conditions are modified, the modification details are retrieved. Based on the acquired modifications and the molding conditions, the molding conditions of the molding machine are regenerated. A computer program that causes a computer to perform a process.

10. By inputting prompts containing molding information related to the molded product or mold into the language model, the molding conditions for the molding machine are generated. When the generated molding conditions are modified, the modification details are retrieved. The molding conditions of the molding machine are regenerated by re-inputting the acquired corrections and the prompts including the molding conditions into the language model. A computer program that causes a computer to perform a process.

Images