Method for modifying operating conditions, device for modifying operating conditions, molding machine, and computer program

The method and device address the issue of counter-proposals in learning devices by employing multiple algorithms to determine and apply appropriate correction amounts, enhancing the accuracy of industrial machine operations and reducing defects.

JP2024157273APending Publication Date: 2024-11-07THE JAPAN STEEL WORKS LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023071532
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-04-25
Publication Date
2024-11-07

AI Technical Summary

Technical Problem

Existing learning devices in industrial machines, such as injection molding machines, often make inappropriate corrections to operating conditions, leading to worsening defects in the molded products due to counter-proposals, which are incorrect suggestions for modifications.

Method used

An operating condition correction method and device that utilize multiple learning devices to calculate correction amounts for industrial machines, determine the appropriateness of these corrections, and exclude counter-proposals by selecting correction amounts in the correct direction using methods like majority voting, direction definition tables, or suitability determination models.

Benefits of technology

This approach effectively excludes counter-proposals and corrects operating conditions more accurately, optimizing the molding process by using unbiased correction amounts derived from multiple algorithms, thereby reducing defects in the molded products.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024157273000001_ABST
    Figure 2024157273000001_ABST
Patent Text Reader

Abstract

To provide a method for modifying operating conditions that can exclude counter-suggestions by the learner and modify operating conditions more appropriately.SOLUTION: A method for modifying operating conditions of an industrial machine, comprising the steps of: acquiring measurement data obtained by measuring a condition of the industrial machine and inspection result data obtained by inspecting a condition of a product manufactured by the industrial machine; calculating modification amounts of the operating conditions based on the acquired measurement data and inspection result data by using a plurality of learning devices that have learnt a relation between the modification amounts of the operating conditions, and the acquired measurement data and inspection result data; judging the appropriateness of the modification direction of each of the plurality of modification amounts calculated by using the plurality of learning devices; and modifying the operating conditions based on one or more modification amounts for which the modification direction is judged to be appropriate.SELECTED DRAWING: Figure 5
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an operating condition correction method, an operating condition correction device, a molding machine, and a computer program. [Background technology]

[0002] There is a molding machine system that uses a machine learning machine to correct the molding conditions of an injection molding machine so that molded products are closer to good products (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-045698 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, because the accuracy of the molding condition corrections proposed by the learning machine is not 100%, there is a problem that the learning machine may make counter-suggestions, which may result in inappropriate corrections to the molding conditions. This problem is not limited to injection molding machines, but other industrial machines that use a learning device to correct operating conditions have similar problems.

[0005] An object of the present disclosure is to provide an operating condition correction method, an operating condition correction device, a molding machine, and a computer program that can exclude counterproposals by a learning device and more appropriately correct operating conditions. [Means for solving the problem]

[0006] An operating condition correction method according to one aspect of the present disclosure is a method for correcting operating conditions of an industrial machine, and includes the steps of acquiring measurement data obtained by measuring a state of the industrial machine and inspection result data obtained by inspecting the state of a product manufactured by the industrial machine, calculating an amount of correction of the operating conditions based on the acquired measurement data and inspection result data using a plurality of learning devices that have learned the relationship between the measurement data and the inspection result data and an amount of correction of the operating conditions, a determination step of determining whether the direction of correction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate, and correcting the operating conditions based on one or more correction amounts for which the direction of correction is determined to be appropriate.

[0007] An operating condition correction device according to one aspect of the present disclosure is an operating condition correction device that corrects the operating conditions of an industrial machine, and includes an acquisition unit that acquires measurement data obtained by measuring the state of the industrial machine and inspection result data obtained by inspecting the state of a product manufactured by the industrial machine, a plurality of learning devices that learn the relationship between the measurement data and the inspection result data, and a correction amount of the operating conditions based on the acquired measurement data and the inspection result data, a judgment unit that judges whether the correction direction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate, and a correction unit that corrects the operating conditions based on one or more correction amounts for which the correction direction is judged to be appropriate.

[0008] A molding machine according to one aspect of the present disclosure includes the above-described operating condition modification device.

[0009] A computer program according to one aspect of the present disclosure is a computer program for causing a computer to execute a process of correcting operating conditions of an industrial machine, and causes the computer to execute the steps of acquiring measurement data obtained by measuring the state of the industrial machine and inspection result data obtained by inspecting the state of a product manufactured by the industrial machine, calculating an amount of correction to the operating conditions based on the acquired measurement data and inspection result data using a plurality of learning devices that have learned the relationship between the measurement data and the inspection result data, and an amount of correction to the operating conditions, a determination step of determining whether the direction of correction for each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate, and correcting the operating conditions based on one or more correction amounts for which the direction of correction is determined to be appropriate. Effect of the Invention

[0010] According to the present disclosure, it is possible to eliminate counter suggestions made by a learning device and to more appropriately correct driving conditions. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a schematic diagram illustrating an example of the configuration of an injection molding machine according to a first embodiment of the present invention. [Diagram 2] 1 is a block diagram showing an example of the configuration of an injection molding machine according to a first embodiment of the present invention. [Diagram 3] FIG. 2 is a conceptual diagram showing an operating condition modification processing method according to the first embodiment. [Figure 4] 4 is a flowchart showing an operating condition modification process procedure according to the first embodiment. [Diagram 5] FIG. 11 is a block diagram showing an example of the configuration of an injection molding machine according to a second embodiment of the present invention. [Figure 6] FIG. 11 is a conceptual diagram showing a direction definition table according to the second embodiment. [Figure 7] FIG. 11 is a conceptual diagram showing an operating condition modification processing method according to the second embodiment. [Figure 8] 7 is a flowchart showing an operating condition modification process procedure according to the second embodiment. [Figure 9] FIG. 11 is a block diagram showing an example of the configuration of an injection molding machine according to a third embodiment. [Figure 10] FIG. 11 is a conceptual diagram showing a suitability judgment model according to the third embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Specific examples of the operating condition correction method, the operating condition correction device, the molding machine, and the computer program according to the embodiments of the present invention will be described below with reference to the drawings. At least some of the embodiments described below may be combined as desired. Note that the present invention is not limited to these examples, but is indicated by the claims, and is intended to include all modifications within the meaning and scope equivalent to the claims.

[0013] (Embodiment 1) 1 is a schematic diagram showing an example of the configuration of an injection molding machine 101 according to this embodiment. The injection molding machine 101 according to this embodiment includes a mold clamping device 2 that clamps a mold 21, an injection device 3 that plasticizes and injects molding material, a control device 4, and an inspection device 5. The mold clamping device 2 and the injection device 3 constitute a molding machine main body 1. The control device 4 functions as an operating condition correction device according to this embodiment.

[0014] The mold clamping device 2 includes a fixed platen 22 fixed on a bed 20, a mold clamping housing 23 provided slidably on the bed 20, and a movable platen 24 which similarly slides on the bed 20. The fixed platen 22 and the mold clamping housing 23 are connected by a plurality of, for example, four 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.

[0015] The mold clamping mechanism 26 is composed of, for example, a toggle mechanism. The mold clamping mechanism 26 may be composed of a direct pressure type mold clamping mechanism, that is, a mold clamping cylinder. The fixed platen 22 and the movable platen 24 are provided with a fixed mold 21a and a movable mold 21b, respectively, and when the mold clamping mechanism 26 is driven, the mold 21 is opened and closed.

[0016] The injection device 3 is provided on a base 30. The injection device 3 includes a heating cylinder 31 having a nozzle 31a at its tip, and a screw 32 arranged within the heating cylinder 31 so as to be rotatable in the circumferential and axial directions. The screw 32 is driven in the rotational and axial directions by a drive mechanism 33. The drive mechanism 33 is composed of a rotary motor that drives the screw 32 in the rotational direction, and a motor that drives the screw 32 in the axial direction. Note that the drive mechanism 33 shown in FIG. 1 is covered with a cover, and therefore the internal configuration is not shown.

[0017] A hopper 34 into which molding material is poured is provided near the rear end of the heating cylinder 31. The injection molding machine 101 also includes a nozzle touch device 35 that moves the injection unit 3 in the front-rear direction (the left-right direction in FIG. 1). When the nozzle touch device 35 is driven, the injection unit 3 moves forward so that the nozzle 31a of the heating cylinder 31 touches the contact portion of the fixed platen 22.

[0018] 2 is a block diagram showing an example of the configuration of an injection molding machine 101 according to this embodiment. The control device 4 is a computer that controls the operations of the mold clamping unit 2 and the injection unit 3, and includes, as its hardware configuration, a processor 41, a storage unit 42, a control signal output unit 43, a first acquisition unit 44, a second acquisition unit 45, and an operation panel 46. The processor 41 has a plurality of learning devices 40 as its functional units. Note that some of the plurality of learning devices 40 may be realized by hardware.

[0019] The control device 4 is a device that modifies molding conditions (operating conditions) of the injection molding machine 101 (see FIG. 1). The control device 4 may be a server device connected to a network. The control device 4 may be configured to perform distributed processing by being configured with multiple computers, may be realized by multiple virtual machines provided in one server, or may be realized by using a cloud server.

[0020] The processor 41 includes a central processing unit (CPU), a multi-core CPU, a graphics processing unit (GPU), a general-purpose computing on graphics processing units (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a neural processing unit (NPU), and other arithmetic circuits, a read only memory (ROM), a random access memory (RAM), and other internal storage devices, an I / O terminal, and a timer. The processor 41 executes a computer program (program product) 42a stored in a storage unit 42 described later, thereby implementing the molding condition correction method according to the present embodiment. Each functional unit of the control device 4 may be realized by software, or a part or all of the functional units may be realized by hardware.

[0021] The storage unit 42 includes a hard disk, an EEPROM (Electrically Erasable Programmable Read Only Memory), The memory unit 42 is a non-volatile memory such as a ROM or a flash memory. The memory unit 42 stores a computer program 42a for machine learning a method of correcting molding conditions according to the state of the injection molding machine 101 and the molded product, and for causing a computer to execute an operating condition correction process. The memory unit 42 stores various coefficients that characterize each of the multiple learning devices 40.

[0022] The computer program 42a according to the present embodiment may be recorded in a computer-readable manner on a recording medium 49. The storage unit 42 stores the computer program 42a read from the recording medium 49 by a reading device. The recording medium 49 is a semiconductor memory such as a flash memory. The recording medium 49 may also be an optical disc such as a CD (Compact Disc)-ROM, a DVD (Digital Versatile Disc)-ROM, or a BD (Blu-ray (registered trademark) Disc). Furthermore, the recording medium 49 may also be a magnetic disc such as a flexible disc or a hard disk, a magneto-optical disc, or the like.

[0023] Furthermore, the computer program 42a according to this embodiment may be downloaded from an external server connected to a communication network and stored in the storage unit .

[0024] The multiple learning devices 40 are multiple models that use different algorithms to learn the amount of correction to the molding conditions set in the injection molding machine 101. Each of the multiple learning devices 40 is a model that outputs the amount of correction to the molding conditions when measurement data and inspection result data are input.

[0025] For example, the multiple learning devices 40 include a reinforcement learning model, a learning model that has been learned with a supervision, a regression model that has been learned by a gradient method, and the like. The multiple learning devices 40 may be different models that have been trained using different learning data using the same algorithm. The types of the learning devices 40 are not limited to those described above. For example, the multiple learning devices 40 may be configured using algorithms such as any neural network model, decision tree, random forest, and SVM (Support Vector Machine).

[0026] The correction amount output by the learning device 40 is generally a vector amount consisting of the change values ​​of multiple setting parameters. The correction amount for correcting molding conditions related to molding defects such as flash and shorts in molded products is expressed as, for example, (change value of holding pressure, change value of holding pressure switching position, change value of injection speed). Hereinafter, the direction of the vector of the correction amount will be referred to as the correction direction.

[0027] The correction amounts output from the multiple learning devices 40 generally include correction amounts related to countersuggestions. The correction amounts related to countersuggestions include correct correction amount changes and change values ​​with opposite positive and negative signs. As described below, the processor 41 according to this embodiment can appropriately correct the operating conditions by excluding the correction amounts related to countersuggestions, selecting correction amounts in the appropriate correction direction, and executing a process to correct the molding conditions.

[0028] The control signal output unit 43 outputs a control signal to the molding machine main body 1 for controlling the operation of the injection molding machine 101 according to the control of the processor 41 based on the molding conditions.

[0029] The operation panel 46 is an interface for setting molding conditions and the like for the injection molding machine 101 and for operating the operation of the injection molding machine 101. The operation panel 46 includes a display panel and an operation device. The display panel is a display device such as a liquid crystal display panel or an organic EL display panel, and displays a reception screen for receiving settings of molding conditions for the injection molding machine 101, the state of the injection molding machine 101, and the implementation status of the molding condition correction method according to this embodiment, under the control of the processor 41. The operation device is an input device for inputting and correcting molding conditions for the injection molding machine 101, and includes operation buttons, a touch panel, and the like. The operation device provides data indicating the received molding conditions to the processor 41.

[0030] In the injection molding machine 101, setting parameters related to various molding conditions are set. The molding conditions include the injection start position, resin temperature in the mold, nozzle temperature, cylinder temperature (heater temperature), hopper temperature, mold clamping force, injection speed, injection acceleration, injection peak pressure (injection pressure), and injection stroke. The molding conditions also include the resin pressure at the end of the cylinder, back prevention ring seating state, dwell pressure, dwell switching speed, dwell switching position, dwell completion position, cushion position, metering back pressure, and metering torque. The molding conditions also include the metering completion position, screw retreat speed, cycle time, mold closing time, injection time, dwell time, metering time, and mold opening time. The molding conditions also include the cooling time, screw rotation speed, mold opening / closing speed, ejection speed, and ejection count.

[0031] The injection molding machine 101, in which these setting parameters have been set, operates according to the setting parameters. Among the above-mentioned molding conditions, the setting parameters such as the holding pressure, the holding pressure switching position, and the injection speed are particularly related to molding defects such as flash and shorts in the molded product. In this embodiment, an example will be described in which the learning device 40 is used to correct the holding pressure [Mpa], the holding pressure switching position [mm], and the injection speed [mm / sec].

[0032] The first acquisition unit 44 is an input circuit that acquires measurement data obtained by measuring the state of the injection molding machine 101.

[0033] The mold clamping unit 2 and the injection unit 3 (see FIG. 1) are provided with one or more sensors 1a for detecting physical quantities necessary for controlling the operation of the molding machine main body 1, which are information indicating the state of the injection molding machine 101. The sensor 1a is connected to a first acquisition unit 44. The physical quantities include, for example, voltage, current, temperature, humidity, torque, pressure, and other forces, the speed of a movable part, acceleration, rotation angle, position, and the flow rate and speed of a fluid. The sensor 1a is, for example, a current sensor, a voltage sensor, a temperature sensor, a humidity sensor, a torque sensor, a pressure sensor, a speed sensor, an acceleration sensor, a rotation angle sensor, a positioning sensor, a flow sensor, a flow rate meter, and the like. The sensor 1a outputs a measurement signal indicating the physical quantity to the control device 4. The measurement signal output from the sensor 1a is input to the first acquisition unit 44, and the first acquisition unit 44 acquires the measurement signal as measurement data.

[0034] The measurement data is data showing the operating status of the injection molding machine 101, and includes, for example, cycle time [sec], injection time [sec], dwell time [sec], dwell switch position [mm], dwell switch speed [mm / sec], dwell switch pressure [MPa], cushion position [mm], dwell completion position [mm], metering time [sec], back pressure [MPa], metering completion position [mm], etc. Note that each position is positive in the retreat direction of the screw 32 (rightward in FIG. 1).

[0035] The second acquisition unit 45 is an input circuit that acquires inspection result data obtained by inspecting the state of a molded product molded by the injection molding machine 101. The second acquisition unit 45 is connected to the inspection device 5. The inspection device 5 is a camera, a distance measuring sensor, a weighing scale, etc. that detects the state of the molded product, measures physical quantities related to the state of the molded product, and obtains inspection result data indicating the state of the molded product based on the physical quantity data obtained by measurement. The inspection result data is, for example, data indicating the burr area and short area of ​​the molded product. The inspection result data is also data indicating burrs and shorts caused by resin viscosity.

[0036] Fig. 3 is a conceptual diagram showing the operating condition correction processing method according to the present embodiment 1. As shown in Fig. 3, the processor 41 includes, in addition to multiple types of learning devices 40, a counter proposal determination processing unit 41a and a molding condition setting processing unit 41b as functional units.

[0037] The measurement data acquired by the first acquisition unit 44 and the inspection result data acquired by the second acquisition unit 45 are input to each of the multiple learning devices 40. The multiple learning devices 40 represent models that operate according to different algorithms 1, 2, ... N. The multiple learning devices 40 calculate the amount of correction of the molding conditions based on the input measurement data and inspection result data.

[0038] The counterproposal determination processor 41a is a functional unit that determines whether or not the correction amounts output from the multiple learning devices 40 are correction amounts related to counterproposals. For example, the counterproposal determination processor 41a determines, by majority vote, whether or not the correction amounts output from the learning devices 40 are correction amounts related to counterproposals. The counterproposal determination processor 41a excludes correction amounts determined to be counterproposals, and selects a correction amount having a correct correction direction.

[0039] <Example of judgment method 1> The counterproposal determination processing unit 41a according to Example 1 identifies, for each of a plurality of setting parameters, the number of positive change values ​​and the number of negative change values ​​output from a plurality of learning devices 40, selects the change value with the greater number, and excludes the smaller number of change values. For example, when correction amounts are output from five learning devices 40, and there are three positive change values ​​and two negative change values ​​for correcting the first setting parameter, the counterproposal determination processor 41a selects the positive change value. When there is one positive change value and four negative change values ​​for correcting the second setting parameter, the counterproposal determination processor 41a selects the negative change value. <Example of judgment method 2> In Example 2, it is assumed that the positive and negative signs of the multiple change values ​​constituting the correction amount of the counterproposal are all opposite to those of the multiple change values ​​constituting the correct correction amount. In other words, it is assumed that the multiple correction amounts output from the multiple learning devices 40 include a correct correction amount and a correction amount whose positive and negative signs are exactly opposite to those of the correct correction amount. In this case, there are two combinations of positive and negative change values ​​constituting the correction amount. For example, the first combination of change values ​​is (positive, negative, positive), and the second combination of change values ​​is (negative, positive, negative).

[0040] The counterproposal determination processor 41a according to Example 2 calculates the number of correction amounts for the first combination and the number of correction amounts for the second combination, and determines that the combination with the larger number of correction amounts indicates the correct positive and negative change values ​​of the correction amounts. The counterproposal determination processor 41a selects the correction amounts for the majority of combinations, and excludes the correction amounts for the minority of combinations.

[0041] <Example 3 of judgment method> The counterproposal determination processor 41a according to Example 3 finds a unit vector for each of a plurality of correction amounts output from a plurality of learning devices 40. This unit vector indicates the correction direction. The counterproposal determination processor 41a clusters the unit vectors of each correction amount, and determines a class to which a large number of unit vectors belong as a class of correct correction directions. The counterproposal determination processor 41a selects correction amounts that belong to the class of correct correction directions, and excludes correction amounts that belong to other classes.

[0042] <Example 4 of judgment method> The counterproposal determination processing unit 41a according to the fourth example creates a reference unit vector for distinguishing between a correct correction amount and a correction amount of a counterproposal, and excludes the correction amount of the counterproposal. The reference unit vector may be selected, for example, so that the sum of absolute values ​​of the inner products of all correction amounts and the reference unit vector is maximized. The counterproposal determination processing unit 41a calculates the inner products of the reference unit vector and each of the multiple correction amounts, and classifies the correction amounts into correction amounts whose inner products are positive and correction amounts whose inner products are negative. The counterproposal determination processing unit 41a then compares the number of correction amounts whose inner products are positive and the number of correction amounts whose inner products are negative, determines that the direction with the greater number is the direction of the correct correction amount, and selects the correction amount in that direction as the correct correction amount. For example, the correction amount whose inner product is positive (the majority of correction amounts are positive) is selected as the correct correction amount, and the correction amount whose inner product is negative (the minority of correction amounts are negative) is excluded as the correction amount of the counterproposal. The correction amount whose inner product is zero may be excluded as the correction amount of the counterproposal, or may be selected as the correct correction amount.

[0043] The above-mentioned selection method is just one example, and any method can be used as long as it can determine the correction direction of the correction amounts output from multiple learning devices 40 and select a large number of correction amounts that are roughly oriented in the same correction direction. Incidentally, if a learning device 40 using a gradient method is provided as an optimization algorithm in which counterproposals do not or are unlikely to occur, it becomes possible to select a correct revision amount with a higher probability by the majority voting algorithm described above. Also, if a learning device 40 that has been trained using Adam, momentum, Adagrad, or the like is provided as an optimization algorithm in which counterproposals do not or are unlikely to occur, it becomes possible to select a correct revision amount with a higher probability in the same manner.

[0044] The molding condition setting processing unit 41b is a functional unit that corrects the molding conditions using the correction amount having the correct correction direction and sets the corrected molding conditions. For example, the molding condition setting processing unit 41b calculates the average of the correction amounts having the correct correction direction and corrects the molding conditions by the average correction amount.

[0045] <Specific example of average correction for "Example 1 of judgment method"> For example, when modifying the setting parameters x, y, and z, the combination of change amounts (Cx, Cy, Cz) constituting the correction amounts output from the three learning devices 40 is First correction amount = (-6, -3, 2) Second correction amount = (3,1,4) Third correction amount = (5,-1,-2) In this case, For Cx, discard the first correction amount of -6 and take the average as (3+5) / 2=4. For Cy, discard the second correction amount of 1 and set the average to (-3-1) / 2=-2. For Cz, the third correction amount of -2 is discarded and the average is (2+4) / 2=3. The setting parameters (x, y, z) of the molding conditions are corrected and set using this average correction amount.

[0046] <Specific example of average correction in "Example of judgment method 2"> For example, when correcting the setting parameters x, y, and z, the combination of change amounts (Cx, Cy, Cz) constituting the correction amounts output from the three learning devices 40 is First correction amount = (6,-3,1) Second correction amount = (-3, 1, -4) Third correction amount = (5,-1,2) In this case, the first correction amount and the third correction amount are selected as the correct correction amount by majority vote, For Cx, discard the second correction amount of -3 and set the average to (6+5) / 2=5.5. For Cy, discard the second correction of 1 and take the average as (-3-1) / 2=-2. For Cz, the second correction amount of -4 is discarded and the average is set to (1+2) / 2=1.5. The setting parameters (x, y, z) of the molding conditions are corrected and set using this average correction amount.

[0047] The molded products produced by the injection molding machine 101 for which the molding conditions have been modified are inspected by the inspection device 5, and the molding conditions are repeatedly modified until the defects in the molded products are eliminated.

[0048] 4 is a flowchart showing the procedure of the operation condition correction process according to the embodiment 1. The processor 41 of the control device 4 controls the operation of the molding machine body 1, and the molding machine body 1 injection molds a molded product (step S111). Next, the processor 41 acquires the measurement data and the inspection result data by the first acquisition unit 44 and the second acquisition unit 45 (step S112).

[0049] Next, the processor 41 inputs the measurement data and the inspection result data acquired in step S112 to each of the multiple learning devices 40, and calculates multiple correction amounts (step S113). The multiple correction amounts are calculated using each of the multiple learning devices 40.

[0050] Next, the processor 41 judges whether the inspection result is good or not, that is, whether the molded product is a good product or not, based on the inspection result data (step S114). For example, the processor 41 judges whether the molded product is good or not by judging whether the flash area and short area of ​​the molded product are less than a predetermined threshold value.

[0051] When it is determined that the inspection result is good (step S114: YES), the processor 41 returns the process to step S111. When it is determined that the inspection result is not good (step S114: NO), it is determined by majority vote for the correction direction whether the correction amount output from each learning device 40 is a correction amount related to a counterproposal, and the correction amount of the counterproposal is excluded (step S115).

[0052] Next, the processor 41 calculates an average correction amount using the correct correction amount excluding the counter-proposed correction amount (step S116), sets the molding conditions corrected using the average correction amount (step S117), and returns the processing to step S111. By repeating steps S111 to S117, the molding conditions are corrected in a direction that eliminates defects in the molded product.

[0053] According to the injection molding machine 101 and the control device 4 (operating condition correction device) of the present embodiment 1 configured as described above, it is possible to exclude counter suggestions by the learning device 40 and correct the molding conditions more appropriately. Specifically, multiple learning devices 40, i.e., multiple algorithms, are used to calculate multiple amounts of correction to the molding conditions, and the correction amounts related to the counterproposals are excluded, making it possible to correct the operating conditions using the correct correction amounts.

[0054] Furthermore, the control device 4 can select, by a simple majority vote, a correction amount in a correct correction direction from among a plurality of correction amounts output from a plurality of learning devices 40. In the first embodiment, even if the positive or negative of each change value constituting the correction amount is unknown, the control device 4 can select a correct correction amount and exclude counterproposals.

[0055] Furthermore, the control device 4 is configured to calculate an average correction amount using multiple correct correction amounts, excluding correction amounts related to counterproposals, and correct the molding conditions. Therefore, the molding conditions can be corrected using unbiased correction amounts, and the molding conditions can be optimized.

[0056] In this embodiment, an example of correcting the molding conditions of the injection molding machine 101 has been described, but the present technology may be applied to a control device that corrects the molding conditions of an extruder or other molding machine. Also, the present technology may be applied to a control device that corrects the operating conditions of any industrial machine that manufactures products.

[0057] (Embodiment 2) The injection molding machine 201 according to the second embodiment differs from the embodiment 1 in the method of removing the correction amount of the counterproposal. Since the other configurations of the injection molding machine 201 are similar to those of the injection molding machine 101 according to the first embodiment, the same reference numerals are used for the similar parts, and detailed description thereof will be omitted.

[0058] 5 is a block diagram showing a configuration example of an injection molding machine 201 according to the embodiment 2. The storage unit 42 of the control device 4 according to the embodiment 2 stores a direction definition table 242b for determining whether or not a correction amount is a counterproposal.

[0059] 6 is a conceptual diagram showing a direction definition table 242b according to the present embodiment 2. The direction definition table 242b stores the types of defects in the molded product and combinations of positive and negative change values ​​that constitute the correct correction amounts, in association with each other.

[0060] For example, for a burr, which is a defect in a molded product, information indicating that the correct correction direction is a positive change value for the packing pressure, a negative change value for the packing pressure switching position, and a positive change value for the injection speed is associated with the molded product. Similarly, for a short, which is a defect in a molded product, information indicating that the correct correction direction is a negative change value for the packing pressure, a positive change value for the packing pressure switching position, and a negative change value for the injection speed is associated with the molded product.

[0061] In particular, in the case of burrs caused by resin viscosity, information indicating that the change value of the holding pressure switching position is positive, the change value of the holding pressure switching speed is negative, and the change value of the holding pressure switching pressure is negative may be associated as the correct correction direction. In other words, the direction definition table 242b stores data indicating burrs caused by resin viscosity, data indicating that the appropriate change value of the holding pressure switching position is positive, data indicating that the appropriate change value of the holding pressure switching speed is negative, and data indicating that the appropriate change value of the holding pressure switching pressure is negative, in association with each other. Similarly, in the case of a short circuit caused by resin viscosity, information indicating that the change value of the holding pressure switching position is negative, the change value of the holding pressure switching speed is positive, and the change value of the holding pressure switching pressure is positive may be associated as the correct correction direction. In other words, the direction definition table 242b stores the inspection result data indicating a short circuit caused by resin viscosity, data indicating that the appropriate change value of the holding pressure switching position is negative, data indicating that the appropriate change value of the holding pressure switching speed is positive, and data indicating that the appropriate change value of the holding pressure switching pressure is positive, in association with each other.

[0062] 7 is a conceptual diagram showing an operating condition correction processing method according to the embodiment 2. The counterproposal determination processor 241a according to the embodiment 2 is a functional unit that determines whether or not the correction amounts output from the multiple learning devices 40 are correction amounts related to counterproposals by referring to information in the direction definition table 242b.

[0063] Specifically, the processor 41 determines the type of defect of the molded product based on the inspection result data. The types of defects include, for example, burrs and shorts. The types of defects may be subdivided based on the cause of the defect. For example, the types of defects may include burrs caused by resin viscosity and shorts caused by resin viscosity. The counterproposal determination processing unit 241a refers to the direction definition table 242b using the type of defect of the molded product as a key, and reads out information indicating the correct correction direction. Then, the counterproposal determination processing unit 241a determines whether the correction amount to be determined is a counterproposal by determining whether the positive and negative change values ​​of the correction direction match the positive and negative change values ​​indicated by the information read out from the direction definition table 242b. For example, the counterproposal determination processing unit 241a selects correction amounts whose positive and negative values ​​all match as correct correction amounts, and excludes other correction amounts as correction amounts of counterproposals.

[0064] 8 is a flowchart showing the procedure of the operating condition correction process according to the embodiment 2. As in steps S111 to S113 of the embodiment 1, the processor 41 executes injection molding, acquires measurement data and inspection result data, and calculates a plurality of correction amounts using a plurality of learning devices 40 (steps S211, S212, and S213). Next, the processor 41 determines whether the inspection result is good or not, that is, whether the molded product is a non-defective product or not, based on the inspection result data (step S214).

[0065] When it is determined that the inspection result is good (step S214: YES), the processor 41 returns the process to step S211. When it is determined that the inspection result is not good (step S214: NO), the processor 41 determines whether or not the appropriateness of all the correction amounts output from the multiple learning devices 40 has been confirmed (step S215). When it is determined that there is a correction amount whose appropriateness has not been confirmed (step S215: NO), the processor 41 as the counterproposal determination processing unit 41a identifies the type of defect of the molded product, refers to the direction definition table 242b using the type of defect as a key, and determines whether or not the n-th correction amount is a counterproposal (step S216). Here, "n" is an integer, starting from n=1, and is incremented by 1 each time the processes of steps S215 to S217 are executed.

[0066] When processor 41 determines that the n-th correction amount is a counterproposal (step S216: YES), it excludes the correction amount (step S217) and returns the process to step S215. When processor 41 determines that the n-th correction amount is the correct correction direction (step S216: NO), it returns the process to step S215 without excluding the correction amount.

[0067] In step S215, if it is determined that the appropriateness of all correction amounts has been determined (step S215: YES), processor 41 calculates an average correction amount using the correct correction amounts excluding the counter-proposed correction amounts (step S218), sets molding conditions corrected using the average correction amount (step S219), and returns the process to step S211.

[0068] According to the injection molding machine 101 and the control device 4 (operating condition correction device) of this embodiment 2 configured as described above, if the direction definition table 242b can be appropriately set, the correct correction amount for the correction direction can be reliably selected and the molding conditions can be appropriately corrected.

[0069] (Embodiment 3) An injection molding machine 301 according to the third embodiment differs from that according to the first embodiment in the correction process after removing the correction amount of the counterproposal. Since the other configurations of the injection molding machine 301 are similar to those of the injection molding machine 101 according to the first embodiment, the same reference numerals are used for the similar parts, and detailed description thereof will be omitted.

[0070] 9 is a block diagram showing an example of the configuration of an injection molding machine 301 according to the present embodiment 3. A processor 41 according to the embodiment 3 has a suitability determination model 340 as a functional unit. A storage unit 42 stores various coefficients that characterize the suitability determination model 340.

[0071] 10 is a conceptual diagram showing an appropriateness judgment model 340 according to the third embodiment. The appropriateness judgment model 340 is a neural network having an input layer 340a to which the inspection result data, the measurement data, and the correction amount are input, an intermediate layer 340b that extracts feature quantities such as the correction amount, and an output layer 340c that outputs appropriateness data indicating the appropriateness of the correction amount calculated based on the extracted feature quantities. The appropriateness data is information indicating whether the correction amount is appropriate or not. For example, the larger the value of the appropriateness data, the more appropriate the correction amount is.

[0072] The input layer 340a of the neural network has a plurality of neurons to which the inspection result data, measurement data, and correction amounts are input, and passes each input data to the intermediate layer 340b.

[0073] The intermediate layer 340b has multiple layers each made up of multiple neurons. Each layer extracts features related to the appropriateness of the correction amount from the input data, and passes them on to the next layer in turn, until the last layer passes them on to the output layer 340c.

[0074] The output layer 340c includes neurons that output the calculation results, and these neurons output whether the correction amount is appropriate or not.

[0075] In the first embodiment, the suitability determination model 340 is an example of a general neural network, but it may be another neural network such as a transformer, a support vector machine (SVM), a Bayesian network, or a regression tree. The suitability determination model 340 may be configured as a reinforcement learning model.

[0076] A learning method for the suitability judgment model 340 will be described. First, learning data is collected that associates inspection result data, measurement data, correction amounts, and suitability data indicating whether the correction amounts are appropriate. The correction amounts used for learning are all correction amounts in the appropriate direction. The suitability data is created based on information indicating the degree to which the molded product has improved when the molding conditions are corrected with the correction amount. For example, the suitability data may be the amount or rate of reduction in burr area, the amount or rate of reduction in short area, etc.

[0077] The model learning computer uses the generated learning data to perform machine learning or deep learning on the pre-learning neural network model, thereby generating the suitability determination model 340.

[0078] Specifically, the computer inputs the inspection result data, the measurement data, and the correction amount included in the learning data into the neural network model before learning, and obtains the suitability data output from the output layer 340c through the arithmetic processing in the intermediate layer 340b. Then, the computer compares the suitability data output from the output layer 340c with the suitability data indicated by the teacher data, and optimizes the parameters used in the arithmetic processing in the intermediate layer 340b so that the suitability data output from the output layer 340c approaches the correct answer value. The parameters are, for example, weights (coupling coefficients) between neurons. There is no particular limit to the method of optimizing the parameters, but for example, the computer optimizes various parameters using the steepest descent method or the like. The computer obtains the learned suitability judgment model 340 by repeating the above processing.

[0079] The processor 41 according to the third embodiment can obtain suitability data by inputting each of the multiple correction amounts from which the counterproposals are excluded to the suitability determination model 340. Then, the processor 41 selects the most appropriate correction amount from the multiple correction amounts from which the counterproposals are excluded based on the suitability data, and corrects the molding conditions.

[0080] According to the injection molding machine 301 of the present embodiment 3 configured as described above, by using the suitability judgment model 340, it is possible to select the most appropriate correction amount from among correction amounts having the correct correction direction, and correct the molding conditions.

[0081] Although the third embodiment has been described as a modification of the first embodiment, the technique of the third embodiment can also be combined with the second embodiment.

[0082] Means for solving the problems of the present disclosure are described below. (Appendix 1) An operating condition correction method for correcting an operating condition of an industrial machine, comprising the steps of: acquiring measurement data obtained by measuring a state of the industrial machine and inspection result data obtained by inspecting a state of a product manufactured by the industrial machine; calculating the amount of correction of the operating condition based on the acquired measurement data and the inspection result data using a plurality of learning devices that have learned the relationship between the measurement data and the inspection result data and the amount of correction of the operating condition; a determination step of determining whether or not a direction of correction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate; modifying the operating conditions based on one or more of the correction amounts for which the correction direction is determined to be appropriate; An operating condition correction method comprising: (Appendix 2) The determining step includes: The appropriateness of the amount of correction is judged by majority vote of the direction of correction. The method for correcting operating conditions described in appendix 1. (Appendix 3) The determining step includes: A table in which the type of defect indicated by the inspection result data is associated with an appropriate direction of correction of the operating conditions is referenced using the acquired inspection result data as a key, and the appropriateness of the amount of correction is determined. 5. The method for modifying operating conditions according to claim 1 or 2. (Appendix 4) the industrial machine is an injection molding machine, The table includes: The data includes, in association with each other, data indicating burrs caused by resin viscosity, data indicating that an appropriate change value for the holding pressure switching position is positive, data indicating that an appropriate change value for the holding pressure switching speed is negative, and data indicating that an appropriate change value for the holding pressure switching pressure is negative. The operating condition correction method described in appendix 3. (Appendix 5) the industrial machine is an injection molding machine, The table includes: The data includes, in association with each other, data indicating a short circuit caused by resin viscosity, data indicating that an appropriate change value of the holding pressure switching position is negative, data indicating that an appropriate change value of the holding pressure switching speed is positive, and data indicating that an appropriate change value of the holding pressure switching pressure is positive. 5. The operating condition correction method according to claim 3 or 4. (Appendix 6) When the correction amount is input, the calculated correction amount is input to an appropriateness judgment model that outputs appropriateness data indicating the appropriateness of the correction amount, thereby judging the appropriateness of the correction amount. 6. The operating condition modification method according to any one of claims 1 to 5. (Appendix 7) At least one of the plurality of learning devices has learned the relationship between the measurement data and the inspection result data and the correction amount by using Adam, momentum, or Adagrad. 7. The operating condition modification method according to any one of claims 1 to 6. (Appendix 8) calculating an average value of a plurality of the correction amounts for which the correction direction is determined to be appropriate; 8. The operating condition modification method according to claim 1, further comprising modifying the operating conditions based on the calculated average value. [Explanation of symbols]

[0083] 1: Molding machine body 1a: Sensor 2: Mold clamping device 3: Injection device 4: Control device 5: Inspection equipment 40: Learning device 41: Processor 41a: Counterproposal judgment processing unit 41b: Molding condition setting processing section 42: Storage section 42a: Computer programs 43: Control signal output section 44:First acquisition part 45:Second acquisition part 46: Operation panel 49: Recording media 101: Injection molding machine 241a: Counterproposal judgment processing unit 242b : Direction definition table 340: Suitability assessment model

Claims

1. An operating condition correction method for correcting an operating condition of an industrial machine, comprising the steps of: acquiring measurement data obtained by measuring a state of the industrial machine and inspection result data obtained by inspecting a state of a product manufactured by the industrial machine; calculating the amount of correction of the operating condition based on the acquired measurement data and the inspection result data using a plurality of learning devices that have learned the relationship between the measurement data and the inspection result data and the amount of correction of the operating condition; a determination step of determining whether or not a direction of correction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate; modifying the operating conditions based on one or more of the correction amounts for which the correction direction is determined to be appropriate; An operating condition correction method comprising:

2. The determining step includes: The appropriateness of the amount of correction is judged by majority vote of the direction of correction. The method for modifying operating conditions according to claim 1 .

3. The determining step includes: A table in which the type of defect indicated by the inspection result data is associated with an appropriate direction of correction of the operating conditions is referenced using the acquired inspection result data as a key to determine whether the amount of correction is appropriate. The method for modifying operating conditions according to claim 1 .

4. the industrial machine is an injection molding machine, The table includes: The data includes, in association with each other, data indicating burrs caused by resin viscosity, data indicating that an appropriate change value for the holding pressure switching position is positive, data indicating that an appropriate change value for the holding pressure switching speed is negative, and data indicating that an appropriate change value for the holding pressure switching pressure is negative. The method for modifying operating conditions according to claim 3.

5. the industrial machine is an injection molding machine, The table includes: The data includes, in association with each other, data indicating a short circuit caused by resin viscosity, data indicating that an appropriate change value of the holding pressure switching position is negative, data indicating that an appropriate change value of the holding pressure switching speed is positive, and data indicating that an appropriate change value of the holding pressure switching pressure is positive. The method for modifying operating conditions according to claim 3.

6. When the correction amount is input, the calculated correction amount is input to an appropriateness judgment model that outputs appropriateness data indicating the appropriateness of the correction amount, thereby judging the appropriateness of the correction amount. The method for modifying operating conditions according to claim 1 .

7. At least one of the plurality of learning devices has learned the relationship between the measurement data and the inspection result data and the correction amount using Adam, Momentum, or Adagrad. The method for modifying operating conditions according to claim 1 .

8. calculating an average value of a plurality of the correction amounts for which the correction direction is determined to be appropriate; The method for modifying an operating condition according to claim 1 , further comprising the step of modifying the operating condition based on the calculated average value.

9. An operating condition modification device for modifying operating conditions of an industrial machine, comprising: an acquisition unit that acquires measurement data obtained by measuring a state of the industrial machine and inspection result data obtained by inspecting a state of a product manufactured by the industrial machine; a plurality of learning devices that learn the relationship between the measurement data and the inspection result data and the correction amount of the operating condition, and calculate the correction amount of the operating condition based on the acquired measurement data and the inspection result data; a determination unit that determines whether or not a direction of correction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate; a correction unit that corrects the operating conditions based on one or more of the correction amounts whose correction direction is determined to be appropriate; An operating condition modification device comprising:

10. A molding machine comprising the operating condition correction device according to claim 9.

11. A computer program for causing a computer to execute a process for correcting an operating condition of an industrial machine, The computer includes: acquiring measurement data obtained by measuring a state of the industrial machine and inspection result data obtained by inspecting a state of a product manufactured by the industrial machine; calculating the amount of correction of the operating condition based on the acquired measurement data and the inspection result data using a plurality of learning devices that have learned the relationship between the measurement data and the inspection result data and the amount of correction of the operating condition; a determination step of determining whether or not a direction of correction of each of the plurality of correction amounts calculated using the plurality of learning devices is appropriate; modifying the operating conditions based on one or more of the correction amounts for which the correction direction is determined to be appropriate; A computer program that executes the following:

Citation Information

Patent Citations

  • Machine learning method, computer program, machine learning device, and molding machine

    JP2022045698A