Automatic evaluation device, automatic evaluation method, and computer program

Abstract

An automatic evaluation device for a learned model that corrects the operating conditions of an industrial machine is provided with: a reception unit that receives a plurality of defect-inducing conditions that induce the occurrence of defective product operations; and a processing unit that generates a plurality of defect conditions on the basis of the plurality of defect inducing conditions received by the receiving unit, sets the generated defect conditions in the industrial machine, and, when operating the industrial machine to which the defect conditions have been set, acquires an operation result when the operation conditions have been corrected by the learned model, and processes the acquired operation result. A plurality of failure conditions are set in the industrial machine, and an evaluation report of the learned model is generated on the basis of operation results obtained by operating the industrial machine a plurality of times with respect to each of the failure conditions.

Description

Technical Field

[0001] The present invention relates to an automatic evaluation device, an automatic evaluation method, and a computer program for a learned model. Background Art

[0002] When performing injection molding using an injection molding machine, it is necessary to first perform a condition setting operation to correct the set values of various molding condition items in order to obtain the molding conditions. The correction of the molding conditions is performed based on the experience of the operator, and repeated trials are required to obtain appropriate molding conditions. Patent Document 1 discloses an injection molding machine system that corrects the molding conditions of an injection molding machine by reinforcement learning.

[0003] Prior Art Documents

[0004] Patent Documents

[0005] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2019-166702 Summary of the Invention

[0006] The performance evaluation of a learned model generated by reinforcement learning or other machine learning is performed as follows: A person sets predetermined defective molding conditions and records at which injection the defect becomes a good product (qualified product), or how many injections maintain a good product within a specified number of injections. The performance evaluation of the learned model is performed manually and requires a lot of time. In addition, there are the same problems with the performance evaluation of a learned model for correcting the operating conditions of industrial machinery.

[0007] An object of the present disclosure is to provide an automatic evaluation device, an automatic evaluation method, and a computer program that can automatically perform performance evaluation of a learned model for correcting the operating conditions of industrial machinery only by setting an action that causes defective products in the industrial machinery.

[0008] An automatic evaluation device according to an aspect of the present disclosure is an automatic evaluation device for a learned model that corrects the operating conditions of industrial machinery, and includes: a reception unit that receives a plurality of defect-inducing conditions that induce an action that causes defective products; and a processing unit that performs the following processing: Based on the plurality of defect-inducing conditions received by the reception unit, generate a plurality of defective conditions; set the generated defective conditions in the industrial machinery; when the industrial machinery set with the defective conditions is operated, obtain an operation result when the operating conditions are corrected by the learned model; set the plurality of defective conditions in the industrial machinery respectively, and generate an evaluation report of the learned model based on the operation results obtained by operating the industrial machinery multiple times for each defective condition.

[0009] An automatic evaluation method according to one aspect of the present disclosure is an automatic evaluation method for a learned model that corrects the operating conditions of industrial machinery. In this method, a computer receives a plurality of defect-inducing conditions that induce defective product actions; generates a plurality of defective conditions based on the received plurality of defect-inducing conditions; sets the generated defective conditions in the industrial machinery; when the industrial machinery set with the defective conditions is operated, obtains the operating result when the learned model corrects the operating conditions; sets the plurality of defective conditions in the industrial machinery respectively, and generates an evaluation report of the learned model based on the operating results obtained by operating the industrial machinery multiple times for each defective condition.

[0010] A computer program (program product) according to one aspect of the present disclosure is a computer program that causes a computer to evaluate a learned model that corrects the operating conditions of industrial machinery. It causes the computer to perform the following processing: receive a plurality of defect-inducing conditions that induce defective product actions; generate a plurality of defective conditions based on the received plurality of defect-inducing conditions; set the generated defective conditions in the industrial machinery; when the industrial machinery set with the defective conditions is operated, obtain the operating result when the learned model corrects the operating conditions; set the plurality of defective conditions in the industrial machinery respectively, and generate an evaluation report of the learned model based on the operating results obtained by operating the industrial machinery multiple times for each defective condition.

[0011] Advantages of the Invention

[0012] According to the present invention, it is possible to automatically perform a performance evaluation of a learned model that corrects the operating conditions of industrial machinery only by setting actions that induce defective product production in the industrial machinery. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram showing a configuration example of an injection molding apparatus according to the present embodiment.

[0014] Figure 2 It is a block diagram showing a configuration example of a control device according to the present embodiment.

[0015] Figure 3 It is a block diagram showing a configuration example of an automatic evaluation device according to the present embodiment.

[0016] Figure 4 It is a flowchart showing the processing procedure of the automatic evaluation method according to the present embodiment.

[0017] Figure 5 It is a flowchart showing the processing procedure of the automatic evaluation method according to the present embodiment.

[0018] Figure 6Schematic diagram showing an example of the evaluation report of this embodiment. Detailed implementation

[0019] The following describes specific examples of the automatic evaluation device, automatic evaluation method, and computer program according to the embodiments of the present invention with reference to the drawings. At least a part of the embodiments described below can be arbitrarily combined. In addition, the present invention is not limited to these examples, but is shown by the claims, and is intended to include all changes within the meaning and scope equivalent to the claims.

[0020] <Composition of injection molding machine>

[0021] Figure 1 Schematic diagram showing a configuration example of the injection molding machine 101 of this embodiment. The injection molding machine 101 of this embodiment includes a mold clamping device 2 for clamping the mold 21, an injection device 3 for plasticizing and injecting the molding material, a control device 4, and an inspection device 5. The mold clamping device 2 and the injection device 3 constitute the molding machine main body 1. The control device 4 functions as a molding condition correction device for correcting the molding conditions using the learned model 40 (refer to Figure 2 ) described later. As Figure 1 shown, the automatic evaluation device 6 of this embodiment is provided in the injection molding machine 101. The automatic evaluation device 6 is a device for evaluating the learned model 40.

[0022] The mold clamping device 2 includes a fixed plate 22 fixed to the base 20, a mold clamping housing 23 provided so as to be slidable on the base 20, and a movable plate 24 that also slides on the base 20. The fixed plate 22 and the mold clamping housing 23 are connected by a plurality of, for example, four tie rods 25, 25,.... The movable plate 24 is configured to be slidable between the fixed plate 22 and the mold clamping housing 23. A mold clamping mechanism 26 is provided between the mold clamping housing 23 and the movable plate 24.

[0023] The mold clamping mechanism 26 is constituted by, for example, a toggle mechanism. In addition, the mold clamping mechanism 26 can be constituted by a direct pressure type mold clamping mechanism, that is, a mold clamping cylinder. Fixed molds 21a and movable molds 21b are provided on the fixed plate 22 and the movable plate 24 respectively, and when the mold clamping mechanism 26 is driven, the mold 21 is opened and closed.

[0024] The injection device 3 is provided on the base 30. The injection device 3 includes a heating cylinder 31 having a nozzle 31a at the tip, and a screw 32 disposed in the heating cylinder 31 so as to be rotatable in the circumferential and axial directions. The screw 32 is driven in the rotational direction and the axial direction by a drive mechanism 33. The drive mechanism 33 includes a rotary motor for driving the screw 32 in the rotational direction, and a motor for driving the screw 32 in the axial direction, etc. In addition, Figure 1The shown drive mechanism 33 is covered by a cover, so the internal structure is not shown.

[0025] Near the rear end portion of the heating cylinder 31, there is a hopper 34 into which the molding material is put. In addition, the injection molding machine 101 includes a nozzle contact device 35 that moves the injection device 3 in the front-rear direction ( Figure 1 the left-right direction in the figure). It is configured that when the nozzle contact device 35 is driven, the injection device 3 advances and the nozzle 31a of the heating cylinder 31 contacts the abutting portion of the fixed plate 22.

[0026] Figure 2 It is a block diagram showing a configuration example of the control device 4 of the present embodiment. The control device 4 is a computer that controls the operations of the mold clamping device 2 and the injection device 3, and includes a control unit 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 as hardware configurations. The control unit 41 has a learned model 40 as a functional unit. The learned model 40 is a model that outputs a correction amount of a specified molding condition when the currently set molding conditions and the inspection results (operation results) obtained by inspecting the state of the molded product are input. The learned model 40 can be a reinforcement learning model or a model obtained by supervised learning. The type and configuration of the learned model 40 are not particularly limited. In addition, a part of the learned model 40 can be implemented in hardware.

[0027] The control device 4 is a device that controls the operation of the injection molding machine 101 (refer to Figure 1 ) and corrects the molding conditions. In addition, the control device 4 can be a server device connected to a network. Further, the control device 4 can be configured by multiple computers for distributed processing, can be implemented by multiple virtual machines provided in one server, or can be implemented using a cloud server.

[0028] The control unit 41 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), NPU (Neural Processing Unit), internal storage devices such as a ROM (Read Only Memory) and RAM (Random Access Memory), I / O terminals, a timing unit, and the like. The control unit 41 executes the correction process of the molding conditions by executing the control program (program product) stored in the storage unit 42 described later. In addition, each functional unit of the control device 4 can be implemented in software, and part or all of it can also be implemented in hardware.

[0029] The storage unit 42 is a non-volatile memory such as a hard disk, EEPROM (Electrically Erasable Programmable ROM), or flash memory. The storage unit 42 stores a control program for performing reinforcement learning on a method for correcting molding conditions corresponding to the injection molding machine 101 and the state of the molded product, and causing the computer to execute the molding condition correction process. The storage unit 42 stores various coefficients characterizing the learned model 40 as a reinforcement learning model.

[0030] The control signal output unit 43 outputs a control signal for controlling the operation of the injection molding machine 101 to the molding machine main body 1 in accordance with the control of the control unit 41 based on the molding conditions.

[0031] The operation panel 46 is an interface for setting the molding conditions of the injection molding machine 101 and 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 the setting of the molding conditions of the injection molding machine 101, and displays the state of the injection molding machine 101 and the implementation status of the molding condition correction method of the present embodiment in accordance with the control of the control unit 41. The operation device is an input device for inputting and correcting the molding conditions of the injection molding machine 101, and includes operation buttons, a touch panel, and the like. The operation device provides data representing the received molding conditions to the control unit 41.

[0032] Set the set values for various molding conditions in the injection molding machine 101. The molding conditions include the injection start position, mold temperature, nozzle temperature, cylinder temperature (heater temperature), hopper temperature, clamping force, injection speed, injection acceleration, injection peak pressure (injection pressure), injection stroke. Additionally, the molding conditions include the resin pressure at the cylinder tip, the seated state of the reflux prevention ring, holding pressure, holding pressure switching speed, holding pressure switching position, holding pressure completion position, cushion position, back pressure, metering torque. Additionally, the molding conditions include the metering completion position, screw retraction speed, cycle time, mold closing time, injection time, holding pressure time, metering time, mold opening time. Additionally, the molding conditions include the cooling time, screw rotation speed, mold opening and closing speed, ejection speed, ejection times.

[0033] The injection molding machine 101 with these set values set operates according to these set values. Among the molding conditions as described above, in particular, the holding pressure, holding pressure switching position, and injection speed are molding conditions related to molding defects such as flash and short shot of the molded product. In the present embodiment, an example of correcting the holding pressure [Mpa], holding pressure switching position [mm], and injection speed [mm / sec] using the learned model 40 will be described.

[0034] The first acquisition unit 44 is an input circuit for acquiring measurement data obtained by measuring the state of the injection molding machine 101.

[0035] One or more sensors 1a are provided in the clamping device 2 and the injection device 3 (refer to Figure 1 ), and the one or more sensors 1a detect physical quantities required for controlling the operation of the molding machine main body 1 as information indicating the state of the injection molding machine 101. The sensor 1a is connected to the first acquisition unit 44. The physical quantities include, for example, forces such as voltage, current, temperature, humidity, torque, pressure, the speed, acceleration, rotation angle, position of a movable part, the flow rate, speed, etc. of a fluid. The sensor 1a is, for example, a current sensor, voltage sensor, temperature sensor, humidity sensor, torque sensor, pressure sensor, speed sensor, acceleration sensor, rotation angle sensor, position measuring sensor, flow sensor, flow velocity meter, etc. 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.

[0036] The measurement data is data representing the operating status of the injection molding machine 101, and includes, for example, cycle time [seconds], injection time [seconds], holding time [seconds], holding pressure switching position [mm], holding pressure switching speed [mm / second], holding pressure switching pressure [Pa], cushioning member position [mm], holding pressure completion position [mm], metering time [seconds], back pressure [Pa], metering completion position [mm], etc.

[0037] The second acquisition unit 45 is an input circuit that acquires inspection result data obtained by inspecting the state of the molded product molded by the injection molding machine 101. An inspection device 5 is connected to the second output unit. The inspection device 5 is a camera, a distance measuring sensor, a weighing instrument, etc. that detect the state of the molded product, measure the physical quantity regarding the state of the molded product, and obtain inspection result data representing the state of the molded product based on the measured physical quantity data. The inspection result data is, for example, data representing the flash area [mm 2 and the short shot area [mm 2 of the molded product.

[0038] In addition, an automatic evaluation device 6 is connected to the control device 4. The automatic evaluation device 6 can be configured to be detachable from the control device 4. When evaluating the learned model 40, the control unit 41 of the control device 4 cooperates with the automatic evaluation device 6 to execute the process related to the evaluation of the learned model 40. Specifically, the control unit 41 executes the following process: receives the information for inducing molding defects (the defective molding conditions and the number of tests described later) output from the automatic evaluation device 6 described later, and based on the received information, changes the molding conditions to induce molding defects, performs molding multiple times, and sends the inspection result data of the molded product to the automatic evaluation device 6.

[0039] Figure 3 FIG. is a block diagram showing a configuration example of the automatic evaluation device 6 of the present embodiment. The automatic evaluation device 6 is a computer that automatically evaluates the performance of the learned model 40, and includes a processor (processing unit) 61, a storage unit 62, a reception unit 63, a data input / output unit 64, and a report output unit 65 as a hardware configuration. The automatic evaluation device 6 can be a server device connected to a network. In addition, the control device 4 and the automatic evaluation device 6 can be one computer. That is, it can be configured that the control device 4 executes the process of the automatic evaluation device 6. In addition, the automatic evaluation device 6 can be configured to perform distributed processing by multiple computers, can be implemented by multiple virtual machines provided in one server, or can be implemented using a cloud server.

[0040] The processor 61 has computing circuits such as a CPU, multi-core CPU, GPU, GPGPU, TPU, ASIC, FPGA, NPU, etc., internal storage devices such as a ROM and RAM, I / O terminals, a timing unit, etc. The processor 61 implements the automatic evaluation method of the present embodiment by executing a computer program (program product) 62a stored in a storage unit 62 described later. In addition, each functional unit of the automatic evaluation device 6 can be implemented in software, and a part or all of it can also be implemented in hardware.

[0041] The storage unit 62 is a non-volatile memory such as a hard disk, EEPROM, or flash memory. The storage unit 62 stores a computer program 62a for automatically performing performance evaluation of the learned model 40. In addition, the storage unit 62 stores an evaluation report 62b as the evaluation result of the learned model 40.

[0042] The computer program 62a of the present embodiment can be in a form stored in a recording medium 69 in a computer-readable manner. The storage unit 62 stores the computer program 62a read out from the recording medium 69 by a reading device. The recording medium 69 is a semiconductor memory such as a flash memory. In addition, the recording medium 69 can be an optical disc such as a CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, BD (Blu-ray (registered trademark) Disc). In addition, the recording medium 69 can be a magnetic disk such as a floppy disk or hard disk, a magneto-optical disc, etc.

[0043] In addition, the computer program 62a of the present embodiment can be downloaded from an external server connected to a communication network and stored in the storage unit 62.

[0044] The reception unit 63 is an input device such as a keyboard, touch panel, or mouse. In order to perform performance evaluation of the learned model 40, the reception unit 63 receives information for inducing defective product generation actions of the injection molding machine 101.

[0045] Specifically, the reception unit 63 receives the upper limit value, lower limit value, division number, and number of test injections (m times) of each of a plurality of defect induction conditions that induce defects in the molded product, which are set in the operating conditions of the molding machine main body 1. Hereinafter, this number will be referred to as the number of tests. The upper limit value and the lower limit value are information for determining the numerical range of the defect induction conditions that can be changed to induce defects in the molded product. The division number is the number of divisions within this numerical range. As long as the division number is information for determining the number of defect induction conditions created by dividing within this numerical range, its definition is not particularly limited.

[0046] The defect-inducing conditions are, for example, the metering completion position of the molding material, the switching position of the injection speed, or the back pressure. By changing these conditions, the filling amount of the molding material into the mold 21 can be increased or decreased.

[0047] The number of test times is the number of test injections for each of a plurality of different defective molding conditions. The receiving unit 63 provides the upper limit value, the lower limit value, the number of divisions, and the number of test times of each of the metering completion position, the first injection switching position, the second injection switching position, and the back pressure to the processor 61.

[0048] The first injection switching position is the first position where the injection speed is switched after the start of injection. The second injection switching position is the second position where the injection speed is switched after the injection speed is switched at the first injection switching position.

[0049] For example, when the upper limit value of the injection switching position is 60 [mm], the lower limit value is 40 [mm], and the number of divisions is n = 20, the set value range (40 to 60 [mm]) of the injection switching position can be divided into n at an interval of (40 - 20) / n. The set values of the injection switching position obtained by dividing into n are 40, 41, 42,... 60. The number of set values is (n + 1). The same applies to other defect-inducing conditions.

[0050] The data input / output unit 64 is an input / output interface for inputting and outputting data between the control device 4. The input / output interface includes terminals for data signal input / output, communication circuits, etc. The data input / output unit 64 is connected to the control device 4 in a wired or wireless manner.

[0051] The processor 61 generates a plurality of (n) defective molding conditions based on the upper limit value, the lower limit value, and the number of divisions of the defect-inducing conditions received by the receiving unit 63. Specifically, the processor 61 divides the numerical range represented by the upper limit value and the lower limit value of the defect-inducing condition by the number of divisions to obtain a plurality of set values of the defect-inducing condition. Similarly, the processor 61 also obtains a plurality of set values for other defect-inducing conditions. And, the processor 61 generates a plurality of defective molding conditions by combining the set values of the plurality of defect-inducing conditions. The processor 61 outputs the data representing the generated defective molding conditions and the number of test times to the control device 4 via the data input / output unit 64. For example, by outputting the n defective molding conditions and the number of test times to the control device 4 in sequence at intervals of m test injections, the defective molding conditions are set. In addition, it can be configured to summarize and output the plurality of (n) defective molding conditions and the number of test times to the control device 4.

[0052] The control device 4 sets the defective molding conditions output from the automatic evaluation device 6 and operates the molding machine main body 1. In addition, the defective molding conditions are a part of various molding conditions that are set. The control device 4 manufactures molded products according to the number of tests accepted by the acceptance unit 63. In addition, the control device 4 corrects the molding conditions using the learned model 40. Specifically, in the case of injection molding, the control device 4 acquires the inspection result data obtained by inspecting the molded product by the inspection device 5, and inputs the current molding conditions and the inspection result data into the learned model 40, so that it outputs the correction amount of the molding conditions. The control device 4 corrects the molding conditions based on the correction amount output from the learned model 40. The control device 4 outputs the correction contents of the molding conditions and the inspection results performed in each of the test injections of the aforementioned number of tests to the automatic evaluation device 6.

[0053] The processor 61 acquires the correction contents and the inspection result data via the data input / output unit 64. The correction contents are, for example, information associated with the output of the learned model 40 such as the correction amount output from the learned model 40, or the molding conditions corrected based on the correction amount. The processor 61 generates an evaluation report 62b of the learned model 40 (refer to Figure 6 ), and this evaluation report 62b is based on a plurality of defective molding conditions set in the control device 4 of the injection molding machine 101, the correction contents of the molding conditions using the learned model 40, and the inspection results of the molded products.

[0054] The report output unit 65 is an output device such as a display device or a printing device. The report output unit 65 outputs the evaluation report 62b generated by the processor 61 to the outside. In addition, the report output unit 65 can be a communication circuit that sends the evaluation report 62b to an external terminal.

[0055] Figure 4 and Figure 5 is a flowchart showing the processing procedure of the automatic evaluation method of the present embodiment. The processor 61 accepts the upper limit value, the lower limit value, and the number of divisions of a plurality of defect-inducing conditions, and the number of tests performed for each defective molding condition (step S11).

[0056] Next, the processor 61 divides the numerical range of the upper limit value and the lower limit value of each defect-inducing condition by the number of divisions, thereby calculating the set value of each of the plurality of defect-inducing conditions (step S12). The processor 61 generates n defective molding conditions by combining the calculated set values of the respective defect-inducing conditions (step S13).

[0057] For example, the processor 61 can generate a plurality of defective molding conditions by sequentially changing the set values of a plurality of defect-inducing conditions in ascending or descending order in a polling scheduling manner. Specifically, the set value of the first defect-inducing condition selected from the plurality of defect-inducing conditions is sequentially increased from the lower limit value to the upper limit value by a change amplitude corresponding to the number of divisions, thereby generating a defect-inducing condition. When the first defect-inducing condition reaches the upper limit value, the set value of the second defect-inducing condition is similarly changed. Similarly below, a plurality of defective molding conditions are obtained by sequentially changing the set values of the plurality of defective molding conditions.

[0058] Of course, the set value of the defect-inducing condition can also be changed by sequentially subtracting the change amplitude corresponding to the number of divisions from the upper limit value to the lower limit value, thereby generating a defect-inducing condition.

[0059] In addition, the polling scheduling method is an example, and it can also be configured to generate a plurality of defective molding conditions by randomly changing a plurality of defect-inducing conditions within the numerical range of the upper limit value and the lower limit value. However, the same defective molding condition is excluded.

[0060] Next, the processor 61 initially sets the defective molding condition variable X to 1 (step S14), and initially sets the test number variable Y to 1 (step S15). The defective molding condition variable is a variable used to distinguish a plurality of defective molding conditions. The test number variable is a variable indicating the number of test injections that have been performed.

[0061] Next, the processor 61 determines whether the defective molding condition variable X is the number n of defective molding conditions generated in step S13 (step S16). That is, the processor 61 determines whether test injections have been performed for all defective molding conditions.

[0062] When it is determined that the defective molding condition variable has not reached the number n of defective molding conditions (step S16: no), the processor 61 sets the Xth defective molding condition to the control device 4 (step S17).

[0063] Next, the processor 61 determines whether the test number variable Y is the test number m accepted in step S11 (step S18). When it is determined that the test number variable Y is the test number m (step S18: yes), the processor 61 returns the process to step S15. When it is determined that the test number variable Y has not reached the test number m (step S18: no), the processor 61 operates the molding machine main body 1 to manufacture a molded product, and inspects the molded product (step S19).

[0064] Further, the control device 4 corrects the molding conditions based on the inspection results of the molded product and the currently set molding conditions (step S20). For example, the control device 4 corrects the holding pressure, the holding pressure switching position, and the injection speed of the third stage using the learned model 40.

[0065] Next, the processor 61 obtains the correction content of the molding conditions and the inspection results from the control device 4 (step S21), and returns the process to step S18.

[0066] By repeating the above steps S15 to S21, it is possible to perform trial injections m times for each of the n defective molding conditions, and collect the correction content of the molding conditions performed using the learned model 40 and the inspection result data of the molded product obtained through this correction.

[0067] When it is determined in step S16 that the defective molding condition variable is the number of defective molding conditions n (step S16: YES), the processor 61 determines the number of injections from when the defective molding condition is set and defective products start to be molded until good molded products are obtained based on the obtained correction content and inspection result data (step S22). The processor 61 calculates the number of injections for each of the multiple defective molding conditions.

[0068] In addition, the processor 61 calculates the good product retention rate of the molded product based on the obtained correction content and inspection results (step S23). For example, the processor 61 calculates the good product retention rate by dividing the maximum number of continuous good product molding times in the case where a certain defective molding condition is set by the number of trials. The processor 61 calculates the good product retention rate for each of the multiple defective molding conditions.

[0069] Next, the processor 61 generates an evaluation report 62b representing the performance of the learned model 40 based on the generated defective molding conditions, the obtained correction content and inspection result data, and the calculated good product retention rate, etc. (step S24). And the processor 61 outputs the generated evaluation report 62b to the outside using the report output unit 65 (step S25), and ends the process.

[0070] Figure 6 FIG. is a schematic diagram showing an example of the evaluation report 62b of the present embodiment. The evaluation report 62b is, for example, a list corresponding to the set defective molding conditions, the correction content performed by the learned model 40, the inspection results of the molded product, and the evaluation for each of the multiple molding conditions. In addition, Figure 6 21 molding conditions are shown, but the number of defective molding conditions is an example.

[0071] More specifically, in each row of the list constituting the evaluation report 62b, there are shown defective molding conditions, correction contents, inspection results, and evaluations. The defective molding conditions include defective inducing conditions such as the metering completion position, the injection first-stage switching position, the injection second-stage switching position, and the back pressure. The correction contents include the holding pressure, the holding pressure switching position, and the injection speed of the third stage. The inspection results include the inspection result data of each molded product obtained during the test injection with the number of test times m. The evaluation includes the number of injection times from the occurrence of defects until good products are obtained and the good product maintenance rate.

[0072] Each column of the list constituting the evaluation report 62b corresponds to each of the multiple defective molding conditions. In each column, there are shown the set values of the above-mentioned defective inducing conditions, the correction amounts of the respective conditions to be corrected, the inspection results of the molded products, the number of injection times until good products are obtained, and the good product maintenance rate.

[0073] According to the automatic evaluation device 6 of the present embodiment configured as such, by simply setting the conditions for inducing defective product actions of the injection molding machine 101, the performance of the learned model 40 for correcting the molding conditions of the injection molding machine 101 can be automatically evaluated.

[0074] In addition, the automatic evaluation device 6 can output the result of the performance evaluation as an evaluation report 62b as shown in Figure 6 The operator of the injection molding machine 101, the technician who introduces the learned model 40 or performs maintenance inspections can confirm the performance of the learned model 40 by referring to the evaluation report 62b. The evaluation report 62b includes defective inducing conditions, correction contents of molding conditions, inspection results of molded products, and evaluations, and can perform detailed evaluation and verification of the performance of the learned model 40.

[0075] In addition, by changing the set values of the multiple defective inducing conditions in a round-robin scheduling manner, multiple defective molding conditions can be simply generated without repetition.

[0076] In addition, the performance of the learned model 40 for correcting the molding conditions of the injection molding machine 101 can be evaluated. In particular, the automatic evaluation device 6 of the present embodiment can efficiently evaluate the performance of the learned model 40 for coping with flash and short shot defects.

[0077] Furthermore, in the present embodiment, an example is described in which the evaluation of the learned model 40 for correcting the molding conditions for eliminating flash and short shot of the molded product is mainly based on the flash area and the short shot area as inspection results. However, the evaluation of the learned model 40 for correcting the molding conditions can also be performed based on the presence or absence of flash and short shot of the molded product. The evaluation report of the learned model 40 includes inspection results indicating the presence or absence of flash and short shot of the molded product.

[0078] In addition, it is also possible to evaluate the learned model 40 for eliminating other molding defects other than flash and short shot.

[0079] For example, when evaluating the learned model 40 for eliminating dents, weld lines, wavy surfaces (flow marks), or cracks, any one of the injection pressure, cylinder temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature can be adopted as the defect-inducing condition.

[0080] The learned model 40 for eliminating dents is a model that, when the molding conditions and the dent area, dent depth, or presence or absence of dents obtained by inspecting the state of the molded product are input, outputs the correction amount of the specified molding conditions. The evaluation report of the learned model 40 includes the inspection results indicating the dent area, dent depth, or presence or absence of dents of the molded product.

[0081] The learned model 40 for eliminating weld lines is a model that, when the molding conditions and the weld line area or presence or absence of weld lines obtained by inspecting the state of the molded product are input, outputs the correction amount of the specified molding conditions. The evaluation report of the learned model 40 includes the inspection results indicating the weld line area or presence or absence of weld lines of the molded product.

[0082] The learned model 40 for eliminating flow marks is a model that, when the molding conditions and the flow mark area or presence or absence of flow marks obtained by inspecting the state of the molded product are input, outputs the correction amount of the specified molding conditions. The evaluation report of the learned model 40 includes the inspection results indicating the flow mark area or presence or absence of flow marks of the molded product.

[0083] The learned model 40 for eliminating cracks is a model that, when the molding conditions and the crack area or presence or absence of cracks obtained by inspecting the state of the molded product are input, outputs the correction amount of the specified molding conditions. The evaluation report of the learned model 40 includes the inspection results indicating the crack area or presence or absence of cracks of the molded product.

[0084] When evaluating the learned model 40 for eliminating burn marks, or bubbles or voids, any one of the injection pressure, cylinder temperature, holding pressure, screw rotation speed, injection speed, and back pressure can be adopted as the defect-inducing condition.

[0085] The learned model 40 for eliminating the burn marks of the molded product is a model that, when the molding conditions and the burn mark area or presence or absence of burn marks obtained by inspecting the state of the molded product are input, outputs the correction amount of the specified molding conditions. The evaluation report of the learned model 40 includes the inspection results indicating the burn mark area or presence or absence of burn marks of the molded product.

[0086] The learned model 40 for eliminating bubbles or voids in a molded product is a model that, when the molding conditions and the number of bubbles or voids obtained by inspecting the state of the molded product are input, outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the number of bubbles or voids in the molded product.

[0087] When evaluating the learned model 40 for eliminating surface haze, or color mixing or specks, any one of the cylinder temperature, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature can be adopted as a defect-inducing condition.

[0088] The learned model 40 for eliminating surface haze in a molded product is a model that, when the molding conditions and the haze area or presence / absence of haze obtained by inspecting the state of the molded product are input, outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the haze area or presence / absence of haze in the molded product.

[0089] The learned model 40 for eliminating color mixing in a molded product is a model that, when the molding conditions and the color mixing area or presence / absence of color mixing obtained by inspecting the state of the molded product are input, outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the color mixing area or presence / absence of color mixing in the molded product.

[0090] The learned model 40 for eliminating specks in a molded product is a model that, when the molding conditions and the number of specks obtained by inspecting the state of the molded product are input, outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the number of specks in the molded product.

[0091] When evaluating the learned model 40 for eliminating surface dirt, any one of the injection pressure, screw temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature can be adopted as a defect-inducing condition.

[0092] The learned model 40 for eliminating surface dirt in a molded product is a model that, when the molding conditions and the area of the dirt portion or presence / absence of dirt obtained by inspecting the state of the molded product are input, outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the area of the dirt portion or presence / absence of dirt in the molded product.

[0093] For example, when evaluating the learned model 40 for eliminating silver streaks, any one of the cylinder temperature, screw rotation speed, injection speed, back pressure, and mold temperature can be adopted as a defect-inducing condition.

[0094] The learned model 40 for eliminating silver streaks in a molded product is as follows: when molding conditions and the presence or absence of silver streaks obtained by inspecting the state of the molded product are input, it outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the presence or absence of silver streaks in the molded product.

[0095] For example, when evaluating the learned model 40 for eliminating gate defects, any one of the injection pressure, holding pressure, and mold temperature can be adopted as a defect-inducing condition.

[0096] The learned model 40 for eliminating gate defects in a molded product is as follows: when molding conditions and the presence or absence of gate defects obtained by inspecting the state of the molded product are input, it outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the presence or absence of gate defects in the molded product.

[0097] For example, when evaluating the learned model 40 for eliminating warping and bending, any one of the injection pressure, barrel temperature, holding pressure, nozzle temperature, injection speed, back pressure, and mold temperature can be adopted as a defect-inducing condition.

[0098] The learned model 40 for eliminating warping or bending of a molded product is as follows: when molding conditions and the presence or absence of warping or bending obtained by inspecting the state of the molded product are input, it outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the presence or absence of warping or bending in the molded product.

[0099] For example, when evaluating the learned model 40 for eliminating cold slugs, any one of the barrel temperature, nozzle temperature, screw rotation speed, back pressure, and mold temperature can be adopted as a defect-inducing condition.

[0100] The learned model 40 for eliminating cold slugs in a molded product is as follows: when molding conditions and the presence or absence of cold slugs obtained by inspecting the state of the molded product are input, it outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the presence or absence of cold slugs in the molded product.

[0101] For example, when evaluating the learned model 40 for eliminating jetting marks, any one of the barrel temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature can be adopted as a defect-inducing condition. The learned model 40 for eliminating jetting marks in a molded product is as follows: when molding conditions and the presence or absence of jetting marks obtained by inspecting the state of the molded product are input, it outputs a correction amount for the specified molding conditions. The evaluation report of this learned model 40 includes the inspection results indicating the presence or absence of jetting marks in the molded product.

[0102] In addition, in the present embodiment, an example of performance evaluation of the learned model 40 for correcting the molding conditions of the modified injection molding machine 101 has been described. However, it may also be configured to perform performance evaluation of the learned model for the operating conditions of a modified extruder or other industrial machines.

[0103] A note is added to the solution for solving the problems of the present disclosure.

[0104] (Supplementary Note 1)

[0105] An automatic evaluation device, which is an automatic evaluation device for a learned model that corrects the operating conditions of an industrial machine, includes:

[0106] A reception unit that receives the upper limit value, lower limit value, and number of divisions of each of a plurality of defect-inducing conditions that induce defective product actions; and

[0107] A processing unit,

[0108] The processing unit performs the following processing:

[0109] By combining the values of each of the plurality of defect-inducing conditions obtained by dividing the numerical range between the upper limit value and the lower limit value by the number of divisions based on the upper limit value, lower limit value, and number of divisions of each of the plurality of defect-inducing conditions received by the reception unit, a plurality of defective conditions are generated;

[0110] The generated defective conditions are set in the industrial machine;

[0111] When the industrial machine set with the defective conditions is operated, the operation result when the operating conditions are corrected by the learned model is obtained;

[0112] The plurality of defective conditions are respectively set in the industrial machine, and an evaluation report of the learned model is generated based on the operation results obtained by operating the industrial machine multiple times for each of the defective conditions.

[0113] (Supplementary Note 2)

[0114] According to the automatic evaluation device described in Supplementary Note 1, wherein,

[0115] The evaluation report includes the plurality of generated defective conditions, the operating conditions corrected by the learned model, and the operation results obtained by operating the industrial machine multiple times for each of the defective conditions.

[0116] (Supplementary Note 3)

[0117] According to the automatic evaluation device described in Supplementary Note 1 or Supplementary Note 2, wherein,

[0118] The processing unit generates a plurality of defective conditions by sequentially changing the values of the plurality of defective induction conditions in ascending or descending order in a polling scheduling manner.

[0119] (Supplementary Note 4)

[0120] The automatic evaluation device according to any one of Supplementary Notes 1 to 3, wherein

[0121] The industrial machine is an injection molding machine.

[0122] (Supplementary Note 5)

[0123] The automatic evaluation device according to Supplementary Note 4, wherein

[0124] The defective induction conditions include the metering completion position of the molding material, the switching position of the injection speed, or the back pressure.

[0125] (Supplementary Note 6)

[0126] The automatic evaluation device according to Supplementary Note 4 or Supplementary Note 5, wherein

[0127] The operating conditions to be corrected include the injection speed, injection pressure, heater temperature, holding pressure, holding pressure switching position, mold temperature, nozzle temperature, clamping force, back pressure, holding time, or screw rotation.

[0128] (Supplementary Note 7)

[0129] The automatic evaluation device according to any one of Supplementary Notes 4 to 6, wherein

[0130] The operating result includes whether there are defects in the molded product or the degree of quality of the molded product, and the type of defect.

[0131] (Supplementary Note 8)

[0132] The automatic evaluation device according to any one of Supplementary Notes 4 to 7, wherein

[0133] The defective induction conditions include the metering completion position of the molding material, the switching position of the injection speed, and the back pressure,

[0134] The operating conditions to be corrected include the holding pressure, the holding pressure switching position, and the injection speed,

[0135] The operating result includes whether there are flash and short shots in the molded product, or the flash area and short shot area of the molded product.

[0136] (Supplementary Note 9)

[0137] The automatic evaluation device according to any one of Supplementary Notes 4 to 8, wherein

[0138] The defective inducing conditions and the operating conditions to be corrected include any one of injection pressure, cylinder temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature.

[0139] The operating results include the dent area, dent depth, or presence / absence of dents on the molded product, the weld line area or presence / absence of weld lines on the molded product, the flow mark area or presence / absence of flow marks on the molded product, or the crack area or presence / absence of cracks on the molded product.

[0140] (Supplementary Note 10)

[0141] The automatic evaluation device according to any one of Supplementary Notes 4 to 9, wherein

[0142] The defective inducing conditions and the operating conditions to be corrected include injection pressure, cylinder temperature, holding pressure, screw rotation speed, injection speed, or back pressure.

[0143] The operating results include the burn mark area or presence / absence of burn marks on the molded product, or the number of bubbles or voids in the molded product.

[0144] (Supplementary Note 11)

[0145] The automatic evaluation device according to any one of Supplementary Notes 4 to 10, wherein

[0146] The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, nozzle temperature, screw rotation speed, injection speed, back pressure, or mold temperature.

[0147] The operating results include the fogging area or presence / absence of fogging on the molded product, the color mixing area or presence / absence of color mixing on the molded product, or the number of spots on the molded product.

[0148] (Supplementary Note 12)

[0149] The automatic evaluation device according to any one of Supplementary Notes 4 to 11, wherein

[0150] The defective inducing conditions and the operating conditions to be corrected include injection pressure, screw temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, or mold temperature.

[0151] The operating results include the area of the soiled part or presence / absence of soiling on the molded product.

[0152] (Supplementary Note 13)

[0153] The automatic evaluation device according to any one of Supplementary Notes 4 to 12, wherein

[0154] The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, screw rotation speed, injection speed, back pressure, or mold temperature.

[0155] The operation result includes whether there are silver streaks on the molded product.

[0156] (Appendix Note 14)

[0157] The automatic evaluation device according to any one of Appendix Notes 4 to 13, wherein

[0158] The defect induction conditions and the operation conditions to be corrected include injection pressure, holding pressure or mold temperature.

[0159] The operation result includes whether there is a gate defect on the molded product.

[0160] (Appendix Note 15)

[0161] The automatic evaluation device according to any one of Appendix Notes 4 to 14, wherein

[0162] The defect induction conditions and the operation conditions to be corrected include injection pressure, cylinder temperature, holding pressure, nozzle temperature, injection speed, back pressure or mold temperature.

[0163] The operation result includes whether there is warping or bending of the molded product.

[0164] (Appendix Note 16)

[0165] The automatic evaluation device according to any one of Appendix Notes 4 to 15, wherein

[0166] The defect induction conditions and the operation conditions to be corrected include cylinder temperature, nozzle temperature, screw rotation speed, back pressure or mold temperature.

[0167] The operation result includes whether there is cold slug on the molded product.

[0168] (Appendix Note 17)

[0169] The automatic evaluation device according to any one of Appendix Notes 4 to 16, wherein

[0170] The defect induction conditions and the operation conditions to be corrected include cylinder temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure or mold temperature.

[0171] The operation result includes whether there is jet mark on the molded product.

[0172] Explanation of reference numerals

[0173] 1: Main body of molding machine

[0174] 1a: Sensor

[0175] 2: Mold clamping device

[0176] 3: Injection device

[0177] 4: Control device

[0178] 5: Inspection device

[0179] 6: Automatic evaluation device

[0180] 40: Learned model

[0181] 61: Processor

[0182] 62: Storage unit

[0183] 62a: Computer program

[0184] 62b: Evaluation report

[0185] 63: Receiving unit

[0186] 64: Data input / output unit

[0187] 65: Report output unit

[0188] 69: Recording medium

[0189] 101: Injection molding machine.

Claims

1. An automatic evaluation device, characterized in that, An automatic evaluation device for a learned model that corrects the operating conditions of industrial machinery The automatic evaluation device includes: A reception unit that receives the upper limit value, lower limit value, and number of divisions of each of a plurality of defect induction conditions that induce defective product actions; and A processing unit The processing unit performs the following processing: By combining the values of each of the plurality of defect induction conditions received by the reception unit based on the upper limit value, lower limit value, and number of divisions of each of the plurality of defect induction conditions, dividing the numerical range between the upper limit value and the lower limit value by the number of divisions, a plurality of defect conditions are generated; The generated defect conditions are set in the industrial machinery; When the industrial machinery set with the defect conditions is operated, the operation result when the operating conditions are corrected by the learned model is obtained; The plurality of defect conditions are respectively set in the industrial machinery, and an evaluation report of the learned model is generated based on the operation results obtained by operating the industrial machinery multiple times for each of the defect conditions.

2. The automatic evaluation device according to claim 1, wherein The evaluation report includes the plurality of generated defect conditions, the operating conditions corrected by the learned model, and the operation results obtained by operating the industrial machinery multiple times for each of the defect conditions.

3. The automatic evaluation device according to claim 1 or claim 2, wherein The processing unit generates a plurality of defect conditions by sequentially changing the values of the plurality of defect induction conditions in ascending or descending order in a round-robin scheduling manner.

4. The automatic evaluation device according to any one of claims 1 to 3, wherein The industrial machinery is an injection molding machine.

5. The automatic evaluation device according to claim 4, wherein The defect induction conditions include the metering completion position of the molding material, or the switching position of the injection speed, and the back pressure.

6. The automatic evaluation device according to claim 4 or claim 5, wherein The operating conditions to be corrected include injection speed, injection pressure, heater temperature, holding pressure, holding pressure switching position, mold temperature, nozzle temperature, clamping force, back pressure, holding time, or screw rotation.

7. The automatic evaluation device according to any one of claims 4 to 6, wherein The operation result includes the presence or absence of defects in the molded product or the degree of good or bad quality of the molded product, and the type of defect.

8. The automatic evaluation device according to any one of claims 4 to 7, wherein The defect induction conditions include the metering completion position of the molding material, the switching position of the injection speed, and the back pressure, The operating conditions to be corrected include holding pressure, holding pressure switching position, and injection speed, The operation result includes the presence or absence of flash and short shot in the molded product, or the flash area and short shot area of the molded product.

9. The automatic evaluation device according to any one of claims 4 to 8, wherein The defective inducing conditions and the operating conditions to be corrected include any one of injection pressure, cylinder temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, and mold temperature. The operating results include the dent area, dent depth, or presence / absence of dents on the molded product, the weld line area or presence / absence of weld lines on the molded product, the flow mark area or presence / absence of flow marks on the molded product, or the crack area or presence / absence of cracks on the molded product.

10. The automatic evaluation device according to any one of claims 4 to 9, characterized in that The defective inducing conditions and the operating conditions to be corrected include injection pressure, cylinder temperature, holding pressure, screw rotation speed, injection speed, or back pressure. The operating results include the burn mark area or presence / absence of burn marks on the molded product, or the number of bubbles or voids in the molded product.

11. The automatic evaluation device according to any one of claims 4 to 10, characterized in that The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, nozzle temperature, screw rotation speed, injection speed, back pressure, or mold temperature. The operating results include the haze area or presence / absence of haze on the molded product, the color mixing area or presence / absence of color mixing on the molded product, or the number of spots on the molded product.

12. The automatic evaluation device according to any one of claims 4 to 11, characterized in that The defective inducing conditions and the operating conditions to be corrected include injection pressure, screw temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, or mold temperature. The operating results include the area or presence / absence of dirt on the dirty part of the molded product.

13. The automatic evaluation device according to any one of claims 4 to 12, characterized in that The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, screw rotation speed, injection speed, back pressure, or mold temperature. The operating results include the presence / absence of silver streaks on the molded product.

14. The automatic evaluation device according to any one of claims 4 to 13, characterized in that The defective inducing conditions and the operating conditions to be corrected include injection pressure, holding pressure, or mold temperature. The operating results include the presence / absence of gate defects on the molded product.

15. The automatic evaluation device according to any one of claims 4 to 14, characterized in that The defective inducing conditions and the operating conditions to be corrected include injection pressure, cylinder temperature, holding pressure, nozzle temperature, injection speed, back pressure, or mold temperature. The operating results include the presence / absence of warping or bending of the molded product.

16. The automatic evaluation device according to any one of claims 4 to 15, characterized in that The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, nozzle temperature, screw rotation speed, back pressure, or mold temperature. The operating results include the presence / absence of cold slugs on the molded product.

17. The automatic evaluation device according to any one of claims 4 to 16, characterized in that The defective inducing conditions and the operating conditions to be corrected include cylinder temperature, holding pressure, nozzle temperature, screw rotation speed, injection speed, back pressure, or mold temperature. The operation result includes whether there are ejection marks on the molded product.

18. An automatic evaluation method, characterized in that, It is an automatic evaluation method for a learned model that corrects the operating conditions of industrial machinery. In the automatic evaluation method, the computer performs the following processing: Accept the upper limit value, lower limit value, and number of divisions of each of the multiple defect-inducing conditions that induce defective product operations. Generate multiple defect conditions by combining the values of each defect-inducing condition obtained by dividing the numerical range between the upper limit value and the lower limit value based on the upper limit value, lower limit value, and number of divisions of each of the multiple defect-inducing conditions accepted. Set the generated defect conditions in the industrial machinery. When operating the industrial machinery set with the defect conditions, obtain the operation result when the operating conditions are corrected by the learned model. Set the multiple defect conditions in the industrial machinery respectively, and generate an evaluation report of the learned model based on the operation results obtained by operating the industrial machinery multiple times for each defect condition.

19. A computer program, characterized in that, Cause the computer to evaluate a learned model that corrects the operating conditions of industrial machinery. The computer program causes the computer to perform the following processing: Accept the upper limit value, lower limit value, and number of divisions of each of the multiple defect-inducing conditions that induce defective product operations. Generate multiple defect conditions by combining the values of each defect-inducing condition obtained by dividing the numerical range between the upper limit value and the lower limit value based on the upper limit value, lower limit value, and number of divisions of each of the multiple defect-inducing conditions accepted. Set the generated defect conditions in the industrial machinery. When operating the industrial machinery set with the defect conditions, obtain the operation result when the operating conditions are corrected by the learned model. Set the multiple defect conditions in the industrial machinery respectively, and generate an evaluation report of the learned model based on the operation results obtained by operating the industrial machinery multiple times for each defect condition.

Images