Monitoring device of injection molding machine

Abstract

A monitoring device of an injection molding machine can make abnormality determination flexible. A monitoring device for an injection molding machine according to an embodiment includes: an acquisition unit configured to acquire first data indicating a detection result from a detection unit provided in the injection molding machine in a time sequence; and a reasoning unit which inputs the first data into a learning completion model for learning training data in which second data indicating a detection result from the detection unit in a time sequence and a second value indicating an evaluation of a product produced by the injection molding machine are combined, obtaining a first value indicating the evaluation of the product from the learned model; and a determination unit that determines, on the basis of the first value acquired by the reasoning unit, whether or not an abnormality has occurred in the detection result of the detection unit.

Description

[0001] This application claims priority based on Japanese Patent Application No. 2023-186605 filed on October 31, 2023. The entire contents of the Japanese Patent Application are incorporated herein by reference. Technical Field

[0002] The invention relates to a monitoring device for an injection molding machine. Background Art

[0003] Conventionally, there is provided a monitoring function for confirming whether a product is properly produced by an injection molding machine. As a monitoring function between molding shots, there is a function of inputting a time into waveform data representing the detection result of a sensor provided on the injection molding machine in a time series, monitoring the detection result of the sensor at that time, or monitoring a value derived by calculating any one or more of the maximum value, minimum value, average value, and integral on the waveform data. Both require a setting for determining whether a product is properly produced while a person visually confirms the waveform data.

[0004] On the other hand, artificial intelligence tends to develop with the improvement of computer processing power. For example, Patent Document 1 proposes a technology for estimating whether a produced product is a good product through machine learning.

[0005] Patent Document 1: Japanese Patent Application Publication No. 2020-075385

[0006] However, in the technology described in Patent Document 1, in order to estimate whether a product is a good product, it is necessary to re-perform machine learning when the judgment criteria are changed. In other words, it is difficult to make product evaluation flexible. Summary of the invention

[0007] One embodiment of the present invention provides a technique that enables flexibility in abnormality determination.

[0008] A monitoring device for an injection molding machine according to one embodiment of the present invention comprises: an acquisition unit, which acquires first data representing in time series a detection result from a detection unit provided in the injection molding machine; an inference unit, which acquires a first value representing an evaluation of the product from the learned model by inputting the first data into a machine learning model of training data that combines second data representing in time series a detection result from the detection unit and a second value representing an evaluation of a product produced by the injection molding machine; and a determination unit, which determines whether an abnormality has occurred in the detection result of the detection unit based on the first value acquired by the inference unit.

[0009] Effects of the Invention

[0010] According to one aspect of the present invention, the accuracy of abnormality determination is improved by making abnormality determination flexible. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 This is a diagram showing a state of the injection molding machine according to one embodiment when mold opening is completed.

[0012] Figure 2 It is a diagram showing a state of the injection molding machine according to one embodiment when the mold is clamped.

[0013] Figure 3 It is a diagram showing an example of the functional configuration of the control device according to the first embodiment.

[0014] Figure 4 FIG. 1 is a diagram illustrating a configuration of a learned model used in the control device according to the first embodiment.

[0015] Figure 5 It is a schematic diagram showing the structure of the learned model according to the first embodiment.

[0016] Figure 6 It is a diagram illustrating a log information screen displayed by the display control unit according to the first embodiment.

[0017] Figure 7 It is a diagram illustrating a monitoring setting screen displayed by the display control unit according to the first embodiment.

[0018] Figure 8 It is a diagram illustrating an example of an operation setting screen displayed by the display control unit during failure detection according to the first embodiment.

[0019] Fig. 9 This is a diagram illustrating quality characteristics output from the learned model according to the first embodiment.

[0020] Fig.10 This is a diagram illustrating waveform data for each (shot) number detected when a molded product is molded by the injection molding machine according to the first embodiment.

[0021] Fig.11 This is a diagram illustrating waveform data for each (shot) number detected when a molded product is molded by the injection molding machine according to the first embodiment.

[0022] Fig.12 This is a schematic diagram showing the cooperation related to the learned model by the learning device and the control device of the injection molding machine according to the second embodiment.

[0023] Fig.13It is a schematic diagram showing the structure of a learned model according to the second embodiment.

[0024] Fig.14 This is a diagram showing an example of the functional configuration of a learning device and a control device according to the second embodiment.

[0025] In the figure: 10-injection molding machine, 700, 700A-control device, 701-CPU, 702-storage medium, 705-communication interface, 711-acquisition unit, 712-training data generation unit, 713-learning unit, 714-injection molding control unit, 715, 715A-determination unit, 716, 716A-inference unit, 717-display control unit, 718-communication control unit, 719-update unit, 721-training data storage unit, 722, 722A-learning completed model storage unit, 760-display device, 1300-learning device, 1301-CPU, 1302-storage medium, 1303-communication interface, 1311-acquisition unit, 1312-training data generation unit, 1313-learning unit, 1314-communication control unit, 1321-training data storage unit, 1322-learning completed model storage unit. DETAILED DESCRIPTION

[0026] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Furthermore, the embodiments described below are not intended to limit the embodiments of the invention but are intended to be illustrative, and all features and combinations thereof described in the embodiments are not necessarily the essence of the invention. In addition, in each of the accompanying drawings, the same or corresponding structures are sometimes marked with the same or corresponding symbols, and the description is omitted.

[0027] Figure 1 It is a diagram showing a state of the injection molding machine according to the first embodiment when mold opening is completed. Figure 2 1 is a diagram showing the state of the injection molding machine involved in the first embodiment when the mold is clamped. In this specification, the X-axis direction, the Y-axis direction and the Z-axis direction are directions perpendicular to each other. The X-axis direction and the Y-axis direction represent horizontal directions, and the Z-axis direction represents a vertical direction. When the mold clamping device 100 is horizontal, the X-axis direction is the mold opening and closing direction, and the Y-axis direction is the width direction of the injection molding machine 10. The negative side of the Y-axis direction is called the operating side, and the positive side of the Y-axis direction is called the opposite side of the operating side.

[0028] like Figure 1-2As shown, the injection molding machine 10 includes: a mold clamping device 100 for opening and closing a mold device 800; an ejection device 200 for ejecting a molded product molded by the mold device 800; an injection device 300 for injecting a molding material into the mold device 800; a moving device 400 for moving the injection device 300 forward and backward relative to the mold device 800; a control device 700 for controlling each component of the injection molding machine 10; and a frame 900 for supporting each component of the injection molding machine 10. The frame 900 includes a mold clamping device frame 910 for supporting the mold clamping device 100 and an injection device frame 920 for supporting the injection device 300. The mold clamping device frame 910 and the injection device frame 920 are respectively installed on the base plate 2 via leveling casters 930. The control device 700 is arranged in the internal space of the injection device frame 920. The components of the injection molding machine 10 are described below.

[0029] (Mold clamping device)

[0030] In the description of the mold clamping device 100, the moving direction of the movable platen 120 when the mold is closed (for example, the positive direction of the X axis) is set to the front, and the moving direction of the movable platen 120 when the mold is opened (for example, the negative direction of the X axis) is set to the rear.

[0031] The mold clamping device 100 performs mold closing, pressurization, mold clamping, depressurization, and mold opening of the mold device 800. The mold device 800 includes a fixed mold 810 and a movable mold 820. The mold clamping device 100 is, for example, horizontal, and the mold opening and closing direction is the horizontal direction. The mold clamping device 100 has a fixed platen 110 on which the fixed mold 810 is mounted, a movable platen 120 on which the movable mold 820 is mounted, and a moving mechanism 102 that moves the movable platen 120 relative to the fixed platen 110 in the mold opening and closing direction.

[0032] The fixed platen 110 is fixed to the mold clamping device frame 910. The fixed mold 810 is attached to the surface of the fixed platen 110 that faces the movable platen 120.

[0033] The movable platen 120 is arranged to be movable in the mold opening and closing direction relative to the mold clamping device frame 910. A guide 101 for guiding the movable platen 120 is laid on the mold clamping device frame 910. The movable mold 820 is attached to the surface of the movable platen 120 facing the fixed platen 110.

[0034] The moving mechanism 102 performs mold closing, pressurization, mold clamping, depressurization, and mold opening of the mold device 800 by moving the movable platen 120 forward and backward relative to the fixed platen 110. The moving mechanism 102 includes a toggle seat 130 disposed with a gap between the fixed platen 110, a connecting rod 140 connecting the fixed platen 110 and the toggle seat 130, a toggle mechanism 150 that moves the movable platen 120 relative to the toggle seat 130 in the mold opening and closing direction, a mold clamping motor 160 that operates the toggle mechanism 150, a motion conversion mechanism 170 that converts the rotational motion of the mold clamping motor 160 into a linear motion, and a mold thickness adjustment mechanism 180 that adjusts the gap between the fixed platen 110 and the toggle seat 130.

[0035] The toggle seat 130 is arranged with a gap between the fixed platen 110 and is placed on the mold clamping device frame 910 so as to be movable in the mold opening and closing direction. In addition, the toggle seat 130 can be arranged so as to be movable along a guide member laid on the mold clamping device frame 910. The guide member of the toggle seat 130 can be common to the guide member 101 of the movable platen 120.

[0036] In addition, in the present embodiment, the fixed pressure plate 110 is fixed relative to the mold clamping device frame 910, and the elbow seat 130 is configured to be freely movable relative to the mold clamping device frame 910 in the mold opening and closing direction, but the elbow seat 130 may also be fixed relative to the mold clamping device frame 910, and the fixed pressure plate 110 may be configured to be freely movable relative to the mold clamping device frame 910 in the mold opening and closing direction.

[0037] The connecting rod 140 connects the fixed platen 110 and the toggle seat 130 at a distance L in the mold opening and closing direction. A plurality of connecting rods 140 (e.g., four) may be used. The plurality of connecting rods 140 are arranged parallel to the mold opening and closing direction and extend according to the mold clamping force. A connecting rod strain detector 141 for detecting the strain of the connecting rod 140 may be provided on at least one of the connecting rods 140. The connecting rod strain detector 141 sends a signal indicating its detection result to the control device 700. The detection result of the connecting rod strain detector 141 is used for detecting the mold clamping force, etc.

[0038] In addition, in this embodiment, the tie rod strain detector 141 is used as the mold clamping force detector for detecting the mold clamping force, but the present invention is not limited to this. The mold clamping force detector is not limited to the strain type, and can also be a piezoelectric type, a capacitive type, a hydraulic type, an electromagnetic type, etc., and its installation position is not limited to the tie rod 140.

[0039] The toggle mechanism 150 is disposed between the movable platen 120 and the toggle seat 130, and moves the movable platen 120 in the mold opening and closing direction relative to the toggle seat 130. The toggle mechanism 150 includes a crosshead 151 that moves in the mold opening and closing direction and a pair of link groups that are bent and extended by the movement of the crosshead 151. The pair of link groups each includes a first link 152 and a second link 153 that are connected by a pin or the like so as to be able to bend and extend freely. The first link 152 is mounted by a pin or the like so as to be able to swing freely relative to the movable platen 120. The second link 153 is mounted by a pin or the like so as to be able to swing freely relative to the toggle seat 130. The second link 153 is mounted to the crosshead 151 via a third link 154. When the crosshead 151 is moved forward and backward relative to the toggle seat 130, the first link 152 and the second link 153 are bent and extended so that the movable platen 120 is moved forward and backward relative to the toggle seat 130.

[0040] In addition, the structure of the toggle mechanism 150 is not limited to Figure 1 and Figure 2 For example, in Figure 1 and Figure 2 In the embodiment, the number of nodes of each link group is 5, but it may be 4, or one end of the third link 154 may be connected to the node between the first link 152 and the second link 153.

[0041] The mold clamping motor 160 is mounted on the toggle base 130 and operates the toggle mechanism 150. The mold clamping motor 160 advances and retreats the crosshead 151 relative to the toggle base 130, thereby bending and extending the first link 152 and the second link 153, thereby advancing and retreating the movable platen 120 relative to the toggle base 130. The mold clamping motor 160 is directly connected to the motion conversion mechanism 170, but may be connected to the motion conversion mechanism 170 via a belt, a pulley, or the like.

[0042] The motion conversion mechanism 170 converts the rotational motion of the mold clamping motor 160 into the linear motion of the cross head 151. The motion conversion mechanism 170 includes a screw shaft and a screw nut screwed with the screw shaft. Balls or rollers may be interposed between the screw shaft and the screw nut.

[0043] The mold clamping device 100 performs a mold closing process, a pressure increasing process, a mold clamping process, a pressure releasing process, a mold opening process, and the like under the control of the control device 700 .

[0044] In the mold closing process, the mold clamping motor 160 is driven to move the crosshead 151 to the mold closing end position at a set moving speed, and the movable platen 120 is moved forward to make the movable mold 820 contact the fixed mold 810. For example, the position and moving speed of the crosshead 151 are detected using a mold clamping motor encoder 161. The mold clamping motor encoder 161 detects the rotation of the mold clamping motor 160 and sends a signal indicating the detection result to the control device 700.

[0045] In addition, the crosshead position detector for detecting the position of the crosshead 151 and the crosshead moving speed detector for detecting the moving speed of the crosshead 151 are not limited to the mold clamping motor encoder 161, and a conventional detector can be used. In addition, the movable platen position detector for detecting the position of the movable platen 120 and the movable platen moving speed detector for detecting the moving speed of the movable platen 120 are not limited to the mold clamping motor encoder 161, and a conventional detector can be used.

[0046] In the pressure increasing step, the mold clamping motor 160 is further driven to further advance the cross head 151 from the mold closing end position to the mold clamping position, thereby generating a mold clamping force.

[0047] In the mold closing process, the mold closing motor 160 is driven to maintain the position of the cross head 151 at the mold closing position. In the mold closing process, the mold closing force generated in the pressure increasing process is maintained. In the mold closing process, a cavity space 801 is formed between the movable mold 820 and the fixed mold 810 (refer to Figure 2 ), the injection device 300 fills the cavity space 801 with liquid molding material. The filled molding material is solidified, thereby obtaining a molded product.

[0048] The number of cavity spaces 801 may be one or more. In the latter case, multiple molded products can be obtained at the same time. An insert may be arranged in a part of the cavity space 801, and a molding material may be filled in another part of the cavity space 801. A molded product in which the insert and the molding material are integrated can be obtained.

[0049] In the depressurization process, the crosshead 151 is retracted from the mold clamping position to the mold opening start position by driving the mold clamping motor 160, so that the movable platen 120 is retracted to reduce the mold clamping force. The mold opening start position and the mold closing end position may be the same position.

[0050] In the mold opening process, the mold clamping motor 160 is driven to move the crosshead 151 backward from the mold opening start position to the mold opening completion position at a set moving speed, and the movable platen 120 is moved backward to separate the movable mold 820 from the fixed mold 810. Then, the ejector device 200 ejects the molded product from the movable mold 820.

[0051] The setting conditions in the mold closing process, the pressure increasing process and the mold clamping process are uniformly set as a series of setting conditions. For example, the moving speed, position (including the mold closing start position, the moving speed switching position, the mold closing end position and the mold clamping position) and the mold clamping force of the crosshead 151 in the mold closing process and the pressure increasing process are uniformly set as a series of setting conditions. The mold closing start position, the moving speed switching position, the mold closing end position and the mold clamping position are arranged in sequence from the back to the front, and indicate the starting point and end point of the interval for setting the moving speed. The moving speed is set for each interval. The moving speed switching position can be one or more. The moving speed switching position can also be not set. Only one of the mold clamping position and the mold clamping force can be set.

[0052] The setting conditions in the pressure release process and the mold opening process are also set in the same way. For example, the moving speed and position (mold opening start position, moving speed switching position and mold opening completion position) of the crosshead 151 in the pressure release process and the mold opening process are uniformly set as a series of setting conditions. The mold opening start position, the moving speed switching position and the mold opening completion position are arranged in sequence from the front to the back, and represent the starting point and end point of the interval for setting the moving speed. The moving speed is set for each interval. The moving speed switching position can be one or more. The moving speed switching position can also be not set. The mold opening start position and the mold closing end position can be the same position. In addition, the mold opening completion position and the mold closing start position can be the same position.

[0053] In addition, the moving speed and position of the movable platen 120 may be set instead of the moving speed and position of the crosshead 151. Furthermore, the mold clamping force may be set instead of the position of the crosshead (eg, mold clamping position) and the position of the movable platen.

[0054] However, the toggle mechanism 150 amplifies the driving force of the mold clamping motor 160 and transmits it to the movable platen 120. The amplification ratio is also called the toggle magnification. The toggle magnification changes according to the angle θ (hereinafter also referred to as "link angle θ") formed by the first link 152 and the second link 153. The link angle θ is obtained from the position of the crosshead 151. When the link angle θ is 180°, the toggle magnification becomes the maximum.

[0055] When the thickness of the mold device 800 changes due to replacement of the mold device 800, temperature change of the mold device 800, etc., mold thickness adjustment is performed to obtain a predetermined mold clamping force when the mold is clamped. In the mold thickness adjustment, for example, the interval L between the fixed platen 110 and the toggle seat 130 is adjusted so that the link angle θ of the toggle mechanism 150 becomes a predetermined angle at the time when the movable mold 820 contacts the fixed mold 810.

[0056] The mold clamping device 100 has a mold thickness adjustment mechanism 180. The mold thickness adjustment mechanism 180 adjusts the interval L between the fixed platen 110 and the toggle seat 130, thereby adjusting the mold thickness. In addition, the mold thickness adjustment is performed, for example, during the period from the end of the molding cycle to the start of the next molding cycle. The mold thickness adjustment mechanism 180 has, for example: a screw shaft 181 formed at the rear end of the connecting rod 140; a screw nut 182 that is held in the toggle seat 130 so as to be rotatable and unable to move forward and backward; and a mold thickness adjustment motor 183 that rotates the screw nut 182 that is threaded with the screw shaft 181.

[0057] A lead screw shaft 181 and a lead screw nut 182 are provided for each connecting rod 140. The rotational driving force of the mold thickness adjustment motor 183 can be transmitted to the plurality of lead screw nuts 182 via the rotational driving force transmission unit 185. The plurality of lead screw nuts 182 can be rotated synchronously. In addition, by changing the transmission path of the rotational driving force transmission unit 185, the plurality of lead screw nuts 182 can also be rotated individually.

[0058] The rotational driving force transmission part 185 is composed of, for example, gears. In this case, a driven gear is formed on the outer periphery of each lead screw nut 182, a driving gear is mounted on the output shaft of the mold thickness adjustment motor 183, and an intermediate gear meshing with a plurality of driven gears and the driving gear is held rotatably at the center of the toggle seat 130. In addition, the rotational driving force transmission part 185 may be composed of a belt, a pulley, etc. instead of a gear.

[0059] The operation of the mold thickness adjustment mechanism 180 is controlled by the control device 700. The control device 700 drives the mold thickness adjustment motor 183 to rotate the lead screw nut 182. As a result, the position of the toggle seat 130 relative to the connecting rod 140 is adjusted, and the interval L between the fixed platen 110 and the toggle seat 130 is adjusted. In addition, a plurality of mold thickness adjustment mechanisms may be used in combination.

[0060] The mold thickness adjustment motor encoder 184 is used to detect the interval L. The mold thickness adjustment motor encoder 184 detects the rotation amount and rotation direction of the mold thickness adjustment motor 183, and sends a signal indicating the detection result to the control device 700. The detection result of the mold thickness adjustment motor encoder 184 is used to monitor and control the position and interval L of the toggle seat 130. In addition, the toggle seat position detector for detecting the position of the toggle seat 130 and the interval detector for detecting the interval L are not limited to the mold thickness adjustment motor encoder 184, and a conventional detector can be used.

[0061] The mold clamping device 100 may include a mold temperature regulator for regulating the temperature of the mold device 800. The mold device 800 includes a flow path of a temperature regulating medium therein. The mold temperature regulator regulates the temperature of the temperature regulating medium supplied to the flow path of the mold device 800, thereby regulating the temperature of the mold device 800.

[0062] Furthermore, the mold clamping device 100 of the present embodiment is a horizontal type in which the mold opening and closing direction is the horizontal direction, but may be a vertical type in which the mold opening and closing direction is the vertical direction.

[0063] The mold clamping device 100 of this embodiment includes the mold clamping motor 160 as a driving source, but may include a hydraulic cylinder instead of the mold clamping motor 160. Furthermore, the mold clamping device 100 may include a mold opening and closing linear motor or a mold clamping electromagnet.

[0064] (Ejector device)

[0065] In the description of the ejection device 200, similar to the description of the mold clamping device 100, etc., the moving direction of the movable platen 120 when the mold is closed (for example, the positive direction of the X-axis) is set to the front, and the moving direction of the movable platen 120 when the mold is opened (for example, the negative direction of the X-axis) is set to the rear for description.

[0066] The ejector device 200 is mounted on the movable platen 120 and moves forward and backward together with the movable platen 120. The ejector device 200 includes an ejector rod 210 for ejecting the molded product from the mold device 800 and a drive mechanism 220 for moving the ejector rod 210 along the moving direction of the movable platen 120 (X-axis direction).

[0067] The ejector rod 210 is arranged to be able to move forward and backward in the through hole of the movable platen 120. The front end of the ejector rod 210 contacts the ejector plate 826 of the movable mold 820. The front end of the ejector rod 210 may be connected to the ejector plate 826 or may not be connected thereto.

[0068] The driving mechanism 220 includes, for example, an ejector motor and a motion conversion mechanism that converts the rotational motion of the ejector motor into the linear motion of the ejector rod 210. The motion conversion mechanism includes a screw shaft and a screw nut screwed with the screw shaft. Balls or rollers may be interposed between the screw shaft and the screw nut.

[0069] The ejector device 200 performs an ejection process under the control of the control device 700. In the ejection process, the ejector rod 210 is moved forward from the standby position to the ejection position at a set moving speed, and the ejector plate 826 is moved forward to eject the molded product. Then, the ejector motor is driven to move the ejector rod 210 backward at a set moving speed, and the ejector plate 826 is moved backward to the original standby position.

[0070] For example, an ejector motor encoder is used to detect the position and moving speed of the ejector rod 210. The ejector motor encoder detects the rotation of the ejector motor and sends a signal indicating the detection result to the control device 700. In addition, the ejector rod position detector for detecting the position of the ejector rod 210 and the ejector rod moving speed detector for detecting the moving speed of the ejector rod 210 are not limited to the ejector motor encoder, and a conventional detector can be used.

[0071] (Injection device)

[0072] In the description of the injection device 300, unlike the description of the mold clamping device 100 and the description of the ejection device 200, the moving direction of the screw 330 during filling (for example, the negative direction of the X-axis) is set to the front, and the moving direction of the screw 330 during metering (for example, the positive direction of the X-axis) is set to the rear for description.

[0073] The injection device 300 is provided on a sliding base 301, and the sliding base 301 is configured to be able to move forward and backward freely relative to the injection device frame 920. The injection device 300 is configured to be able to move forward and backward freely relative to the mold device 800. The injection device 300 contacts the mold device 800, and fills the molding material measured in the cylinder 310 into the cavity space 801 in the mold device 800. The injection device 300 includes, for example, a cylinder 310 for heating the molding material, a nozzle 320 provided at the front end of the cylinder 310, a screw 330 configured to be able to move forward and backward freely and to rotate freely in the cylinder 310, a metering motor 340 for rotating the screw 330, an injection motor 350 for moving the screw 330 forward and backward, and a load detector 360 for detecting the load transmitted between the injection motor 350 and the screw 330.

[0074] The cylinder 310 heats the molding material supplied from the supply port 311. The molding material includes, for example, resin. The molding material is formed into, for example, granules and supplied to the supply port 311 in a solid state. The supply port 311 is formed at the rear of the cylinder 310. A cooler 312 such as a water-cooled cylinder is provided on the periphery of the rear of the cylinder 310. A heater 313 such as a belt heater and a temperature detector 314 are provided on the periphery of the cylinder 310 in front of the cooler 312.

[0075] The cylinder 310 is divided into a plurality of regions along the axial direction (e.g., the X-axis direction) of the cylinder 310. Heaters 313 and temperature detectors 314 are respectively provided in the plurality of regions. Set temperatures are respectively set for the plurality of regions, and the control device 700 controls the heater 313 so that the detected temperature of the temperature detector 314 becomes the set temperature.

[0076] The nozzle 320 is provided at the front end of the cylinder 310 and presses the mold device 800. A heater 313 and a temperature detector 314 are provided on the outer periphery of the nozzle 320. The control device 700 controls the heater 313 so that the detected temperature of the nozzle 320 becomes a set temperature.

[0077] The screw 330 is configured to be rotatable and to be able to move forward and backward freely in the cylinder 310. When the screw 330 is rotated, the molding material is transported forward along the spiral groove of the screw 330. While being transported forward, the molding material is gradually melted by the heat from the cylinder 310. As the liquid molding material is transported forward to the front of the screw 330 and accumulated in the front part of the cylinder 310, the screw 330 is retracted. Then, when the screw 330 is advanced, the liquid molding material accumulated in front of the screw 330 is injected from the nozzle 320 and filled in the mold device 800.

[0078] The check ring 331 is installed at the front of the screw 330 so as to be able to move forward and backward. The check ring 331 acts as a check valve to prevent the molding material from flowing backward from the front to the rear of the screw 330 when the screw 330 is pushed forward.

[0079] When the screw 330 is advanced, the check ring 331 is pushed backward by the pressure of the molding material in front of the screw 330, and moves back relative to the screw 330 to a closed position that blocks the flow path of the molding material (see FIG. Figure 2 ). Thus, the molding material accumulated in front of the screw 330 is prevented from flowing backward.

[0080] On the other hand, when the screw 330 is rotated, the check ring 331 is pushed forward by the pressure of the molding material conveyed to the front along the spiral groove of the screw 330, and moves forward relative to the screw 330 to an open position (refer to Figure 1 ). As a result, the molding material is transported to the front of the screw 330.

[0081] The check ring 331 may be a co-rotating type that rotates together with the screw 330 or a non-co-rotating type that does not rotate together with the screw 330 .

[0082] In addition, the injection device 300 may include a driving source that moves the check ring 331 forward and backward relative to the screw 330 between the open position and the closed position.

[0083] The metering motor 340 rotates the screw 330. The driving source for rotating the screw 330 is not limited to the metering motor 340, and may be, for example, a hydraulic pump or the like.

[0084] The injection motor 350 moves the screw 330 forward and backward. A motion conversion mechanism that converts the rotational motion of the injection motor 350 into the linear motion of the screw 330 is provided between the injection motor 350 and the screw 330. The motion conversion mechanism, for example, includes a screw shaft and a screw nut screwed with the screw shaft. Balls, rollers, etc. may be provided between the screw shaft and the screw nut. The driving source that moves the screw 330 forward and backward is not limited to the injection motor 350, and may be, for example, a hydraulic cylinder.

[0085] The load detector 360 detects the load transmitted between the injection motor 350 and the screw 330. The detected load is converted into pressure by the control device 700. The load detector 360 is provided in the transmission path of the load between the injection motor 350 and the screw 330, and detects the load acting on the load detector 360.

[0086] The load detector 360 sends a signal of the detected load to the control device 700. The load detected by the load detector 360 is converted into a pressure acting between the screw 330 and the molding material, and is used to control and monitor the pressure on the screw 330 from the molding material, the back pressure on the screw 330, and the pressure acting on the molding material from the screw 330.

[0087] In addition, the pressure detector for detecting the pressure of the molding material is not limited to the load detector 360 , and a conventional detector can be used. For example, a nozzle pressure sensor or a mold internal pressure sensor can be used. The nozzle pressure sensor is provided at the nozzle 320 .

[0088] The injection device 300 performs a metering process, a filling process, a pressure-maintaining process, etc. under the control of the control device 700. The filling process and the pressure-maintaining process can be collectively referred to as an injection process.

[0089] In the metering process, the metering motor 340 is driven to rotate the screw 330 at a set speed, and the molding material is transported to the front along the spiral groove of the screw 330. As a result, the molding material is gradually melted. As the liquid molding material is transported to the front of the screw 330 and accumulated in the front part of the cylinder 310, the screw 330 is moved backward. For example, the metering motor encoder 341 is used to detect the rotation speed of the screw 330. The metering motor encoder 341 detects the rotation of the metering motor 340 and sends a signal indicating the detection result to the control device 700. In addition, the screw speed detector that detects the rotation speed of the screw 330 is not limited to the metering motor encoder 341, and a conventional detector can be used.

[0090] In order to limit the rapid retreat of the screw 330 during the metering process, the injection motor 350 may be driven to apply a set back pressure to the screw 330. For example, the back pressure on the screw 330 may be detected using a load detector 360. When the screw 330 retreats to the metering end position and a predetermined amount of molding material is accumulated in front of the screw 330, the metering process is completed.

[0091] The position and speed of the screw 330 in the metering process are uniformly set as a series of setting conditions. For example, the metering start position, the speed switching position and the metering end position are set. These positions are arranged in sequence from the front to the back, and represent the starting point and the end point of the interval for setting the speed. The speed is set for each interval. The speed switching position can be one or more. The speed switching position can also be not set. In addition, the back pressure can be set for each interval.

[0092] In the filling process, the injection motor 350 is driven to make the screw 330 move forward at a set moving speed, and the liquid molding material accumulated in front of the screw 330 is filled into the cavity space 801 in the mold device 800. For example, the injection motor encoder 351 is used to detect the position and moving speed of the screw 330. The injection motor encoder 351 detects the rotation of the injection motor 350 and sends a signal indicating its detection result to the control device 700. If the position of the screw 330 reaches the set position, the switch from the filling process to the pressure holding process (the so-called V / P switch) is performed. The position where the V / P switch is performed is also referred to as the V / P switching position. The set moving speed of the screw 330 can be changed according to the position of the screw 330, time, etc.

[0093] The position and moving speed of the screw 330 in the filling process are uniformly set as a series of setting conditions. For example, the filling start position (also called the "injection start position"), the moving speed switching position and the V / P switching position are set. These positions are arranged in sequence from the back to the front, and represent the starting point and the end point of the interval for setting the moving speed. The moving speed is set for each interval. There can be one or more moving speed switching positions. The moving speed switching position can also be not set.

[0094] The upper limit value of the pressure of the screw 330 is set for each interval of the moving speed of the screw 330. The pressure of the screw 330 is detected by the load detector 360. When the pressure of the screw 330 is below the set pressure, the screw 330 moves forward at the set moving speed. On the other hand, when the pressure of the screw 330 exceeds the set pressure, the screw 330 moves forward at a moving speed slower than the set moving speed for the purpose of protecting the mold so that the pressure of the screw 330 becomes below the set pressure.

[0095] In addition, in the filling process, after the position of the screw 330 reaches the V / P switching position, the screw 330 can be paused at the V / P switching position and then the V / P switching can be performed. Instead of stopping the screw 330 before the V / P switching is to be performed, the screw 330 can be moved forward or backward at a slow speed. In addition, the screw position detector for detecting the position of the screw 330 and the screw moving speed detector for detecting the moving speed of the screw 330 are not limited to the injection motor encoder 351, and a conventional detector can be used.

[0096] In the pressure holding process, the injection motor 350 is driven to push the screw 330 forward, and the pressure of the molding material at the front end of the screw 330 (hereinafter also referred to as "holding pressure") is maintained at the set pressure, and the molding material remaining in the cylinder 310 is pushed toward the mold device 800. The insufficient amount of molding material in the mold device 800 due to cooling shrinkage can be supplemented. For example, the holding pressure is detected using a load detector 360. The setting value of the holding pressure can be changed according to the elapsed time since the start of the pressure holding process. The holding pressure and the holding time of the holding pressure in multiple pressure holding processes can be set separately, or they can be set uniformly as a series of setting conditions.

[0097] In the pressure holding process, the molding material in the cavity space 801 in the mold device 800 is gradually cooled, and at the end of the pressure holding process, the entrance of the cavity space 801 is blocked by the solidified molding material. This state is called gate sealing, which prevents the backflow of the molding material from the cavity space 801. After the pressure holding process, the cooling process begins. In the cooling process, the molding material in the cavity space 801 is solidified. In order to shorten the molding cycle time, a metering process can be performed in the cooling process.

[0098] In addition, the injection device 300 of this embodiment is a coaxial screw type, but it can also be a pre-plastic type. The injection device of the pre-plastic type supplies the molding material melted in the plasticizing cylinder to the injection cylinder, and injects the molding material into the mold device from the injection cylinder. In the plasticizing cylinder, the screw is configured to be able to rotate freely and cannot move forward or backward, or the screw is configured to be able to rotate freely and move forward and backward freely. On the other hand, in the injection cylinder, the plunger is configured to move forward and backward freely.

[0099] Furthermore, the injection device 300 of the present embodiment is a horizontal type in which the axial direction of the cylinder 310 is in the horizontal direction, but it may be a vertical type in which the axial direction of the cylinder 310 is in the up-down direction. The mold clamping device combined with the vertical injection device 300 may be a vertical type or a horizontal type. Similarly, the mold clamping device combined with the horizontal injection device 300 may be a horizontal type or a vertical type.

[0100] (Mobile device)

[0101] In the description of the moving device 400, similar to the description of the injection device 300, the moving direction of the screw 330 during filling (e.g., the negative direction of the X axis) is set to the front, and the moving direction of the screw 330 during metering (e.g., the positive direction of the X axis) is set to the rear.

[0102] The moving device 400 moves the injection device 300 forward and backward relative to the mold device 800. The moving device 400 also presses the nozzle 320 relative to the mold device 800 to generate nozzle contact pressure. The moving device 400 includes a hydraulic pump 410, a motor 420 as a driving source, and a hydraulic cylinder 430 as a hydraulic actuator.

[0103] The hydraulic pump 410 has a first port 411 and a second port 412. The hydraulic pump 410 is a bidirectionally rotatable pump, and generates hydraulic pressure by sucking working fluid (e.g., oil) from one of the first port 411 and the second port 412 and discharging it from the other port by switching the rotation direction of the motor 420. Alternatively, the hydraulic pump 410 can also suck working fluid from a tank and discharge the working fluid from one of the first port 411 and the second port 412.

[0104] The motor 420 operates the hydraulic pump 410. The motor 420 drives the hydraulic pump 410 with a rotation direction and torque corresponding to a control signal from the control device 700. The motor 420 may be an electric motor or an electric servomotor.

[0105] The hydraulic cylinder 430 includes a cylinder body 431, a piston 432, and a piston rod 433. The cylinder body 431 is fixed to the injection device 300. The piston 432 divides the interior of the cylinder body 431 into a front chamber 435 as a first chamber and a rear chamber 436 as a second chamber. The piston rod 433 is fixed to the fixed platen 110.

[0106] The front chamber 435 of the hydraulic cylinder 430 is connected to the first port 411 of the hydraulic pump 410 via the first flow path 401. The working fluid discharged from the first port 411 is supplied to the front chamber 435 via the first flow path 401, thereby pushing the injection device 300 forward. The injection device 300 moves forward and the nozzle 320 is pressed against the fixed mold 810. The front chamber 435 functions as a pressure chamber that generates the nozzle contact pressure of the nozzle 320 by the pressure of the working fluid supplied from the hydraulic pump 410.

[0107] On the other hand, the rear chamber 436 of the hydraulic cylinder 430 is connected to the second port 412 of the hydraulic pump 410 via the second flow path 402. The hydraulic fluid discharged from the second port 412 is supplied to the rear chamber 436 of the hydraulic cylinder 430 via the second flow path 402, thereby pushing the injection device 300 backward. The injection device 300 retreats and the nozzle 320 separates from the fixed mold 810.

[0108] In this embodiment, the moving device 400 includes the hydraulic cylinder 430 , but the present invention is not limited thereto. For example, instead of the hydraulic cylinder 430 , an electric motor and a motion conversion mechanism that converts the rotational motion of the electric motor into the linear motion of the injection device 300 may be used.

[0109] (Control device)

[0110] The control device 700 is constituted by a computer, for example. Figure 1-2 As shown, the control device 700 has a CPU (Central Processing Unit) 701, a storage medium such as a memory 702, an input interface 703, an output interface 704, and a communication interface 705. The control device 700 performs various controls by causing the CPU 701 to execute a program stored in the storage medium 702. In addition, the control device 700 receives signals from the outside through the input interface 703 and sends signals to the outside through the output interface 704.

[0111] The control device 700 repeatedly produces molded products by repeatedly performing metering steps, mold closing steps, pressure increasing steps, mold clamping steps, filling steps, pressure holding steps, cooling steps, pressure releasing steps, mold opening steps, and ejection steps. A series of actions for obtaining molded products, such as the actions from the metering step to the next metering step, are also referred to as "injection" or "molding cycle". In addition, the time required for one injection is also referred to as "molding cycle time" or "cycle time".

[0112] A molding cycle includes, for example, a metering process, a mold closing process, a pressure increasing process, a mold closing process, a filling process, a pressure holding process, a cooling process, a pressure releasing process, a mold opening process, and an ejection process in sequence. The order here is the order in which each process starts. The filling process, the pressure holding process, and the cooling process are performed during the mold closing process. The start of the mold closing process may also coincide with the start of the filling process. The end of the pressure releasing process coincides with the start of the mold opening process.

[0113] In addition, multiple processes can be performed simultaneously for the purpose of shortening the molding cycle time. For example, the metering process can be performed during the cooling process of the previous molding cycle, or during the mold closing process. In this case, the mold closing process can be performed at the beginning of the molding cycle. In addition, the filling process can be started during the mold closing process. In addition, the ejection process can be started during the mold opening process. When an on-off valve is set to open and close the flow path of the nozzle 320, the mold opening process can be started during the metering process. Because even if the mold opening process is started during the metering process, as long as the on-off valve closes the flow path of the nozzle 320, the molding material will not leak from the nozzle 320.

[0114] In addition, a primary molding cycle may include processes other than the metering process, the mold closing process, the pressure increasing process, the mold clamping process, the filling process, the pressure holding process, the cooling process, the pressure releasing process, the mold opening process, and the ejection process.

[0115] For example, after the pressure holding process is completed and before the metering process begins, a pre-metering suction process may be performed to retract the screw 330 to a preset metering start position. The pressure of the molding material accumulated in front of the screw 330 can be reduced before the metering process begins, and the screw 330 can be prevented from retreating sharply when the metering process begins.

[0116] Furthermore, after the metering process is completed and before the filling process begins, a post-metering backsucking process may be performed in which the screw 330 is retracted to a preset filling start position (also referred to as an "injection start position"). The pressure of the molding material accumulated in front of the screw 330 can be reduced before the filling process begins, and leakage of the molding material from the nozzle 320 before the filling process begins can be prevented.

[0117] The control device 700 is connected to an operation device 750 that receives input operations from a user and a display device 760 that displays a screen. The operation device 750 and the display device 760 are composed of, for example, a touch panel 770 and can be integrated. The touch panel 770 as the display device 760 displays a screen under the control of the control device 700. Information such as the settings of the injection molding machine 10 and the current state of the injection molding machine 10 can be displayed on the screen of the touch panel 770. The touch panel 770 can receive operations in the displayed screen area. In addition, operation parts such as buttons and input fields for receiving input operations from the user can be displayed in the screen area of ​​the touch panel 770. The touch panel 770 as the operation device 750 detects the input operation of the user on the screen and outputs a signal corresponding to the input operation to the control device 700. Thus, for example, the user can confirm the information displayed on the screen while operating the operation part set on the screen to set the injection molding machine 10 (including the input of the set value). In addition, the user operates the operation part set on the screen, thereby enabling the injection molding machine 10 corresponding to the operation part to operate. In addition, the operation of the injection molding machine 10 may be, for example, the operation (including stopping) of the mold clamping device 100, the ejection device 200, the injection device 300, the moving device 400, etc. Furthermore, the operation of the injection molding machine 10 may be the switching of the screen displayed on the touch panel 770 as the display device 760, etc.

[0118] In addition, the operating device 750 and the display device 760 of this embodiment are described as being integrated into the touch panel 770, but they may be provided independently. In addition, a plurality of operating devices 750 may be provided. The operating device 750 and the display device 760 are arranged on the operating side (negative direction of the Y axis) of the mold clamping device 100 (more specifically, the fixed platen 110).

[0119] Figure 3FIG. 1 is a diagram showing an example of the functional structure of the control device 700 according to the present embodiment. Figure 3 As shown, in Figure 3 Components of the control device 700 of the injection molding machine 10 are shown as functional blocks. Figure 3 The functional blocks shown in the figure are conceptual functional blocks and do not have to be physically constructed as shown in the figure. All or part of each functional block can be functionally or physically dispersed / integrated in any unit to form a structure. All or any part of each processing function performed in each functional block is implemented by a program executed by CPU701. Alternatively, each functional block can also be implemented as hardware based on wired logic. Figure 3 As shown in FIG. 1 , the CPU 701 of the control device 700 includes an acquisition unit 711, a training data generation unit 712, a learning unit 713, an injection molding control unit 714, a determination unit 715 having an inference unit 716, and a display control unit 717. In addition, the control device 700 includes a training data storage unit 721 and a learning model storage unit 722 in the storage medium 702. In this embodiment, the control device 700 functions as a monitoring device of the injection molding machine 10.

[0120] The training data storage unit 721 stores training data used for learning the learned model LM. The structure of the training data will be described later.

[0121] The learned model storage unit 722 stores the learned model LM.

[0122] The acquisition unit 711 acquires waveform data representing the detection results detected by the sensor provided in the injection molding machine 10 in a time series from the injection molding machine 10. In the present embodiment, the acquisition unit 711 acquires waveform data from the start to the end of the molding cycle of the injection molding machine 10. In addition, in the present embodiment, the acquired waveform data is not limited to waveform data from the start to the end of the molding cycle, as long as it is waveform data for each injection. For example, the waveform data may also be waveform data representing a part or the whole of the molding cycle, or waveform data similar to the previous and next injections.

[0123] The acquired waveform data include, for example: the first waveform data representing the mold clamping force detection value detected by the mold clamping force detector in time series, the second waveform data representing the pressure exerted on the screw 330 from the molding material in time series calculated by the load detector 360 based on the detected load signal, and the third waveform data representing the position of the screw 330 detected by the injection motor encoder 351 in time series. In addition, the present embodiment shows an example of waveform data, but is not limited to the first to third waveform data. For example, it is also possible to use not all of the first to third waveform data, but any one or more of them. Moreover, waveform data representing the detection results detected by other sensors provided in the injection molding machine 10 in time series may also be used.

[0124] The training data generation unit 712 generates training data that combines the waveform data acquired by the acquisition unit 711 and correct information indicating the evaluation of the molded product produced when the waveform data is detected. The training data generation unit 712 stores the generated training data in the training data storage unit 1321.

[0125] The correct information involved in this embodiment is set as quantitative data (continuous scale) representing the evaluation of the product. Moreover, quantitative data includes both discrete values ​​such as integers and continuous values ​​such as real numbers. In this embodiment, the weight of the molded product measured after production is used as the correct information. In this embodiment, an example of using the weight of the molded product as the correct information is described, but the correct information is not limited to the weight. That is, the correct information can be quantitative data representing the state of the produced molded product or the molding material during molding.

[0126] That is, when the user refers to the first to third waveform data, it is difficult to determine whether an abnormality has occurred in the molded product. In contrast, the learned model LM according to the present embodiment outputs quantitative data indicating the state of the molded product or the molding material being molded when the first to third waveform data are input. Since the output information is quantitative data, it is easy to determine whether an abnormality has occurred.

[0127] The learning unit 713 generates a learned model LM by performing machine learning based on the training data group stored in the training data storage unit 721. The learned model LM is generated by applying supervised learning to a basic learning model.

[0128] Specifically, the learning unit 713 performs machine learning based on the waveform data (an example of the second data) and the weight of the molded product (an example of the second value) included in the training data set, thereby generating a learned model LM.

[0129] When waveform data (an example of the first data) representing the respective timing of the mold clamping force detection value detected during molding by the injection molding machine 10, the pressure received from the molding material, and the position of the screw 330 are input from the input layer, the learned model LM outputs the weight of the molded product (an example of the first value) from the output layer. Thus, in the present embodiment, the output data from the output layer of the learned model LM is set as a regression model.

[0130] Furthermore, the learned model LM may be updated by additionally learning a new training data set to the existing learned model LM.

[0131] Figure 4 2 is a diagram illustrating a configuration of a learned model LM used in the control device 700 according to the first embodiment.

[0132] Figure 4 The learned model LM shown is composed of an input layer 1411, one or more intermediate layers 1412 connected to the input layer, and an output layer 1413 connected to the intermediate layer. The input layer has, for example, 2200 nodes, the intermediate layer has, for example, 200 nodes, and the output layer has 1 node.

[0133] When input data 1401 corresponding to 2200 nodes is input, the learned model LM outputs output data y corresponding to 1 node.

[0134] The input data 1401 is set to, for example: a first waveform data which represents in time series the mold clamping force detection value detected by the mold clamping force detector, a second waveform data which represents in time series the pressure exerted on the screw 330 from the molding material calculated by the load detector 360 based on the signal of the detected load, and a third waveform data which represents in time series the position of the screw 330 detected using the injection motor encoder 351.

[0135] The output data y outputs, for example, the weight of the molded product after injection molding.

[0136] When the learning unit 713 performs learning using the training data, the waveform data included in the training data may be preprocessed. In addition, the preprocessing may be performed when the inference unit 716 described later performs inference.

[0137] exist Figure 4 In the example shown, in the input layer, the variables corresponding to each node are defined as the input variable vector x 0～2199 In the middle layer, the variables corresponding to each node are defined as the intermediate variable vector z 0～199 In the output layer, the variable corresponding to the node is defined as the output data y.

[0138] At this time, the formula for calculating the intermediate variable vector z0 of the intermediate layer is calculated by the following formula (1). In formula (1), the weight vector w 0,0～2199 Let z be a (2200-dimensional) weight vector for outputting the value of each intermediate variable vector z0 from each node of the input layer. 1～199 It can also be calculated by using the same formula, so the description is omitted. Figure 4 The deviation b is omitted 0～199 , but it is set as the bias used in the neural network. The activation function φ can be considered to use the ReLU function, for example, but other functions can also be used.

[0139] z0=φ(w 0,0～2199 ·x 0～2199 +b0)……(1)

[0140] The formula for calculating the output data y of the output layer is calculated by the following formula (2). In formula (2), the weight vector w' 0～199 Set as the weight vector for outputting values ​​from each node in the middle layer to the output data y. Figure 4 The bias b' is omitted in , but it is set as the bias used in the neural network. The activation function φ' can be considered to use the ReLU function, for example, but other functions can also be used.

[0141] y=φ'(w' 0～199 ·z 0～199 +b')……(2)

[0142] Figure 5 Schematic diagram showing the structure of the learning model LM involved in this embodiment. Figure 5 As shown, the learned model LM according to the present embodiment includes a body LM1 and a head LM2.

[0143] The main body LM1 is a neural network for extracting feature quantities from input waveform data. As a learning method, deep learning can be applied. As a deep learning algorithm, a deep neural network (DNN), a convolutional neural network (CNN), or a recurrent neural network RNN ​​(Recurrent Neural Networks) can also be applied. In addition, LSTM (Long Short Term Memory) and Transformer can also be applied.

[0144] The neural network of the main body LM1 is a neural network designed to extract feature quantities from waveform data (connections between nodes are limited, parameters are shared between different nodes, or there are feedback connections, etc.).

[0145] As a result of machine learning, by extracting feature quantities through the main body LM1, it is possible to extract the relationship between the detection results of each sensor included in a plurality of waveform data.

[0146] The head LM2 is a structure for outputting quality characteristics of a molded product or a molding material, etc. from the extracted feature quantity, and can also be set as a neural network with a fully connected layer.

[0147] The quality characteristics of the molded product may be any characteristics that can be expressed as quantitative data by measuring instruments or tests after molding (shape or size (width, height, thickness), size of poor appearance (discoloration, silver streaks, cold blocks, etc.), warping, strength, etc.). Furthermore, the quality characteristics used in the present embodiment are not limited to the quality characteristics of the molded product. In other words, the data that can be measured after injection molding as described above, as quantitative data representing the state of the molding material in injection molding, such as viscosity, specifically the fluidity of the molding material, may also be the melt mass flow rate (MFR) or the melt volume flow rate (MVR), etc. Any method may be used to obtain the quality characteristics of the molded product or molding material.

[0148] Furthermore, the present embodiment is not limited to using a neural network as a method of calculating the head LM2 for deriving the quality characteristics of a molded product or a molded material based on the feature quantity, and for example, a non-neural network regression model may also be used. In this case, after machine learning is performed on the structure including the main body LM1 and the head LM2 during the first machine learning, the parameters until the feature quantity of the main body LM1 is extracted may be fixed during the second machine learning and inference, and other algorithms (e.g., decision tree, random forest, gradient boosting, SVR, etc.) may also be replaced in the head LM2.

[0149] In this embodiment, the case where the input data 1501 is time series data of the results detected by the sensor at each injection and includes the first waveform data that represents the mold clamping force detection value detected by the mold clamping force detector in time series, the second waveform data that represents the pressure exerted on the screw 330 from the molding material in time series calculated by the load detector 360 based on the signal of the detected load, and the third waveform data that represents the position of the screw 330 detected by the injection motor encoder 351 in time series is described. However, this embodiment does not limit the input data input to the learned model LM. For example, the input data can be any one or more of the first waveform data to the third waveform data, and can also include time series data of the results detected by other sensors.

[0150] Furthermore, transfer learning can also be used in the learning method of the learned model LM. For example, a part of the neural network included in the learned model LM (for example, the main body LM1) can be subjected to the first machine learning without using the correct answer information, and then the correct answer information can be used to perform the second machine learning.

[0151] In this method, the first machine learning can be performed without using the correct answer information, and after fixing the parameters until the feature amount of the main body LM1 is extracted, the second machine learning can be performed according to the type of correct answer information. In this way, multiple learned models LM are generated, and each of the multiple learned models LM can output different quality characteristics. For example, one of the learned models LM can output the weight of the molded product when the first waveform data to the third waveform data are input, and another learned model LM can output the viscosity of the molding material when the first waveform data to the third waveform data are input.

[0152] Then, the output data y is output from the learned model LM as quantitative data based on the regression model. This can be achieved by configuring the final output of the learned model LM as a single node and applying MSE (mean square error) or MAE (mean absolute error) as an error function.

[0153] Furthermore, the output data y outputted from the learned model LM outputs the quality characteristics of the molded product other than the weight, and can be converted from the quality characteristics to the weight.

[0154] Furthermore, the output data y outputted by the learned model LM may be quality characteristics of the molded product, molding material, or components of the injection molding machine 10. That is, the output data y outputted by the learned model LM may be quantitative data that can be used to determine whether the product is abnormal or not.

[0155] In the present embodiment, the quality characteristics of the molded product or the molding material that can be expressed as quantitative data are output from the learned model LM as output data y. Moreover, in the present embodiment, the quality characteristics that can be output as output data y are not limited to the molded product or the molding material. The outputtable quality characteristics as output data y may be, for example, the set value of the molding condition, the detection value that serves as the basis for the molding condition, the detection value of the molded product after molding, the state of the molding material during molding, the size of the components of the injection molding machine 10 (for example, the mold device 800), the amount of degradation, etc. Moreover, these quality characteristics may also be used to determine whether an abnormality has occurred.

[0156] Return to Figure 3 The learning unit 713 stores the generated learned model LM in the learned model storage unit 722.

[0157] The injection molding control unit 714 executes processing for producing a molded product by the injection molding machine 10. For example, when producing a molded product, the injection molding control unit 714 may set each item constituting the molding conditions and then perform injection molding.

[0158] The determination unit 715 includes an inference unit 716 , and determines whether an abnormality (defect) has occurred in the molded product based on information acquired from the inference unit 716 .

[0159] The inference unit 716 inputs waveform data (an example of first data) indicating the result detected by the mold clamping force detector during the production of the molded product by the injection molding control unit 714 to the input layer of the learned model LM, and acquires a value indicating the evaluation of the molded product as quantitative data from the output layer of the learned model LM. In this embodiment, a case where the acquired value (an example of the first value) is the weight of the molded product or the viscosity of the molding material during molding is described.

[0160] Furthermore, the inference unit 716 may pre-process the waveform data before inputting the waveform data to the input layer.

[0161] Furthermore, the determination unit 715 determines whether an abnormality has occurred in the detection result of the sensor provided in the injection molding machine 10 based on the quantitative data of the quality characteristics of the molded product (for example, the weight of the molded product or the viscosity of the molding material during molding) received from the inference unit 716. The criterion for determining whether an abnormality has occurred can also be set by the user.

[0162] The display control unit 717 displays information on the display device 760 .

[0163] Figure 6 FIG. 7 is a diagram illustrating a log information screen displayed by the display control unit 717 according to the present embodiment. Figure 6 In the log information screen shown in , the user sets the criteria for determining whether or not there is an abnormality.

[0164] exist Figure 6 In the log information screen 1600 shown, a monitoring setting menu 1601 , a monitoring range setting menu 1602 , a statistics list 1610 , and a performance list 1620 are displayed.

[0165] The statistical overview 1610 displays the statistical values ​​(e.g., average, range, maximum, minimum, standard deviation) of each setting column 1611-1612. The contents displayed in the setting columns 1611-1612 can be set by the user. In this embodiment, the items displayed in the setting columns 1611-1612 can be displayed, monitored, and log information can be saved. In addition, the monitoring of this embodiment indicates whether it is a good product based on a specified standard.

[0166] The “monitoring”, “monitoring value”, and “range” of the monitoring items 1650 included in the statistical list are information set to determine whether or not the molded product is defective.

[0167] When "Monitoring" in the statistics list 1610 is "off", the control device 700 does not monitor, and when it is "fixed" or "automatic", the control device 700 monitors. When it is "fixed" or "automatic", the control device 700 determines whether the detected value meets the criteria shown in the "monitoring value" and "range" in the items shown in the setting columns 1611 and 1612. The switching of the monitoring is performed through the monitoring setting menu 1601.

[0168] "Monitoring value" sets the center value for determining the quality characteristics of molded products. In "Range", set the "Monitoring value" to the center and set the upper and lower widths for determining whether it is defective.

[0169] “Defective” in the statistical list 1610 indicates the number of molded products that do not satisfy the criteria shown in “Monitoring value” and “Range”.

[0170] The “weight g” of the “AI inference quality” in the setting column 1611 and the “viscosity Pa·s” of the “AI inference quality” in the setting column 1612 are items set to monitor the weight output by the learned model LM and the viscosity output by the learned model LM.

[0171] That is, in Figure 6 In the example shown in , the learned model LM outputting "weight" and the learned model LM outputting "viscosity" are stored in advance in the learned model storage unit 722. Furthermore, the inference unit 716 inputs the waveform data (for example, the first waveform data to the third waveform data mentioned above) representing the detection results detected by the sensor for each injection in a time series to each of the learned model LM outputting "weight" and the learned model LM outputting "viscosity". Thus, the inference unit 716 receives the inference result of "weight" and the inference result of "viscosity". Furthermore, the determination unit 715 determines whether the molded product is abnormal for each of the inference result of "weight" and the inference result of "viscosity".

[0172] Furthermore, in this embodiment, an example is described in which the weight of the molded product output from the learned model LM is expressed in the measurement unit [g] and the viscosity of the molding material is expressed in the measurement unit [Pa·s]. That is, quantitative data indicating the quality characteristics of the molded product or molding material are expressed based on the specific index of the measurement unit, so that the user can easily understand the quality characteristics.

[0173] In the display Figure 6In the case of the screen shown in FIG. 1 , the determination unit 715 determines whether the molded product is abnormal based on whether the inference result of "weight" and the inference result of "viscosity" meet the criteria shown in "monitoring value" and "range" respectively. The determination criteria shown in "monitoring value" and "range" are set according to the measurement unit of the determination object (for example, weight or viscosity).

[0174] For example, in the setting column 1611, the determination unit 715 determines whether the inference result of "weight" is included within the "range", i.e., ± "1.000" [g], based on the "monitoring value", i.e., "45.000" [g]. That is, the determination unit 715 determines that it is abnormal when the "weight" is greater than "46.000 [g]" (an example of a preset threshold value), and determines that it is abnormal when the "weight" is less than "44.000 [g]" (an example of a preset threshold value).

[0175] Similarly, in the setting column 1612, the determination unit 715 determines whether the inference result of "viscosity" is included in the "range", that is, ± "10" [Pa·s], based on the "monitoring value", that is, "340" [Pa·s]. That is, the determination unit 715 determines that it is abnormal when the "viscosity" is greater than "350" [Pa·s] (an example of a preset threshold value), and determines that it is abnormal when the "weight" is less than "330" [Pa·s] (an example of a preset threshold value).

[0176] Then, the determination result of the determination unit 715 is displayed in the performance list 1620 .

[0177] In addition, the setting fields 1611 to 1612 are configured to be changeable to items that the user wishes to monitor. The description of the changing method is omitted.

[0178] The performance list 1620 shows a list of quality characteristics of molded products output by the learned model LM in the items set in the setting columns 1611 to 1612 for each shot.

[0179] In addition, in this embodiment, an example of outputting the quality characteristics of the molded product output by the learned model LM to the performance overview 1620 is described, but the information output to the performance overview 1620 is not limited to quality characteristics, and for example, it may include setting information set for injection molding and any one or more of the detection values ​​measured by various sensors.

[0180] The output results of the items set in the setting columns 1611 and 1612 are shown in the "CH-1" column 1621 and the "CH-2" column 1622. In the performance list 1620, each time a shot is injected, the "shot number", the "time" of injection molding, and the "discrimination" of injection molding are associated with each other as information indicating the injection.

[0181] Furthermore, in the "CH-1" column 1621 to the "CH-2" column 1622, when "Monitoring" is "Fixed" or "Automatic", if the quality characteristic is greater than the upper limit value set in the "Monitoring value" and "Range", "H" is displayed. On the other hand, if the quality characteristic is less than the lower limit value set in the "Monitoring value" and "Range", "L" is displayed.

[0182] In the shot, if it is determined that at least one of the CHs of the monitoring target does not meet the standard, "E (bad)" is displayed in "Discrimination". For example, in the shot where "H" or "L" is displayed in any of the "CH-1" column 1621 and the "CH-2" column 1622, "E (bad)" is displayed.

[0183] Furthermore, every time "E" is displayed in "Judgment", the value in "Defective" in the corresponding setting column of the statistical list 1610 increases by "1".

[0184] The monitoring setting menu 1601 is provided as a button for accepting whether to monitor the items to be monitored in the statistical list 1610. When "ON" is selected in the monitoring setting menu 1601, each injection is monitored to see whether there is a defective product, and the monitoring result is displayed in "Discrimination" in the performance list 1620. The monitoring setting menu 1601 is switched to "OFF" or "ON" according to the operation from the user.

[0185] The monitoring range setting menu 1602 is selected from "[±]" and "[+ to -]". For "[±]", the plus side and minus side of the same range are monitored with the "monitoring value" as the center. For "[+ to -]", the range is set to monitor the plus side and minus side separately with the "monitoring value" as the center.

[0186] In the injection molding machine 10 , it is also possible to instruct an external device to remove a molded product determined to be “defective” (abnormal) from shipment.

[0187] Also, you can save Figure 6 The information shown in the statistical list 1610 and the actual performance list 1620 can be used to confirm the quality characteristics of the molded product when the molded product is suspected to be defective after shipment. This can be used to trace the inference results of the defective quality characteristics.

[0188] Furthermore, in the present embodiment, the numerical value of the quality characteristic of the molded product output by the learned model LM may not be accurate. That is, the numerical value (quantitative data) of the quality characteristic of the molded product output by the learned model LM is a value used instead of the monitoring waveform data, so it is sufficient as long as it is a value that can determine whether the molded product is defective based on the above-mentioned "monitoring value" and "range" and the like. For example, in the case where there is a corresponding relationship between the presence or absence of burrs and the weight received from the learned model LM, in the present embodiment, by setting a threshold value for the weight received from the learned model LM, the presence or absence of burrs can be detected. However, the weight received from the learned model LM does not have to be exactly consistent with the weight of the molded product output from the injection molding machine 10. In other words, the weight received from the learned model LM may not be an accurate weight as long as it is related to the abnormality that is the detection object (for example, whether burrs are generated).

[0189] Figure 6 The case where the monitoring function is included in the log information screen has been described. However, the present embodiment is not limited to the case where the monitoring function is included in the log information screen. A monitoring setting screen may be prepared separately from the log information screen.

[0190] Figure 7 This is a diagram illustrating a monitoring setting screen displayed by the display control unit 717 according to the present embodiment.

[0191] exist Figure 7 The monitoring setting screen 1700 shown includes a monitoring setting menu 1701 , a monitoring range setting menu 1702 , an item selection field 1703 , a monitoring menu 1704 , a monitoring value setting field 1705 , and a range setting field 1706 .

[0192] The monitoring setting menu 1701 and the monitoring range setting menu 1702 are Figure 6 The monitoring setting menu 1601 and the monitoring range setting menu 1602 are the same.

[0193] The item selection field 1703 is a field for selecting the type of quality characteristic output by the learned model LM. In the present embodiment, the item selection field 1703 is selected according to the type of the learned model LM stored in the learned model storage unit 722.

[0194] The monitoring menu 1704 is a menu that receives a selection from “off,” “fixed,” and “automatic.” In the case of “off,” the control device 700 does not monitor, and in the case of “fixed” or “automatic,” the control device 700 monitors.

[0195] The monitoring value setting column 1705 is a column for setting a center value for determining the quality characteristics of the molded product.

[0196] The range setting column 1706 is a column for setting the upper and lower widths for determining whether or not a value is defective, with the “monitoring value” as the center.

[0197] Moreover, in the Figure 7 In the case of the settings of the columns shown, the control device 700 determines whether the molded product is abnormal based on the quality characteristics. Furthermore, when the molded product is determined to be abnormal, the molded product can be removed from the production line as abnormal, the user can be notified of the abnormality, or the production of the molded product can be stopped. In this way, any measures can be taken as abnormal processing regardless of the known measures.

[0198] Furthermore, in the control device 700, an abnormality (defect) of the molded product can be detected by the above-mentioned method. Furthermore, it can be said that what kind of control is to be performed when a defect is detected can be set.

[0199] Figure 8 1 is a diagram illustrating an operation setting screen displayed by the display control unit 717 according to the present embodiment when a failure is detected.

[0200] exist Figure 8 The operation setting screen 1800 for detecting a defect shown includes a defect stop setting column 1801 , a defect scrapping column 1802 , and a scrapping after molding interruption setting column 1803 .

[0201] The failure stop setting column 1801 includes a denominator column 1811 and a numerator column 1812. The denominator column 1811 sets the number of shots that will become the denominator when the failure stops, and the numerator column 1812 sets the number of shots that will become the numerator when the failure stops.

[0202] That is, when the injection molding machine 10 produces a molded product, if the judgment unit 715 determines, based on the quality characteristics output by the learned model LM, that an abnormality (defect) of the number shown in the numerator column 1812 has occurred in the number shown in the denominator column 1811, the injection molding control unit 714 performs control to stop the injection molding.

[0203] The defective scrap column 1802 includes the number of scraps after defect generation 1821. The number of scraps after defect generation 1821 is set with the number of molded products that are subsequently scrapped when the determination unit 715 determines that an abnormality has occurred based on the quality characteristics output by the learned model LM.

[0204] When it is determined that an abnormality has occurred based on the quality characteristics outputted from the learned model LM, the control device 700 instructs to scrap only molded products produced after the determination of abnormality, which is the number of scraps set as 1821 after the defect has occurred.

[0205] The scrapping after molding interruption setting column 1803 includes a scrapping after molding interruption setting 1831, a molding interruption time setting 1832, and a scrap number after molding interruption 1833. In the scrapping after molding interruption setting 1831, whether to scrap the molded product is set by "on" or "off" after the injection molding is interrupted. In the molding interruption time setting 1832, it is set to be a column for setting the following time, which is used to determine that the injection molding is interrupted. When the interruption time is longer than the time set in the molding interruption time setting 1832, it is judged to be scrapped after the interruption. The control device 700 indicates as follows: after the interruption, only the number of molded products set to the scrap number after molding interruption 1833 is scrapped. Scrap after molding interruption is not limited to the case where the user interrupts the injection molding, but also includes the case where it is interrupted based on the judgment result of the defect (abnormality) of the molded product.

[0206] In the present embodiment, the control device 700 includes the above-described configuration, and thereby can perform abnormality determination and processing when abnormality is determined based on the quality characteristics output from the learned model LM.

[0207] In the present embodiment, the waveform data input to the learned model LM can be monitored based on the quality characteristics output from the learned model LM.

[0208] Fig. 9 is a diagram illustrating the quality characteristics output from the learned model LM according to the present embodiment. Fig. 9 In the example shown, two quality characteristics (weight and viscosity) output from the learned model LM are shown for each (shot) number.

[0209] exist Fig. 9 In the example shown, the user can recognize the state of the molded product or the molding material by referring to the weight and viscosity for each (shot) number.

[0210] Fig.10 and Fig.11 The following is an example of the detection of the molded product when the injection molding machine 10 according to the first embodiment is used. Fig. 9 A graph of the waveform data for each (shot) number shown in . Fig.10 and Fig.11 In the example shown, the vertical axis is the detection value and the horizontal axis is time. Fig.10 and Fig.11 In the above description, the detection value detected as the waveform data may be any detection result of any sensor provided in the injection molding machine 10 .

[0211] Fig.10 The waveform data 2001 shown corresponds to Fig. 9The (injection) number in is "1", the waveform data 2002 corresponds to the number "2", and the waveform data 2003 corresponds to the number "3".

[0212] Fig.11 The waveform data 2001 shown corresponds to Fig. 9 The (injection) number in is "1", the waveform data 2101 corresponds to the number "4", and the waveform data 2102 corresponds to the number "5".

[0213] That is, the user can understand the difference in weight, in other words, the difference in quality characteristics of the molded product by understanding the difference in waveform data 2001 to 2003 when referring to waveform data 2001 to 2003. However, it is difficult for an unskilled technician to understand the difference in waveform data 2001 to 2003.

[0214] Similarly, when the user refers to the waveform data 2001, 2101-2102, the user can understand the difference between the waveform data 2001, 2101-2102, and in other words, the difference in the quality characteristics of the molding material. However, it is difficult for an unskilled technician to understand the difference between the waveform data 2001, 2101-2102.

[0215] That is, the user should be able to identify whether an abnormality has occurred in the molded product or the molding material by monitoring the waveform data, but a technique is required to identify the abnormality. Furthermore, when an abnormality occurs in the molded product or the molding material, it occurs as a difference in the shape of the waveform data. Therefore, the user may consider setting a threshold value relative to the waveform data in order to detect the abnormality. However, the user is often unable to grasp the difference in the shape of the waveform data when an abnormality occurs in the molded product and when no abnormality occurs. Therefore, it is often difficult for the user to set a threshold value for the waveform data.

[0216] Therefore, in the present embodiment, the waveform data is input to the learned model LM, and inference is performed using the learned model LM.

[0217] That is, the control device 700 according to the present embodiment outputs the characteristics generated by the waveform data as the quality characteristics of the molded product or the molded material, etc. The user can identify the characteristics of the waveform data by referring to the output quality characteristics. In other words, the user can identify an abnormality based on the waveform data.

[0218] The criteria for determining whether a molded product or the like is defective often vary depending on the needs. In contrast, in the present embodiment, the learned model LM outputs quantitative data as the quality characteristics of the molded product or the molding material. Therefore, the threshold used for abnormality determination by the determination unit 715 also becomes a numerical value. Figure 6As shown in FIG. 1 , the user can adjust the determination criterion by adjusting the value of the threshold. That is, the control device 700 according to the present embodiment can make the determination criterion of whether a molded product or the like is defective flexible.

[0219] In the present embodiment, an example is described in which the learned model LM is generated in the control device 700. However, the learned model LM involved in the present embodiment is not limited to the method of being generated in the control device 700. The learned model LM generated by other information processing devices may be installed in the control device 700.

[0220] In this embodiment, the case where the value of the evaluation of the product is represented by quantitative data is described. However, this embodiment does not limit the evaluation of the product to quantitative data, as long as it is a parameter that serves as a reference for abnormality determination. A classification model in which the numerical value that will serve as the evaluation of the product is divided into each interval may also be used.

[0221] (Second embodiment)

[0222] In the above-mentioned embodiment, the following example is shown: in the injection molding machine 10, a molding material (for example, one kind) used to mold a molded product is limited, and reasoning using the molding material is performed, thereby performing high-precision reasoning. That is, in order to accurately reason about the weight of the molded product, etc., information on the inherent properties of the molding material is required, so the injection molding machine 10 prepares a learned model corresponding to the molding conditions. However, the above-mentioned embodiment is not limited to the method of preparing a learned model corresponding to the molding conditions in order to reason about the quality characteristics of the molded product. Therefore, in this embodiment, a case is described in which a learned model capable of outputting quality characteristics is generated when waveform data is input regardless of the molding conditions including the molding material.

[0223] In this embodiment, the learned model can be used regardless of the molding material used in the molded product, the type of mold device 800, and the molding conditions. That is, the learned model generated by the learning device can also be installed on each of the multiple injection molding machines. Therefore, in this embodiment, an example with a learning device and a control device is described.

[0224] Fig.12 This is a schematic diagram showing the cooperation between the learning device and the control device of the injection molding machine according to the present embodiment and the learned model.

[0225] exist Fig.12In the example shown, a test injection molding machine 1350, a learning device 1300, and a control device 700A for the injection molding machine 10 are shown. In this embodiment, an example of a learned model LM' having a different structure from the learned model LM of the above-mentioned embodiment is used. The structure of the learned model LM' will be described later.

[0226] exist Fig.12 In the example shown, it can be considered that the test injection molding machine 1350 and the learning device 1300 are owned by, for example, the manufacturer of the injection molding machine 10, but it is not limited to the method of being owned by the manufacturer of the injection molding machine. For example, they can be owned by the shipping location of the injection molding machine, or they can also be owned by the service provider who generates the learned model. Fig.12 The injection molding machine 10 and the control device 700A shown may show the status before shipment, for example.

[0227] The test injection molding machine 1350 is an injection molding machine for machine learning. The structure of the test injection molding machine 1350 is the same as that of the injection molding machine 10, so the description thereof is omitted.

[0228] The test injection molding machine 1350 according to the present embodiment produces a molded product (an example of a product) according to the settings based on the user. In addition, the test injection molding machine 1350 outputs the waveform data measured during the injection molding to the learning device 1300. In the present embodiment, data representing the detection results of the sensors provided in the test injection molding machine 1350 or the injection molding machine 10 in a time series is referred to as waveform data.

[0229] The learning device 1300 generates training data based on the input waveform data, and stores the training data in the training data storage unit 1321 .

[0230] In order to utilize artificial intelligence, it is necessary to perform a learning (training) phase and an inference (inference) phase.

[0231] The learning device 1300 executes the learning phase. Therefore, the learning unit 1313 of the learning device 1300 causes the neural network to read the training data stored in the training data storage unit 1321, and generates a network with adjusted synaptic weights and biases as a learned model LM'.

[0232] For example, the learning device 1300 reads a large amount of training data and performs machine learning based on back propagation using a neural network, thereby generating a learned model LM′.

[0233] The learning device 1300 may be, for example, a local server installed in a factory or a cloud server. Furthermore, the learning device 1300 may be a fixed terminal device or a portable terminal device (mobile terminal) installed in a factory or the like. For example, a desktop PC (Personal Computer) may be included in the fixed terminal device. Furthermore, for example, a smart phone, a tablet terminal, a portable PC, etc. may be included in the mobile terminal device.

[0234] The control device 700A of the injection molding machine 10 executes the inference phase. The control device 700A inputs data into the learned model LM' and causes the learned model LM' to perform inference. Furthermore, the control device 700A sets predetermined items of molding conditions based on the output results from the learned model LM'.

[0235] In this embodiment, the structure of the learned model LM' generated by the learning unit 1313 of the learning device 1300 and the learned model LM' used by the inference unit 716A of the control device 700A can be generalized. In addition, the learning device 1300 can transmit the weights and deviations of the learned model LM' to the control device 700A. The control device 700A updates the learned model LM' based on the received weights and deviations, thereby making it consistent with the learned model LM' of the learning unit 1313 of the learning device 1300. Thus, the learning unit 1313 of the learning device 1300 and the inference unit 716A of the control device 700A of the injection molding machine 10 can be installed in different languages. For example, the learning unit 1313 of the learning device 1300 can also be installed using a computer high-level programming language (Python). On the other hand, the inference unit 716A of the control device 700A can also be installed using C++, for example.

[0236] Furthermore, the learning device 1300 and the control device 700A can have different CPUs, OSs, development languages, etc. The control device 700A can use a CPU with a lower processing speed than the learning device 1300. This can achieve cost reduction.

[0237] Furthermore, the structural information of the transmitted weights and biases may include the values ​​of each weight, the values ​​of each bias and parameter structural information. For example, it may be an object that describes the values ​​of each weight and each bias in accordance with a data structure and in a prescribed method (for example, JSON (JavaScript Object Notation: data exchange format), XML (Extensible Markup Language: Extensible Markup Language), ONNX (Open Neural Network Exchange: Open Neural Network Exchange), etc.).

[0238] Thus, the learning device 1300 can transmit information indicating the structure of the weights and the biases to the control device 700A of the injection molding machine 10 .

[0239] In addition, the present embodiment is not limited to the method of transmitting information indicating the structure of weights and biases from the learning device 1300 to the injection molding machine 10, and the learning model LM' itself may be transmitted from the learning device 1300 to the control device 700A of the injection molding machine 10. In this case, the CPU, OS, development language, etc. may be made common between the control device 700A and the learning device 1300.

[0240] Thus, the control device 700A can perform inference using the learned model LM' generated in the learning phase of the learning device 1300. Next, the structure of the learned model LM' will be described.

[0241] Fig.13 Schematic diagram showing the structure of the learning model LM' involved in this embodiment. Fig.13 As shown, the learned model LM' according to the present embodiment includes a body LM1' and a head LM2'.

[0242] The learned model LM′ according to the present embodiment has two channels (CH_A and CH_B), and waveform data 2401 of different shots are input to the two channels, respectively.

[0243] The input layer of the learned model LM' is configured, for example, to input the waveform data of the previous shot into the channel CH_A and the waveform data of the current shot into the channel CH_B.

[0244] The main body LM1' is set as a neural network for extracting feature amounts from two or more waveform data inputted. The learning method is the same as that in the above-mentioned embodiment, and thus the description thereof is omitted.

[0245] The main body LM1 ′ extracts a feature value from each of the waveform data of the previous shot input to the channel CH_A and the waveform data of the current shot input to the channel CH_B.

[0246] The head LM2' outputs output data representing the difference between the molded product molded by the previous injection molding and the molded product molded by the current injection molding in terms of quality characteristics from the feature quantity extracted from each of the two or more waveform data. Figure 5 In the example shown, the weight ratio of the molded product of the current shot compared to the molded product of the previous shot is output. Figure 5 In the example shown, it is shown that the weight of the molded product this time has increased by 0.50% compared with the previous molded product.

[0247] That is, the learned model LM' involved in this embodiment extracts feature values ​​from the difference in waveform data of two or more different shots, and outputs the difference in quality characteristics of the molded product between these shots. The learned model LM' performs inference based on the difference between two or more waveform data, so it can be used regardless of the molding material, the mold device 800, the molding conditions, etc.

[0248] Fig.14 1 is a diagram showing an example of the functional configuration of the learning device 1300 and the control device 700A according to the present embodiment. Fig.14 As shown, the learning device 1300 and the control device 700A are connected via a communication line NW.

[0249] The functions of the learning device 1300 are implemented by any hardware or any combination of hardware and software. Fig.14 As shown, the learning device 1300 includes a CPU 1301 , a storage medium 1302 , and a communication interface 1303 .

[0250] The storage medium 1302 stores various installed programs, and stores files and data required for various processes, etc. The storage medium 1302 includes, for example, a HDD (Hard Disc Drive), an SSD (Solid State Drive), a flash memory, and the like.

[0251] The storage medium 1302 according to this embodiment includes a training data storage unit 1321 and a learned model storage unit 1322 .

[0252] The training data storage unit 1321 stores training data used for learning the learned model LM'. The structure of the training data will be described later.

[0253] The learned model storage unit 1322 stores the learned model LM′.

[0254] The communication interface 1303 is used as an interface for connecting to external devices so as to enable communication. Thus, the learning device 1300 can communicate with external devices such as the injection molding machine 10 through the communication interface 1303. In addition, the communication interface 1303 may have multiple communication interfaces depending on the communication method between the connected devices.

[0255] The CPU 1301 of the learning device 1300 executes a program stored in the storage medium 1302. Thus, the CPU 1301 includes an acquisition unit 1311, a training data generation unit 1312, a learning unit 1313, and a communication control unit 1314 as functional units.

[0256] Each time injection molding is performed, the acquisition unit 1311 acquires waveform data (an example of first data) indicating a detection result detected by a sensor provided in the test injection molding machine 1350 from the test injection molding machine 1350. In the present embodiment, the acquisition unit 1311 acquires waveform data from the start to the end of the molding cycle of the test injection molding machine 1350. The types of waveform data acquired are the same as those in the above-mentioned embodiment, and thus description thereof is omitted.

[0257] Furthermore, the acquisition unit 1311 acquires the quality characteristics of the molded product of the injection molding machine through the test injection molding machine 1350 .

[0258] Furthermore, the quality characteristics output from the learned model LM' may be normalized. For example, the quality characteristics output from the learned model LM' may be adjusted to output a value between 0.0 and 1.0. Furthermore, the quality characteristics may be standardized.

[0259] The training data generation unit 1312 generates training data by combining the waveform data of the n-1th shot of each (shot) number acquired by the acquisition unit 1311, the waveform data of the nth shot, and the difference in quality characteristics (e.g., increase or decrease in weight, increase or decrease in viscosity) between the n-1th molded product and the nth molded product. The training data generation unit 1312 stores the generated training data in the training data storage unit 1321. In addition, n is set to a natural number greater than 2.

[0260] This embodiment illustrates a method for generating training data. In addition to the above method, training data can also be generated. In addition, it is not limited to the method in which the learning device 1300 automatically performs all the learning until the generation of the model LM' is completed. For example, after the learning device 1300 generates a plurality of training data, the user can confirm the content of each of the plurality of training data.

[0261] In this embodiment, a plurality of training data may be referred to as a training data set. The training data set is a collection of training data based on a combination of waveform data as input and values ​​representing quality characteristics of molded products or molding materials as output in the form of quantitative data.

[0262] Then, after the user confirms and corrects the contents included in the training data set, the learning device 1300 can generate a learned model LM′ based on the training data set.

[0263] The learning unit 1313 generates a learned model LM' by performing machine learning based on the training data set stored in the training data storage unit 1321. The learned model LM' is generated by applying supervised learning to the basic learning model. As a variation, there is also a method of using transfer learning to generate a learned model LM'. When using transfer learning to generate the learned model LM', for example, a learning method can also be used, which uses unsupervised learning for the main body LM1' and supervised learning for the head LM2'.

[0264] Specifically, the learning unit 1313 generates a learned model LM' by performing machine learning based on the waveform data of the first shot (an example of the third cycle), the waveform data of the second shot (an example of the fourth cycle) included in the training data group, and the value (an example of the second value) representing the difference in quality characteristics between the molded product produced in the first shot and the molded product produced in the second shot as quantitative data.

[0265] When the waveform data of the last shot and the waveform data of the current shot detected during molding by the injection molding machine 10 or the test injection molding machine 1350 are input from the input layer, the learned model LM' outputs the difference in quality characteristics between the molded product produced by the last shot and the molded product produced by the current shot from the output layer.

[0266] Furthermore, the learned model LM′ can be updated by additionally learning a new training data set to the existing learned model LM′.

[0267] As machine learning for generating the learned model LM', for example, a neural network can be applied, and as a more specific example, deep learning can be applied. As a deep learning algorithm, a deep neural network (DNN: Deep Neural Network), a convolutional neural network (CNN), or a recurrent neural network RNN ​​(Recurrent Neural Networks) can also be applied. In addition, LSTM (Long Short Term Memory: Long Short Term Memory Network) and Transformer can also be applied.

[0268] The communication control unit 1314 uses the communication I / F 1303 to send and receive information to and from an external device such as the injection molding machine 10. For example, the communication control unit 1314 may send the learned model LM' stored in the learned model storage unit 1322 to the control device 700A of the injection molding machine 10. In addition, the communication control unit 1314 may extract a structure representing parameters (e.g., weights and biases) set in each layer constituting the learned model LM' stored in the learned model storage unit 1322, and send the structure to the control device 700A.

[0269] exist Fig.14 Components of the control device 700A of the injection molding machine 10 are shown as functional blocks. Fig.14 The functional blocks shown in the figure are conceptual functional blocks and do not have to be physically constructed as shown in the figure. All or part of each functional block can be functionally or physically dispersed / integrated in any unit to form a structure. All or any part of each processing function performed in each functional block is implemented by a program executed by CPU701. Alternatively, each functional block can also be implemented as hardware based on wired logic. Fig.14 As shown, the CPU 701 of the control device 700A includes a communication control unit 718, an update unit 719, an acquisition unit 711, an injection molding control unit 714, a determination unit 715A including an inference unit 716A, and a display control unit 717. In addition, the control device 700A includes a learned model storage unit 722A in the storage medium 702.

[0270] The learned model storage unit 722A stores the learned model LM′. The structure of each layer of the learned model LM′ is the same as that of the learned model LM′ stored in the learned model storage unit 1322 of the learning device 1300 .

[0271] The communication control unit 718 uses the communication I / F 705 to send and receive information to and from an external device such as the learning device 1300. For example, the communication control unit 718 may receive the learned model LM' from the learning device 1300. In addition, the communication control unit 718 may receive information representing a structure of parameters (e.g., weights and biases) set in each layer constituting the learned model LM'.

[0272] The updating unit 719 updates the learned model LM′ stored in the learned model storage unit 722A using the received learned model LM′ or information indicating the structure. Thus, the updating unit 719 can make the learned model LM′ stored in the learned model storage unit 722A consistent with the learned model LM′ of the learning device 1300 .

[0273] In the case of updating by means of information representing a structure (representing parameters (e.g., weights and biases) set in each layer constituting the learned model LM'), the update unit 719 may be configured as a program for updating the learned model LM'. The update unit 719 may also determine whether the number of layers defined and the number of nodes per layer are consistent between the learned model LM' stored in the learned model storage unit 722A and the received information representing the structure. Furthermore, the update unit 719 updates the learned model LM' when it is determined that they are consistent. When it is determined that they are inconsistent, the update unit 719 may output an alarm screen having a different structure to the display device 760, etc., without updating the learned model LM'.

[0274] The injection molding control unit 714 executes processing for producing a molded product by the injection molding machine 10. For example, when producing a molded product, the injection molding control unit 714 may set each item constituting the molding conditions and then perform injection molding.

[0275] The determination unit 715A includes an inference unit 716A, and determines whether an abnormality (defect) has occurred in the molded product based on information acquired from the inference unit 716A.

[0276] The reasoning unit 716A, for example, inputs the waveform data of the last injection (for example, the first molding cycle) and the waveform data of the injection this time (for example, the second molding cycle) representing the results detected by various sensors during the production of the molded product by the injection molding control unit 714 into the input layer of the learned model LM', thereby obtaining a value (for example, an increase or decrease in weight) representing the change in the quality characteristics of the molded product this time (the difference between the molded product of the first molding cycle and the molded product of the second molding cycle) since the last time from the output layer of the learned model LM' in the form of quantitative data.

[0277] Furthermore, the inference unit 716A may pre-process the waveform data before inputting the waveform data to the input layer.

[0278] Furthermore, the determination unit 715A determines whether an abnormality has occurred based on the weight of the molded product (an example of a quality characteristic) received from the inference unit 716A. The criterion for determining whether an abnormality has occurred can also be set by the user. For example, when the increase or decrease in weight compared to the previous time is greater than a predetermined criterion (for example, a change of 0.4% or more), the determination unit 715A can also determine that an abnormality has occurred.

[0279] The display control unit 717 displays information on the display device 760. For example, the display control unit 717 may display the determination result of the determination unit 715A in the performance list of the log information screen. The determination result may be displayed in any manner, and may be displayed in the same manner as in the first embodiment.

[0280] In this embodiment, the waveform data of the last injection is input into the channel CH_A of the input layer, and the waveform data of the current injection is input into the channel CH_B of the input layer. However, the corresponding relationship between the channels of the input layer and the input waveform data is not limited, and it can also be designed by any method. In addition, it is not limited to the method of setting channels corresponding to waveform data in order to input multiple waveform data, and multiple waveform data can be input into the input layer at will.

[0281] In the present embodiment, an example of determining whether an abnormality is present based on the waveform data of the last injection and the waveform data of this injection and based on the amount of change since the last injection is described. However, the present embodiment is not limited to the method of determining whether an abnormality is present based on the amount of change since the last injection. For example, the user may also select the waveform data of the injection that serves as a benchmark. At this time, in the learned model LM', the waveform data of the injection (serving as a benchmark) selected by the user is input to the channel CH_A of the input layer, and the waveform data of this injection is input to the channel CH_B of the input layer. Moreover, the learned model LM' outputs the amount of change in the quality characteristics of the molded product relative to the injection that serves as the benchmark from the output layer. Thus, it is possible to determine whether an abnormality is present based on the difference from the molded product that serves as the benchmark.

[0282] In this embodiment, an example of inputting waveform data of two shots is described. However, this embodiment is not limited to the example of inputting waveform data of two shots. The inference unit 716A may input waveform data of three or more shots to the learned model LM' and receive the difference in quality characteristics between three or more molded products. Furthermore, the determination unit 715A may also perform abnormality determination based on the difference in quality characteristics output from the learned model LM'.

[0283] In addition, the present embodiment is not limited to the method of learning only by the learning device 1300 , and additional learning may be performed on the learned model LM′ stored in the control device 700A of the injection molding machine 10 .

[0284] In the above embodiment, an example of determining abnormality (defect) of a molded product using the learned models LM and LM′ mounted on the injection molding machine 10 is described. However, the above embodiment is not limited to the method of determining abnormality (defect) of a molded product in the injection molding machine 10.

[0285] (Variation Example)

[0286] In the above embodiment, an example is described in which the control device 700, 700A of the injection molding machine 10 uses the learned model to perform abnormality determination as a monitoring device of the injection molding machine 10. However, the above embodiment is not limited to the method in which the control device 700, 700A of the injection molding machine 10 uses the learned model to perform abnormality determination. Therefore, in a modified example, the group management device that controls the injection molding machine 10 can also determine the abnormality of the molded product produced by the injection molding machine 10.

[0287] For example, the group management device manages a plurality of injection molding machines 10. In the storage device of the group management device, a learned model storage unit 722 or a learned model storage unit 722A is provided. In addition, in the CPU of the above-mentioned embodiment, a determination unit 715 or a determination unit 715A is provided. In addition, the group management device receives waveform data from a plurality of injection molding machines 10, and determines abnormalities of molded products in the same manner as in the above-mentioned embodiment based on the received waveform data.

[0288] Similar to the above-mentioned embodiment, the group management device may also perform processing based on the result of abnormality determination. For example, the group management device may display the result of abnormality determination. As another example, when it is determined that an abnormality has occurred, the group management device may also instruct the injection molding machine 10 to stop, etc.

[0289] <Function>

[0290] In the above-mentioned embodiment and modification, when waveform data is input to the input layer of the learned model LM, LM', a value representing the evaluation of the molded product as quantitative data is obtained, and it is determined whether an abnormality has occurred based on the obtained value. Therefore, when monitoring the waveform data, it is not the waveform data itself that is monitored, but the quality characteristics that change according to the waveform data are monitored, thereby making it easy to intuitively grasp whether an abnormality has occurred.

[0291] That is, when waveform data is monitored, basic statistics or cumulative values ​​such as the maximum value, minimum value, and average value of the waveform data are monitored. However, it is difficult for the user to set any benchmark for the statistics or cumulative values ​​of the waveform data to detect whether the quality is abnormal. In contrast, in the above-mentioned embodiment and modified example, when waveform data is input into the learned model, the presence or absence of abnormality is detected based on the value of the quality characteristics of the molded product output from the learned model as quantitative data. Therefore, the user can easily and intuitively grasp the benchmark for determining abnormality.

[0292] Furthermore, in the above-mentioned embodiment, the waveform data is input into the learned model, and the abnormality is determined based on the quality characteristics of the molded product, etc. output from the learned model. That is, the learned model outputs the result that takes into account various characteristics included in the waveform data as the quality characteristics of the molded product, etc. Therefore, the abnormality determination based on the quality characteristics of the molded product, etc. can be regarded as a determination based on the results of various studies (understanding) of the waveform data. Therefore, the above-mentioned embodiment makes it possible to determine the abnormality taking into account the complex characteristics included in the waveform data.

[0293] Furthermore, in the above-mentioned embodiment, a machine learning model is used as training data for establishing a correspondence between waveform data and quality characteristics of molded products, so even if the user cannot recognize the correspondence between waveform data and quality characteristics, it is possible to recognize whether an abnormality has occurred in the waveform data by referring to the quality characteristics. In other words, it is possible to determine an abnormality based on the correspondence between quality characteristics and changes in waveform data that are difficult for users to understand.

[0294] Furthermore, when changing the judgment criteria for whether an abnormality has occurred, it is only necessary to adjust the threshold value (numerical value) relative to the quality characteristic represented by the quantitative data, so it can be flexibly adjusted according to the quality of the molded product, etc. Therefore, it is possible to make flexible adjustments when making abnormality judgments, thereby achieving an improvement in the accuracy of abnormality judgments.

[0295] The above is an explanation of the embodiment of the monitoring device for the injection molding machine involved in the present invention, but the present invention is not limited to the above embodiment, etc. Various changes, corrections, replacements, additions, deletions and combinations can be made within the scope of the technical solution. Of course, these also fall within the technical scope of the present invention.

Claims

1. A monitoring device for an injection molding machine, comprising: an acquisition unit that acquires first data representing a detection result from a detection unit provided in the injection molding machine in a time series; an inference unit that inputs the first data into a learned model that has machine-learned training data that combines second data that represents a detection result from the detection unit in a time series and a second value that represents an evaluation of the product produced by the injection molding machine, thereby acquiring a first value that represents an evaluation of the product from the learned model; and The determination unit determines whether an abnormality has occurred in the detection result of the detection unit based on the first value acquired by the inference unit.

2. The monitoring device for an injection molding machine according to claim 1, wherein: The first value and the second value are quantitative data.

3. The monitoring device for an injection molding machine according to claim 2, wherein: The determination unit determines whether an abnormality has occurred in the detection result of the detection unit based on whether the first value is larger than a preset threshold value.

4. The monitoring device for an injection molding machine according to claim 3, wherein: The threshold value used by the determination unit for determination is set according to a unit of measurement of a determination object.

5. The monitoring device for an injection molding machine according to claim 2, wherein: The acquisition unit acquires the first data in a second molding cycle of the injection molding machine after acquiring the first data in a first molding cycle of the injection molding machine. The inference unit inputs the first data of each of the first molding cycle and the second molding cycle into the learned model of training data that has been machine-learned by combining the second data in the third cycle, the second data in the fourth cycle, and the second value of the difference in evaluation between the product of the third cycle and the product of the fourth cycle represented by quantitative data, thereby obtaining the first value of the difference in evaluation between the product of the first molding cycle and the product of the second molding cycle represented by quantitative data.

6. The monitoring device for an injection molding machine according to any one of claims 1 to 5, wherein: A display control unit is further provided, and the display control unit is configured to display one or more of a determination result based on the determination unit and the first value of the product on a display device every time the product is produced by the injection molding machine.

Images