JP7424191B2 - Molding system, abnormality prediction device, abnormality prediction method, program and learned model - Google Patents

Description

The present invention relates to a molding system, an abnormality prediction device, an abnormality prediction method, a program, and a learned model.

2. Description of the Related Art Injection molding apparatuses are known that mold a molded product by supplying a melt containing a molding material to a cavity formed between a plurality of molds. For example, Patent Document 1 discloses an injection molding apparatus that injects a molding material into a mold.

Patent Document 1 discloses that, as a countermeasure against molding defects caused by residual gas in the cavity (gas residual defects), when a gas residual defect is confirmed through a quality check, the forward position of the screw is retracted by a predetermined amount. Change the condition settings. In this manner, Patent Document 1 discloses a technique for changing molding conditions so that the gas discharge ability is superior to the amount of resin filled into the cavity.

JP 2015-24568 Publication

In injection molding, the same mold is used repeatedly, so the mold wears out with each molding cycle. Since wear of the mold affects the quality of the molded product, currently the mold is replaced after a certain number of moldings, or the injection molding time is determined based on intuition and tips based on the skill level of the injection molding worker. The mold is being replaced. Therefore, if the mold is replaced at the wrong timing, the molded product may be defective or the number of man-hours may increase due to unnecessary mold replacement.

In addition, when deciding when to replace a mold based on the operator's intuition and tips, the quality of the molded product should be checked to determine whether the molded product is defective due to mold wear or has a tendency to become defective. After assessing the situation, we decide to replace the mold. Since the quality check is performed by visual inspection by the worker, the accuracy of the quality check largely depends on the skill level of the worker. In addition, since it requires a lot of man-hours for the worker, the burden placed on the inspection is also large. Therefore, there is a need for a technology that automatically determines the appropriate timing for replacing the mold, regardless of the skill level of the operator.

Therefore, an object of the present invention is to provide a molding system, an abnormality prediction device, an abnormality prediction method, a program, and a learned model that predict the wear state of a gate.

(1) A molding system according to the present invention includes a molding device that molds a molded product and an abnormality prediction device that predicts abnormalities in the molding device, the molding device having a cavity inside. a mold part having a gate that opens to the cavity side; a filling operation of filling the molding material in a molten state into the cavity via the gate; and a pressure of the molding material filled in the cavity. and a pressure sensor that detects the pressure of the molding material in the cavity; The abnormality prediction device includes a data acquisition unit that acquires a maximum value of pressure detected by the pressure sensor between the pressure holding operation and the pressure release release operation as a first evaluation value; and an abnormality prediction unit that acquires prediction information for predicting the wear state of the gate based on the above.

As a result of extensive research, the inventors have discovered that the maximum pressure during the pressure holding process of the molding material in the cavity is greater when the gate is abnormal than when the gate is normal. In other words, if we focus on the maximum value of the pressure of the molding material in the cavity during the pressure holding process, we can detect abnormalities due to wear of the gate, as well as abnormalities in the molded product caused by such abnormalities (e.g., voids, sink marks, warpage, dimensional differences, etc.). ) was found to be able to be predicted. Therefore, the molding system according to the present invention acquires prediction information for predicting the wear state of the gate based on the maximum value of the pressure of the molding material in the cavity between the holding pressure operation and the holding pressure release operation. . This makes it possible to predict the wear state of the gate from the information acquired by the pressure sensor provided in the molding device.

(2) Preferably, the abnormality prediction unit determines that the molded product to be predicted is higher than the first reference value generated based on the first evaluation value obtained when the wear state of the gate is normal. When the first evaluation value acquired during molding becomes large, the prediction information indicating that wear has occurred in the gate is acquired. With this configuration, the wear state of the gate can be easily predicted by comparing the first reference value and the first evaluation value.

(3) Preferably, the abnormality prediction unit molds a plurality of the molded products to be predicted by the molding apparatus, and the first evaluation value continuously acquired over a predetermined plurality of times is The predicted information indicating that wear has occurred in the gate is acquired when the predicted value is greater than one reference value. With this configuration, it is possible to exclude the influence of sudden disturbances and predict the wear state of the gate more accurately.

(4) Preferably, the injection section includes a cylinder that stores the molding material in a molten state and connects to the mold section, and a cylinder that is inserted into the cylinder and supplies the molding material in a molten state in the direction of the mold section. the molding device further includes a second pressure sensor that detects the pressure of the molding material in the cylinder or the pressure that the screw receives from the molding material, and the molding device further includes a second pressure sensor that detects the pressure of the molding material in the cylinder, and the data acquisition unit acquires a first reference value based on the maximum value of the pressure detected by the second pressure sensor between the pressure holding operation and the pressure releasing operation, and the abnormality prediction unit The prediction information is acquired based on the difference or ratio between the evaluation value and the first reference value.

With this configuration, both the first reference value and the first evaluation value are obtained when the molded product to be predicted is molded once, regardless of whether the wear state of the gate is normal or not. be able to. Therefore, for example, even if the molding conditions (pressurization, mold clamping force, type of molding material, etc.) are changed while using the same mold part, the first standard can be easily adjusted after the change. It becomes possible to obtain the value.

(5) Preferably, the abnormality prediction unit inputs the first evaluation value into a trained model in which a correlation between the first evaluation value and the wear state of the gate is machine-learned, so that the prediction information is obtained, the explanatory variables of the learned model include the first evaluation value, and the target variables of the learned model include the inner diameter of the gate or the amount of wear of the gate. With this configuration, the wear state of the gate can be predicted more accurately even when there are variations in the molding conditions.

(6) Preferably, the data acquisition unit acquires a second evaluation value regarding a change in pressure of the molding material after the holding pressure release operation, based on time series data of the pressure detected by the pressure sensor. , the abnormality prediction unit acquires one or more pieces of prediction information based on the first evaluation value and the second evaluation value.

As a result of intensive research, the inventors found that the degree of pressure change (pressure slope) after the holding pressure release operation of the molding material in the cavity is larger in the negative direction when the gate is abnormal than when the gate is normal. I discovered that. In other words, if we focus on the change in pressure of the molding material in the cavity after the holding pressure is released, we can detect abnormalities due to wear of the gate, and also abnormalities in the molded product caused by such abnormalities (e.g., voids, sink marks, warpage, dimensional differences, etc.). ) was found to be able to be predicted. Therefore, the molding system according to the present invention provides both a first evaluation value based on the maximum pressure during the pressure holding process and a second evaluation value based on the change in the pressure of the molding material in the cavity after the pressure release operation. Based on this, prediction information for predicting the wear state of the gate is obtained. This makes it possible to more accurately predict the wear state of the gate from the information acquired by the pressure sensor provided in the molding device.

(7) The abnormality prediction device according to the present invention is an abnormality prediction device that predicts abnormalities in a molding device that molds a molded product, and the molding device has a cavity formed inside and a gate that opens toward the cavity side. a filling operation for filling the molding material in a molten state into the cavity via the gate, a pressure holding operation for maintaining the pressure of the molding material filled in the cavity, and the molding an injection unit that performs a holding pressure release operation to release the holding of the pressure of the material; and a pressure sensor that detects the pressure of the molding material in the cavity, and the abnormality prediction device a data acquisition unit that acquires, as a first evaluation value, a maximum value of the pressure detected by the pressure sensor between the time and the holding pressure release operation; and a data acquisition unit that predicts a wear state of the gate based on the first evaluation value. An abnormality prediction device that includes an abnormality prediction unit that acquires prediction information for.

The abnormality prediction device according to the present invention acquires prediction information for predicting the wear state of the gate based on the maximum value of the pressure of the molding material in the cavity between the pressure holding operation and the holding pressure release operation. This makes it possible to predict the wear state of the gate from the information acquired by the pressure sensor provided in the molding device.

(8) The abnormality prediction method according to the present invention includes a mold part that has a cavity formed inside and a gate that opens toward the cavity, and a molding material in a molten state that is filled into the cavity via the gate. an injection unit that performs a filling operation to hold the pressure of the molding material filled in the cavity, a pressure holding operation to hold the pressure of the molding material, and a holding pressure release operation to release the pressure of the molding material. An abnormality prediction method for predicting an abnormality in a molding device for molding, the pressure detected by a pressure sensor that detects the pressure of the molding material in the cavity between the holding pressure operation and the holding pressure release operation. Abnormality prediction comprising: a data acquisition step of acquiring a maximum value of as a first evaluation value; and an abnormality prediction step of acquiring prediction information for predicting the wear state of the gate based on the first evaluation value. It's a method.

The abnormality prediction method according to the present invention acquires prediction information for predicting the wear state of the gate based on the maximum value of the pressure of the molding material in the cavity between the pressure holding operation and the holding pressure release operation. This makes it possible to predict the wear state of the gate from the information acquired by the pressure sensor provided in the molding device.

(9) The program according to the present invention includes a mold part that has a cavity formed inside and a gate that opens toward the cavity, and a filling that fills the molding material in a molten state into the cavity via the gate. an injection unit that performs a pressure holding operation that holds the pressure of the molding material filled in the cavity, and a holding pressure release operation that releases the holding of the pressure of the molding material. A program for predicting an abnormality in a molding device, wherein the program detects the maximum pressure detected by a pressure sensor that detects the pressure of the molding material in the cavity between the pressure holding operation and the holding pressure release operation. causing a computer device to execute a data acquisition step of acquiring the value as a first evaluation value; and an abnormality prediction step of acquiring prediction information for predicting the wear state of the gate based on the first evaluation value. It is a program.

The program according to the present invention executes a process in which a computer acquires prediction information for predicting the wear state of the gate based on the maximum value of the pressure of the molding material in the cavity from the holding pressure operation to the holding pressure release operation. Let the device execute it. This makes it possible to predict the wear state of the gate from the information acquired by the pressure sensor provided in the molding device.

(10) The trained model according to the present invention includes a mold part that has a cavity formed inside and has a gate that opens to the cavity side, and fills the molding material in a molten state into the cavity via the gate. an injection unit that performs a filling operation to hold the pressure of the molding material filled in the cavity, a pressure holding operation to hold the pressure of the molding material, and a holding pressure release operation to release the pressure of the molding material. A trained model for predicting abnormalities in a molding device that molds a molding device, wherein the explanatory variable is a pressure that detects the pressure of the molding material in the cavity between the holding pressure operation and the holding pressure release operation. A learned model including a first evaluation value obtained based on a maximum value of pressure detected by a sensor, and a target variable including an inner diameter of the gate or an amount of wear of the gate.

In the trained model according to the present invention, the explanatory variable is the first evaluation value regarding the maximum value of the pressure of the molding material in the cavity from the holding pressure operation to the holding pressure release operation, and the objective variable is the inner diameter of the gate or the gate inner diameter. Since it is the amount of wear, if the first evaluation value is input to the learned model, the predicted wear state of the gate will be output. Thereby, the wear state of the gate can be predicted using the first evaluation value obtained based on the detection value of the pressure sensor.

According to the present invention, it is possible to predict the wear state of the gate of a molding device.

FIG. 1 is a block diagram schematically showing a molding system according to a first embodiment. 2 is an explanatory diagram conceptually showing the molding apparatus according to FIG. 1. FIG. 2 is an explanatory diagram conceptually showing the molding apparatus according to FIG. 1. FIG. FIG. 3 is a cross-sectional view of the mold section schematically showing the state at the end of the filling process. FIG. 5 is a cross-sectional view of the mold section taken along the cutting line of arrow V in FIG. 4; FIG. 5 is a cross-sectional view showing the mold section taken along the cutting line of arrow VI in FIG. 4; This is an example of a graph showing time series data of pressure. FIG. 3 is an explanatory diagram schematically showing the state at the start and end of the pressure holding process when the gate is normal. FIG. 3 is an explanatory diagram schematically showing the state at the start and end of a pressure holding process when a gate abnormality occurs. 1 is a block diagram showing a functional configuration of a learning device according to a first embodiment. FIG. It is a graph schematically explaining evaluation values according to the first embodiment. 1 is a block diagram showing a functional configuration of an abnormality prediction device according to a first embodiment. FIG. FIG. 2 is a block diagram schematically showing a molding system according to a second embodiment. It is a block diagram showing the functional composition of the abnormality prediction device concerning a 2nd embodiment. 12 is a graph schematically illustrating an abnormality determination method according to a modification.

<First embodiment>
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.

<Overall configuration of molding system>
FIG. 1 is a block diagram schematically showing a molding system 10 according to the first embodiment. The molding system 10 includes a plurality of molding devices 20, a learning device 30, an abnormality prediction device 40, an input section 50, and a display section 60.

The molding device 20, the learning device 30, the abnormality prediction device 40, the input section 50, and the display section 60 are each provided to be able to communicate wirelessly or by wire. The learning device 30 and the abnormality prediction device 40 are configured by an information processing device (computer device) having a calculation unit (eg, CPU, GPU, etc.) and a storage unit (eg, HDD, SSD, etc.). The learning device 30 and the abnormality prediction device 40 may be configured by the same information processing device, or may be configured by separate information processing devices.

In this embodiment, a plurality of molding devices 20 are connected to one learning device 30 and one abnormality prediction device 40, and the learning device 30 and the abnormality prediction device 40 respond to various data transmitted from the plurality of molding devices 20. Based on this, learning and anomaly prediction are performed. Note that the molding device 20 may have a one-to-one correspondence with the learning device 30 and the abnormality prediction device 40. That is, in the molding system 10, there may be one molding device 20, or a plurality of learning devices 30 and abnormality prediction devices 40 may be provided.

The input unit 50 is, for example, a keyboard or a mouse, and accepts various inputs from an operator. The display unit 60 is, for example, a display or a speaker, and displays various information on the molding system 10. The input section 50 and the display section 60 may be integrated, for example, like a touch panel. Further, the input unit 50 and the display unit 60 may be provided as portable terminal devices so as to be movable at a location away from the molding device 20, the learning device 30, and the abnormality prediction device 40.

<Schematic configuration of molding device>
2 and 3 are explanatory diagrams conceptually showing the molding apparatus 20. FIG. In FIGS. 2 and 3, portions shown as cross sections are hatched. The molding apparatus 20 includes a bed 21 , an injection section 22 , a mold clamping section 23 , a mold section 24 , a pressure sensor 25 , a temperature sensor 26 , and a control panel 27 . FIG. 2 shows the molding apparatus 20 with the mold part 24 open, and FIG. 3 shows the molding apparatus 20 with the mold part 24 assembled. The molding device 20 is a device that performs mold clamping type injection molding.

The control panel 27 includes a control section 271 and a communication section 272. The control unit 271 is electrically connected to each drive unit (motor 237, etc.) of the molding device 20, and outputs an operation command to each drive unit. Further, the control section 271 is electrically connected to each sensor (pressure sensor 25, etc.) of the molding device 20, and signals detected by the respective sensors are input to the control section 271. The control unit 271 is constituted by an information processing device having a calculation unit (eg, CPU, GPU, etc.) and a storage unit (eg, HDD, SSD, etc.).

The communication unit 272 communicates with other units of the molding system 10 (learning device 30, etc.). For example, the communication unit 272 transmits the signals detected by the respective sensors to the learning device 30 or the abnormality prediction device 40. The communication unit 272 also receives determination information and prediction information, which will be described later, from the abnormality prediction device 40.

The mold clamping section 23 includes a fixed platen 231, a movable platen 232, a tie bar 233, a ball screw 234, a support plate 235, a mold clamping force sensor 236, and a motor 237. The fixed plate 231 and the support plate 235 are fixed to the bed 21. The support plate 235 supports the ball screw 234. Ball screw 234 is connected to motor 237. When the motor 237 is rotated by an operation command from the control unit 271, the ball screw 234 moves. A movable platen 232 is fixed to an end of the ball screw 234 opposite to the end connected to the motor 237.

Here, in the molding device 20, the direction in which the ball screw 234 moves is referred to as the "axial direction." The side on which the motor 237 is located relative to the ball screw 234 is referred to as "one side" in the axial direction, and the side on which the movable platen 232 is located relative to the ball screw 234 is referred to as "the other side" in the axial direction.

The movable platen 232 moves in the axial direction as the ball screw 234 moves. The movable platen 232 is formed with a through hole 232a that penetrates in the axial direction. The tie bar 233 has one end in the axial direction fixed to the support plate 235 and the other end in the axial direction fixed to the fixed plate 231. The tie bar 233 is inserted into the through hole 232a of the movable platen 232. Thereby, the tie bar 233 guides the movement of the movable platen 232 in the axial direction.

The mold clamping force sensor 236 detects the pressure (reaction force of the mold clamping force) applied from the ball screw 234 to the support plate 235. The mold clamping force sensor 236 outputs a detection signal related to pressure to the control unit 271. Note that the mold clamping force sensor 236 may be installed at any other position as long as it can detect the mold clamping force in the mold section 24, which will be described later. The fixed platen 231 is formed with a through hole 231a whose diameter increases toward the other side in the axial direction. A cylinder 222, which will be described later, is inserted into the through hole 231a.

The mold section 24 has a plurality of molds 241 and 242. The mold 241 is fixed to the movable platen 232. When the ball screw 234 moves in the axial direction, the mold 241 also moves in the axial direction together with the movable platen 232. That is, the mold 241 is a movable mold. The mold 242 is fixed to the fixed platen 231. That is, the mold 242 is a fixed mold. A flow path 243 is formed in the mold 242 .

See FIG. 3. When the mold 241 is moved to the other side in the axial direction by the ball screw 234 and the mold 241 comes into contact with the mold 242 (that is, when the plurality of molds 241 and 242 are combined), a gap is created between the molds 241 and 242. A cavity C1 is formed. The cavity C1 of this embodiment is an annular space filled with molding material.

See FIG. 2. The injection unit 22 includes a hopper 221, a cylinder 222, a screw 223, a ball screw 225, a motor 226, a pressurization sensor 227, a pressure sensor 227a, a movement sensor 228, and a heater 229. The hopper 221 is connected to the cylinder 222 and supplies molding material into the cylinder 222. The cylinder 222 is a hollow cylindrical member extending in the axial direction. The diameter of the end of the cylinder 222 on one axial side becomes narrower as it approaches the end on the one side in the radial direction, and a nozzle 224 is provided at the end of the cylinder 222 on the one side in the axial direction. The nozzle 224 is connected to a flow path 243 of the mold 242.

The screw 223 is inserted into the cylinder 222 from the other end of the cylinder 222 in the axial direction. A ball screw 225 is connected to the other axial side of the screw 223, and a motor 226 is connected to the other axial side of the ball screw 225. When the motor 226 rotates according to an operation command from the control unit 271, the ball screw 225 moves in the axial direction. Along with this, the screw 223 also moves in the axial direction. At this time, the screw 223 rotates in the circumferential direction with the axial direction as the central axis.

The pressurization sensor 227 detects the pressure (reaction force of the pushing force of the screw 223) applied from the ball screw 225 to the motor 226. That is, the pressurization sensor 227 detects the pressure that the screw 223 receives from the molding material L1. The pressurization sensor 227 outputs a detection signal related to pressure to the control unit 271. Note that the pressurization sensor 227 may be installed at any other position as long as it can detect the pushing force of the screw 223.

The pressure sensor 227a is installed on the inner peripheral surface of the cylinder 222, and detects the pressure that the cylinder 222 receives from the molding material L1. The pressure sensor 227a outputs a detection signal related to pressure to the control unit 271. At least one of the pressurization sensor 227 and the pressure sensor 227a functions as the "second pressure sensor" of the present invention.

The movement amount sensor 228 detects the amount of movement of the ball screw 225 in the axial direction. The movement amount sensor 228 outputs a detection signal regarding the movement amount to the control unit 271. Note that the movement amount sensor 228 may be installed at any other position as long as it can detect the amount of movement of the ball screw 225 in the axial direction.

The heater 229 is, for example, a resistance heater made of a resistance wire wound into a coil. The heater 229 heats the inside of the cylinder 222 with resistance heat when a current is passed through the resistance wire according to an operation command from the control unit 271 .

The pressure sensor 25 (the "pressure sensor" of the present invention) is installed on the surface of the molds 241 and 242 that is exposed to the cavity C1. More specifically, the pressure sensor 25 is installed on a surface exposed to the cavity C1 that is close to a gate 243b (FIG. 4), which will be described later. Pressure sensor 25 detects the pressure within cavity C1. In particular, the pressure sensor 25 detects the pressure of the molding material (in a molten state, a solidified state, or a mixed state of a molten state and a solidified state) supplied into the cavity C1. The pressure sensor 25 outputs a detection signal related to pressure to the control unit 271. In this embodiment, one pressure sensor 25 is installed in the mold 242. However, a plurality of pressure sensors 25 may be installed in both molds 241 and 242, respectively.

The temperature sensor 26 is built into the mold 241 and detects the temperature of the mold 241. The temperature sensor 26 outputs a detection signal related to temperature to the control unit 271. Note that the temperature sensor 26 may be installed in a region of the mold 241 facing the cavity C1, or may be installed in the mold 242. Additionally, the temperature sensor 26 may be installed within the cylinder 222. That is, the temperature sensor 26 only needs to be able to directly or indirectly detect the temperature of the molding material supplied into the cavity C1.

<Manufacturing method using molding equipment>
A method for manufacturing a molded product using the molding apparatus 20 will be described with reference to FIGS. 2 to 5 as appropriate. The method for manufacturing a molded product using the molding apparatus 20 includes a pre-process ST1, a mold clamping process ST2, a filling process ST3, a holding pressure process ST4, a holding pressure release process ST5, and a mold release process ST6 in this order. executed. In this embodiment, the molded product is a resin retainer used in a rolling bearing. However, this is an example of a molded product, and the molded product molded by the molding apparatus according to the present invention may have other shapes and uses.

See FIG. 2. First, a pre-process ST1 is executed. In the pre-process ST1, the screw 223 is rotated by the motor 226, and pellets of molding material are supplied from the hopper 221 into the cylinder 222 while the inside of the cylinder 222 is heated by the heater 229. The molding material pellets are melted in the cylinder 222 by frictional heat caused by the rotation of the screw 223 and heating by the heater 229, and become a molten molding material L1. When a predetermined amount of molding material L1 is stored in the cylinder 222, the pre-process ST1 ends.

Next, when the mold clamping process ST2 is started, in the molding apparatus 20 in the state shown in FIG. 241 is brought into contact with the mold 242. With the molds 241 and 242 combined in this manner, the ball screw 234 further presses the mold 241 against the mold 242 in the other axial direction with a predetermined mold clamping force. That is, the plurality of molds 241 and 242 are tightened. Thereby, an annular cavity C1 is formed between the plurality of molds 241 and 242. With the above, the mold clamping step ST2 is completed.

Here, the mold clamping force is one of the molding conditions, and is determined according to other molding conditions such as the shapes of the molds 241 and 242. The mold clamping force is detected by a mold clamping force sensor 236.

Subsequently, when the filling process ST3 is started, the ball screw 225 moves to one side in the axial direction while maintaining the above mold clamping force. As a result, the screw 223 pushes the molding material L1 to one side in the axial direction, and the molten molding material L1 is injected from the nozzle 224 of the cylinder 222 into the cavity C1 via the flow path 243 of the mold 242 (filling operation). .

FIG. 4 is a cross-sectional view of the mold part 24 schematically showing the state at the end of the filling step ST3. FIG. 5 is a cross-sectional view of the mold part 24 taken along the cutting line indicated by arrow V in FIG. FIG. 6 is a cross-sectional view of the mold part 24 taken along the cutting line of arrow VI in FIG.

See FIG. 4. The flow path 243 has a first opening 243a that opens toward the nozzle 224 and a second opening 243b that opens toward the cavity C1. Since the second opening 243b functions as a gate for supplying the molding material L1 from the flow path 243 into the cavity C1, it is hereinafter referred to as a "gate 243b."

See FIG. 5. In this embodiment, the cavity C1 is formed in an annular shape. Moreover, in this embodiment, the pressure sensor 25 is installed so as to face the region of the cavity C1 closest to the gate 243b. For example, when the cavity C1 is viewed in the axial direction as shown in FIG. 5, the pressure sensor 25 is installed close to the direction toward the gate 243b (in the radial direction of the cavity C1 in this embodiment).

See FIG. 6. In this embodiment, the gate 243b has a circular inner peripheral surface. Further, the gate 243b has an inner diameter smaller than the inner diameter of other portions of the flow path 243. With this configuration, it is possible to realize a good gate cut (cutting the connection between the molded product in the cavity C1 portion and the runner in the flow path 243 portion) when molding the molded product. On the other hand, since the gate 243b has a smaller inner diameter than other parts of the flow path 243, it has a characteristic of being easily worn out. Wear of the gate 243b particularly affects the quality of the molded product, so when the gate 243b wears out to some extent, it becomes necessary to replace the mold part 24.

As shown in FIGS. 4, 5, and 6, the molding material L1 is supplied from the flow path 243 to the cavity C1, and when the inside of the cavity C1 is completely filled with the molding material L1, the filling step ST3 is completed. In the filling step ST3, the molten molding material L1 is supplied into the cavity C1 from near the surfaces of the molds 241 and 242 while gradually solidifying.

Subsequently, when the pressure holding step ST4 is started, the screw 223 further pushes the molding material L1 to one side in the axial direction, and the molding material L1 is further injected from the nozzle 224 of the cylinder 222 into the cavity C1. As a result, a predetermined pressure (for example, several tens to several hundred MPa) is applied to the molding material L1 filled in the cavity C1. Then, the screw 223 maintains this state for a predetermined time to continue applying a predetermined pressure to the molding material L1 for a predetermined time (for example, several seconds) (pressure holding operation). The pressure (pressure) with which the screw 223 pushes out the molding material L1 into the cavity C1 is detected by the pressurization sensor 227 and the pressure sensor 227a.

Subsequently, when the holding pressure release step ST5 is started, the screw 223 moves to the other side in the axial direction and releases the pressure holding of the molding material L1 (pressure releasing operation). After the holding pressure release operation, when a predetermined time has elapsed and the pressure of the molding material L1 in the cavity C1 becomes equal to or less than a predetermined value, the holding pressure release step ST5 is completed. Thereafter, when the mold release step ST6 is started, the mold part 24 is cooled, so that the molding material L1 in the cavity C1 hardens, and a molded product is formed. Then, the ball screw 234 moves to one side in the axial direction, and the mold 241 separates from the mold 242, so that the molded product is taken out. Note that cooling of the mold part 24 may be started simultaneously with the holding pressure release step ST5.

FIG. 7 is an example of a graph showing time-series data of the pressure detected by each sensor of the molding device 20 in the filling process ST3, the pressure holding process ST4, and the holding pressure release process ST5. In FIG. 7, the vertical axis is pressure P, and the horizontal axis is time t. The graph line F0 is time series data of the pressure detected by the pressurization sensor 227 ("second time series data" of the present invention). Graph lines F1 and F2 are time series data of pressure detected by the pressure sensor 25 ("time series data" of the present invention).

Graph line F1 shows the pressure sensor when there is no wear on the gate 243b, or when there is wear, the wear is to the extent that it does not adversely affect the quality of the molded product (hereinafter referred to as "gate normal"). This is time-series data of pressure detected by No. 25. For example, the graph line F1 is time series data of the pressure detected when the mold section 24 performs molding for the first time. Graph line F2 is time series data of the pressure detected by the pressure sensor 25 when there is abnormal wear on the gate 243b (more specifically, when there is wear to the extent that it adversely affects the quality of the molded product).

In this embodiment, the origin of time (t=0) is the point in time when the control unit 271 commands the motor 226 to move the screw 223 to the other side in the axial direction. That is, this is the point in time when the injection section 22 starts supplying the molding material L1 to the mold section 24 in the filling step ST3. Note that the origin of time (t=0) is not limited to this, and may be the time point when the pressurization sensor 227 or the pressure sensor 227a detects a predetermined pressure, or the time point when the movement amount sensor 228 detects a predetermined movement amount. It may be at the point in time.

Focus on graph line F0. The graph line F0 tends to be substantially the same regardless of whether the gate 243b is normal or abnormal, if the molding conditions (mold clamping force, pressurization, pressure holding time, etc.) are the same. The graph line F0 rises to the pressure Ps1 in the filling process ST3.

Here, the pressure Ps1 is the "set pressure" of the pressure holding step ST4. That is, during the pressure holding step ST4, the control unit 271 performs feedback control on the motor 226 so that the pressure detected by the pressurization sensor 227 is maintained at Ps1. Therefore, during the pressure holding step ST4, the pressure detected by the pressurization sensor 227 remains approximately constant at Ps1. The pressure holding step ST4 starts at time X1 and ends at time X2. That is, the pressure holding time is (X2-X1). After that, the pressure decreases in the holding pressure release step ST5. The holding pressure release step ST5 is started from time point X2. That is, the holding pressure release operation is executed at time point X2.

Next, pay attention to the graph line F1. The graph line F1 rises after the graph line F0 rises in the filling process ST3, the pressure rises to the maximum pressure Pt1 in the pressure holding process ST4, the pressure decreases thereafter, and the pressure suddenly increases in the pressure holding process ST5. descend.

Next, pay attention to graph line F2. The graph line F2 rises after the graph line F0 rises in the filling process ST3, the pressure rises to the maximum pressure Pt2 in the pressure holding process ST4, the pressure decreases thereafter, and the pressure suddenly increases in the pressure holding process ST5. descend.

Here, there are two main differences between the graph line F1 (when the gate is normal) and the graph line F2 (when the gate is abnormal). The first point is the maximum pressure in the pressure holding step ST4. The maximum pressure Pt2 of the graph line F2 tends to be larger than the maximum pressure Pt1 of the graph line F1. This is because the inner diameter of the gate 243b becomes larger than normal due to wear, and the flow path resistance at the gate 243b becomes lower than normal, so the pressure applied from the injection part 22 to the molding material L1 in the cavity C1 increases. This is because the maximum pressure increases as a result of no damping. That is, the maximum pressure of the molding material L1 in the cavity C1 in the pressure holding step ST4 increases as the inner diameter of the gate 243b becomes larger (that is, as the gate 243b wears out).

The second point is the degree of pressure change (that is, the slope of pressure) after the pressure holding release operation (that is, after time point X2). The slope of the pressure after the pressure-holding release operation shown by the graph line F2 tends to be larger in the negative direction than the slope of the pressure after the pressure-holding release action shown by the graph line F1. The cause of this will be explained using FIGS. 8 and 9.

FIG. 8 is an explanatory diagram schematically showing the state at the start (time point X1) and end (time point X2) of the pressure holding step ST4 when the gate is normal. FIGS. 8A and 8B are enlarged views showing a region including the gate 243b in FIG. Moreover, FIG. 8(a) shows the state at the time when the filling process ST3 is completed and the pressure holding process ST4 is started. The inner diameter d1 of the gate 243b is a normal inner diameter that does not affect the quality of the molded product. At the start of the pressure holding step ST4 (time point X1), the molding material L1 in the cavity C1 and the flow path 243 is almost in a molten state.

FIG. 8(b) shows the state at the time when the pressure holding process ST4 is completed and the pressure holding release process ST5 is started. During the pressure holding step ST4, heat is removed from the molding material L1 by the mold 242, so that a portion of the molding material L1 solidifies at the edge of the flow path 243, forming a solidified body S1. Then, by the end of the pressure holding step ST4, the solidified body S1 completely covers the gate 243b, and the cavity C1 and the flow path 243 are cut off by the solidified body S1, as shown in FIG. 8(b), in a "gate seal" state. becomes.

After the gate is sealed, the holding pressure release operation is performed, so that the molding material L1 in the cavity C1 solidifies and contracts while remaining in the cavity C1. As a result, the pressure of the molding material L1 in the cavity C1 after the holding pressure release operation decreases rapidly, as shown by the graph line F1 in FIG.

FIG. 9 is an explanatory diagram schematically showing the state at the start (time point X1) and end (time point X2) of the pressure holding step ST4 when the gate is abnormal. 9A and 9B are enlarged views showing the region including the gate 243b in FIG. 5. FIG. Moreover, FIG. 9(a) shows the state at the time when the filling process ST3 is completed and the pressure holding process ST4 is started. The inner diameter d2 of the gate 243b is larger than the normal inner diameter d1, and is an abnormal inner diameter that affects the quality of the molded product. At the start of the pressure holding step ST4 (time point X1), the molding material L1 in the cavity C1 and the flow path 243 is almost in a molten state.

FIG. 9(b) shows the state at the time when the pressure holding process ST4 is completed and the pressure holding release process ST5 is started. During the pressure holding step ST4, heat is removed from the molding material L1 by the mold 242, so that a portion of the molding material L1 solidifies at the edge of the flow path 243, forming a solidified body S1. However, since the diameter d2 is larger than normal, the coagulated body S1 cannot completely cover the gate 243b by the time the pressure holding step ST4 is completed. Therefore, as shown in FIG. 9(b), a gap G1 remains in the solidified body S1 of the gate 243b, and the cavity C1 and the flow path 243 are not cut off by the solidified body S1. Such a state is referred to as a "gate seal incomplete state."

Then, when the holding pressure release operation is performed while the gate sealing is not completed, the molding material L1 in the cavity C1 flows back into the flow path 243 (that is, in the direction of the arrow AR1). Further, the molding material L1 remaining in the cavity C1 solidifies and contracts. As a result, the pressure of the molding material L1 in the cavity C1 after the holding pressure release operation is reduced by the amount of the backflow of the molding material L1 into the flow path 243, as shown by the graph line F2 in FIG. decreases more rapidly than

In this way, if the holding pressure release operation is performed while the gate sealing is not completed, the filling amount of the molding material in the cavity C1 will be reduced by the backflow of the molding material L1 into the flow path 243, so that the molding This has the negative effect of increasing variations in product size, weight, and quality (eg, strength). In addition, when the filling amount of molding material decreases, the molded product may have voids (unintended spaces created inside the molded product), sink marks (unintended dents created on the outer surface of the molded product), warps (unintended deformation of the molded product), etc. There is a risk that such defects may occur. Therefore, it is important to have a technology that can detect abnormalities in the gate 243b in the molding apparatus 20 and automatically determine the abnormality.

As explained above, as a result of intensive research, the inventors found that the maximum pressure of the molding material L1 in the cavity C1 during the pressure holding step ST4 is larger when the gate is abnormal than when the gate is normal (Pt1< Pt2). That is, if we focus on the maximum value of the pressure of the molding material L1 in the cavity C1 during the pressure holding step ST4 (hereinafter appropriately referred to as "first evaluation value R1"), we can see that the abnormality due to wear of the gate, and furthermore, due to the abnormality. It was discovered that it is possible to predict abnormalities in molded products (e.g., voids, sink marks, warpage, dimensional differences, etc.).

In addition, as a result of intensive research, the inventors have found that the degree of pressure change (pressure slope) after the holding pressure release operation of the molding material L1 in the cavity C1 is more negative when the gate is abnormal than when the gate is normal. I discovered that it grows in the direction. In other words, if we focus on the slope of the pressure of the molding material L1 in the cavity C1 after the holding pressure release operation (hereinafter referred to as the "second evaluation value R2"), we can see that the abnormality is caused by the wear of the gate, and by extension the abnormality. We have discovered that it is possible to predict abnormalities in molded products (e.g., voids, sink marks, warpage, dimensional differences, etc.).

Therefore, in the molding system 10 according to the present embodiment, the learning device 30 generates a learned model Tm1 in which the correlation between the first evaluation value R1, the second evaluation value R2, and the wear state of the gate 243b is learned. , the abnormality prediction device 40 acquires prediction information for predicting the wear state of the gate 243b based on the learned model Tm1, the first evaluation value R1, and the second evaluation value R2. The learning device 30 and the abnormality prediction device 40 will be explained below.

<Description of learning device>
FIG. 10 is a block diagram showing the functional configuration of the learning device 30 according to this embodiment. The learning device 30 includes a training data acquisition section 31 , a learning calculation section 32 , a shaping information storage section 33 , and a learned model storage section 34 . Each of these units is realized by a computer device having a calculation unit such as a CPU and a storage unit such as an HDD.

The molding information storage section 33 stores various molding information. The molding information is, for example, information in a table format in which various types of first information and second information are associated with each other. For example, when the first information is the type of mold, the second information includes various dimensions of the mold and the volume of the cavity C1. When the first information is the type or lot number of the molding material, the second information includes the physical properties of the molding material (viscosity, moisture content, etc.).

The training data acquisition unit 31 acquires information regarding training data from each part of the molding system 10. The training data includes a first evaluation value R1, a second evaluation value R2, a third evaluation value R3, molding information, and wear information of the gate 243b, which will be described later. The training data is acquired based on data detected by each part of the molding system 10 (for example, the pressure sensor 25) when molding a molded product to be studied, for example.

For example, the training data acquisition unit 31 sets the maximum pressure of the molding material L1 in the cavity C1 during the pressure holding step ST4 as the first evaluation value R1 based on the pressures detected by the pressure sensor 25 and the pressurization sensor 227, respectively. get. Specifically, based on the fact that the pressure detected by the pressurization sensor 227 is the set pressure Ps1, the start time X1 and the end time X2 of the pressure holding step ST4 are acquired. Then, the maximum value Ptn of the pressure detected by the pressure sensor 25 between time point X1 (pressure holding operation time point) and time point X2 (pressure holding release operation time point) is acquired as the first evaluation value R1.

The training data acquisition unit 31 also performs a second evaluation regarding the change (inclination) in the pressure of the molding material L1 after time X2 (the time of the holding pressure release operation) based on the time series data of the pressure detected by the pressure sensor 25. Obtain the value R2.

FIG. 11 is a graph schematically explaining the second evaluation value R2 according to the present embodiment. FIG. 11A is a graph showing enlarged graph lines F1 and F2 after time X2 in FIG. First, time points are divided at predetermined time intervals T1 from X2. For example, time X3 is the time when X2+T1, and time X4 is the time when X3+T1. In FIG. 11(a), the time points are divided into four time points X3 to X6, but the time points may be divided into more time points (for example, 10 times).

Then, the rate of change in graph lines F1 and F2 from time point X2 to time point X3 is obtained as the slope of each pressure. For example, as shown in FIG. 11(a), in the case of graph line F1, the pressure at time X2 is Pa1 and the pressure at time X3 is Pa2, so the pressure slope dPa1 is (Pa2-Pa1)/( X3-X2).

FIG. 11(b) is a graph plotting the slope of pressure acquired at predetermined time intervals for graph lines F1 and F2. The vertical axis of the graph is the slope of pressure, and the horizontal axis is time. In FIG. 11(b), the slope of the pressure obtained from the graph line F1 is plotted by a black circle, and the slope of the pressure obtained from the graph line F2 is plotted by a white circle. In FIG. 11A, since the time point is divided into four time points after time point X2, four pressure slopes are obtained for each graph line F1 and F2. For example, the pressure slope dPa1 of the graph line F1 from time point X2 to time point X3 is plotted on time point X2 in FIG. 11(b).

Then, the average value (or median value) of the slopes of these plural pressures is obtained as the second evaluation value R2. For example, if the time-series data of the pressure detected by the pressure sensor 25 when molding the molded product to be learned is the graph line F1, the average value of the slopes of four pressures for the graph line F1 indicated by a black circle. dPv1 is acquired as the second evaluation value R2. Similarly, if the time-series data of the pressure detected by the pressure sensor 25 when molding the molded product to be learned is the graph F2, the average value of the four pressure slopes for the graph line F2 indicated by the white circle. dPv2 is acquired as the second evaluation value R2.

In FIG. 11A and FIG. 7, pressure time series data is displayed as a curve, but in reality, pressure time series data is composed of a plurality of points. Then, "spike noise" may occur in which a point at a predetermined point in time takes an extremely different value than other adjacent points. Therefore, if the second evaluation value R2 is obtained only based on the pressure value at the time when spike noise occurs, an accurate value cannot be obtained. In this embodiment, in the time-series data of the pressure after the pressure holding release operation, a plurality of pressure gradients are acquired at predetermined time intervals, and the average value or median value of the plurality of pressure gradients is set as the second evaluation value R2. By doing so, the influence of spike noise can be reduced.

Note that the second evaluation value R2 may be obtained by a method other than the above, as long as it is a value that indicates a change in pressure after the pressure-holding release operation. For example, the second evaluation value R2 may be obtained by deriving an approximate curve function from pressure time series data made up of a plurality of points and differentiating the function. Moreover, instead of the slope of pressure, the amount of change in pressure may be acquired as the second evaluation value R2. In the example of FIG. 11A, the amount of change in pressure (Pa2-Pa1) during the predetermined time T1 from the pressure release operation (time point X2) may be obtained as the second evaluation value R2.

Further, the training data acquisition unit 31 acquires wear information indicating the wear state of the gate 243b. The wear state of the gate 243b is, for example, the inner diameter of the gate 243b or the amount of wear of the gate 243b. After molding the molded product to be studied, the operator checks the mold part 24 using, for example, a ruler, and measures the inner diameter of the gate 243b. Then, the measured inner diameter of the gate 243b is input to the input section 50. Thereby, the training data acquisition unit 31 acquires the inner diameter of the gate 243b as wear information.

Furthermore, when acquiring the wear amount of the gate 243b as wear information, first, the inner diameter (initial inner diameter) of the gate 243b in the initial state of the mold part 24 (for example, a state where no molding has been performed) is measured. Next, after molding the molded product to be studied, the operator measures the inner diameter of the gate 243b using a ruler, and subtracts the initial inner diameter of the gate 243b from the measured inner diameter. Thereby, the amount of wear from the initial inner diameter of the gate 243b is acquired. Then, the operator inputs the calculated wear amount of the gate 243b to the input section 50. Thereby, the training data acquisition unit 31 acquires the amount of wear on the gate 243b as wear information.

The training data acquisition section 31 stores the initial inner diameter of the gate 243b in a storage section (not shown), and based on the inner diameter of the gate 243b inputted by the operator into the input section 50 and the initial inner diameter, the training data acquisition section 31 may be configured to calculate the wear amount of the gate 243b.

Further, the training data acquisition unit 31 acquires a plurality of third evaluation values R3. The third evaluation value R3 includes, for example, a value related to the temperature detected by the temperature sensor 26, a value related to the environment around and inside the molding device 20 detected by another sensor (for example, a humidity sensor) not shown, and including.

See FIG. 10. The training data acquisition unit 31 acquires molding information based on information input by the operator into the input unit 50 and the molding information storage unit 33. For example, when molding a molded product to be studied, the operator inputs the lot number of the molding material related to the molded product. The training data acquisition unit 31 acquires molding information regarding the physical properties (for example, viscosity) of the molding material corresponding to the lot number from the molding information storage unit 33.

For example, when molding one molded product to be studied, the training data acquisition unit 31 acquires a set of training data (a first evaluation value R1, a second evaluation value R2, acquired during molding of the molded product, A set of third evaluation value R3, molding information, and wear information) is obtained. Similarly, when molding multiple molded products to be studied, multiple sets of training data are acquired.

The learning calculation unit 32 generates a learned model Tm1 that models the correlation between the first evaluation value R1 and the wear information by performing calculations for performing supervised machine learning based on multiple sets of training data. do. In addition, the learning calculation unit 32 performs calculations to perform supervised machine learning based on a plurality of sets of training data, thereby generating a learned model Tm2 that models the correlation between the second evaluation value R2 and the wear information. generate. That is, in this embodiment, two learned models Tm1 and Tm2 are generated.

In this embodiment, a convolutional neural network (CCN) is used as a machine learning model, but other models may also be used. For example, it may be a regression tree model that is a model for grouping data.

Specifically, when generating the learned model Tm1, the first evaluation value R1, third evaluation value R3, and molding information (these three types of information are collectively referred to as "first evaluation information") are explained. By using the wear information as a variable and the wear information as an objective variable, the correlation between the first evaluation information and the wear information is modeled. In addition, when generating the learned model Tm2, the second evaluation value R2, the third evaluation value R3, and the molding information (these three types of information are collectively referred to as "second evaluation information") are used as explanatory variables, By using the wear information as an objective variable, the correlation between the second evaluation information and the wear information is modeled.

The trained models Tm1 and Tm2 generated by the learning calculation unit 32 are stored in the trained model storage unit 34. When new information is input to the learning device 30 and new training data is acquired by the training data acquisition unit 31, the learned models Tm1 and Tm2 stored in the learned model storage unit 34 change the content of the training data. It will be updated accordingly. Further, the learned models Tm1 and Tm2 are transmitted from the learning device 30 to an abnormality prediction device 40, which will be described later, and are also stored in the learned model storage section 45 of the abnormality prediction device 40.

In this embodiment, the first evaluation value R1 regarding the maximum value of pressure and the second evaluation value R2 regarding the slope of pressure are separated to generate learned models Tm1 and Tm2, respectively. However, the first evaluation value R1, the second evaluation value R2, the third evaluation value R3, and the molding information (these four types of information are collectively referred to as "third evaluation information") are used as explanatory variables, and the wear information is used as the objective variable. By doing so, it may be configured to generate a learned model Tm3 that models the correlation between the third evaluation information and the wear information. Further, in this embodiment, two trained models Tm1 and Tm2 are generated, but it may be configured such that only one of the trained models is generated.

<How to generate a trained model>
Next, a method of generating trained models Tm1 and Tm2 by the learning device 30 will be described. The method for generating trained models Tm1 and Tm2 includes a training data acquisition step and a learning calculation step. First, when the training data acquisition process is started, the training data acquisition unit 31 acquires multiple sets of training data. For example, based on the time series data of the pressure detected by the pressure sensor 25, the first evaluation value R1 and the second evaluation value R2 are acquired. In addition, wear information is acquired by a worker actually inspecting the mold section 24 that has molded the molded product for which the first evaluation value R1 and the second evaluation value R2 have been obtained. Next, when the learning calculation process is started, the learning calculation unit 32 generates learned models Tm1 and Tm2 based on the plurality of sets of training data.

<Description of abnormality prediction device>
FIG. 12 is a block diagram showing the functional configuration of the abnormality prediction device 40 according to this embodiment. The abnormality prediction device 40 includes a data acquisition section 41 , an abnormality prediction section 42 , an output section 43 , a forming information storage section 44 , and a learned model storage section 45 . Each of these units is realized by a computer device having a calculation unit such as a CPU and a storage unit such as an HDD. The calculation unit executes a data acquisition process and an abnormality prediction process, which will be described later, based on a program stored in the storage unit.

Similar to the molding information storage section 33, the molding information storage section 44 stores molding information in a table format in which various kinds of first information and second information are associated with each other. The trained model storage unit 45 stores trained models Tm1 and Tm2 generated by the learning device 30.

The shaping information storage unit 44 and the learned model storage unit 45 may be realized by the same storage area as the shaping information storage unit 33 and the learned model storage unit 34 of the learning device 30 in the computer device, or may be realized by separate storage areas. It may also be realized by a region. That is, the learning device 30 and the abnormality prediction device 40 may be configured to share the same molding information storage unit 33 and the learned model storage unit 34, or the learning device 30 and the abnormality prediction device 40 may be configured to share the same molding information storage unit 33 and the learned model storage unit 34, or the learning device 30 and the abnormality prediction device 40 may be configured to It may be configured to include information storage units 33 and 44 and trained model storage units 34 and 45.

The data acquisition unit 41 executes a data acquisition process to acquire information for predicting abnormality from each part of the molding system 10. The information for performing abnormality prediction is, for example, a set of first evaluation value R1, second evaluation value R2, third evaluation value R3, and molding information acquired when molding a molded product to be predicted.

The abnormality prediction unit 42 inputs a set of first evaluation value R1, third evaluation value R3, and molding information acquired by the data acquisition unit 41 to the learned model Tm1. The learned model Tm1 outputs prediction information D1 for predicting the wear state of the gate 243b based on these inputs. Further, the abnormality prediction unit 42 inputs a set of second evaluation value R2, third evaluation value R3, and molding information acquired by the data acquisition unit 41 to the learned model Tm2. Based on these inputs, the learned model Tm2 outputs prediction information D2 for predicting the wear state of the gate 243b.

The prediction information D1, D2 is information corresponding to the objective variables used when generating the trained models Tm1, Tm2. For example, when the target variable is the amount of wear on the gate 243b, the prediction information D1 is information that includes the predicted probability of the amount of wear on the gate 243b. More specifically, the prediction information D1 includes, for example, a first probability that the wear amount of the gate 243b is 1 mm or less, a second probability that it is longer than 1 mm and 3 mm or less, and a third probability that it is larger than 3 mm. As an example, when a set of first evaluation value R1, third evaluation value R3, and shaping information are input to the learned model Tm1, the learned model Tm1 has a first probability of 10%, a second probability of 10%, and a third probability. Prediction information D1 with a probability of 80% is output. The prediction information D2 is also output in the same way.

The output unit 43 determines whether the wear state of the gate 243b is normal or abnormal based on the prediction information D1 and D2 acquired by the abnormality prediction unit 42. The output unit 43 obtains first preliminary determination information PD1 of the gate 243b based on, for example, a predetermined threshold and prediction information D1. The predetermined threshold value is, for example, the maximum allowable wear amount (or the maximum allowable inner diameter). For example, when the maximum allowable wear amount is 3 mm, when the third probability is the highest among the first to third probabilities above, the output unit 43 indicates that the wear state of the gate 243b is abnormal. The first preliminary determination information PD1 shown is acquired (for example, PD1=1). Further, when the first probability or the second probability is the highest among the above-mentioned first to third probabilities, the output unit 43 acquires first preliminary determination information PD1 indicating that the wear state of the gate 243b is normal. (for example, PD1=0).

Similarly, the output unit 43 obtains second preliminary determination information PD2 of the gate 243b based on the prediction information D2. For example, the output unit 43 outputs second preliminary determination information PD2 (PD2=1) indicating that the wear condition of the gate 243b is abnormal, or second preliminary determination information PD2 indicating that the wear condition of the gate 243b is normal. (PD2=0) is obtained.

Then, based on the first and second preliminary determination information PD1 and PD2, it is determined whether the wear state of the gate 243b is normal or abnormal. For example, only when the first and second preliminary determination information PD1 and PD2 indicate that both are abnormal (that is, when PD1×PD2=1), it is determined that the wear state of the gate 243b is abnormal. Then, the output unit 43 acquires "judgment information" that is information for determining whether the wear state of the gate 243b is normal or abnormal.

Here, the first preliminary determination information PD1 is acquired based on the prediction information D1. The prediction information D1 is obtained from the first evaluation value R1 obtained based on the maximum pressure during the pressure holding step ST4 and the learned model Tm1. Moreover, the second preliminary determination information PD2 is acquired based on the prediction information D2. The prediction information D2 is obtained from the second evaluation value R2 obtained based on the slope of the pressure after the pressure holding release operation and the learned model Tm2.

In this way, the first preliminary determination information PD1 and the second preliminary determination information PD2 are two pieces of information ( are obtained based on the first evaluation value R1 and the second evaluation value R2). Therefore, by determining that the wear condition of the gate 243b is abnormal only when both the first preliminary determination information PD1 and the second preliminary determination information PD2 indicate an abnormality, the wear condition of the gate 243b can be determined more reliably. It becomes possible to extract only abnormal cases. Thereby, the frequency of inspection and replacement of the mold part 24 can be reduced.

Note that the output unit 43 determines that the wear state of the gate 243b is abnormal when at least one of the first and second preliminary determination information PD1 and PD2 indicates that it is abnormal (that is, when PD1+PD2≧1). It may be configured to make a determination. With this configuration, it is possible to extract all cases where the wear state of the gate 243b is abnormal. Thereby, it becomes possible to respond to abnormalities in the mold part 24 more quickly, and the yield of molded products can be improved.

The output unit 43 outputs the determination information and prediction information D1 and D2 to the display unit 60 and the control unit 271. The display unit 60 displays determination information and prediction information D1 and D2. In particular, when it is determined that the wear state of the gate 243b is abnormal, the display unit 60 displays the determination information in a highlighted color such as red, and the speaker issues an alert. Good too.

Further, if the wear state of the gate 243b is determined to be abnormal, the molding apparatus 20 including the gate 243b determined to be abnormal is stopped with the mold part 24 open, according to an operation command from the control unit 271. It may be configured as follows. In this case, the operator inspects the mold section 24 based on an alert from the display section 60 and replaces the mold section 24 as necessary.

In addition, in this embodiment, the prediction information D1 and D2 obtained by the abnormality prediction part 42 may be displayed on the display part 60 as it is, without providing the output part 43. In this case, the operator may determine whether the wear state of the gate 243b is normal or abnormal based on the prediction information D1 and D2 displayed on the display unit 60.

<Anomaly prediction method using an abnormality prediction device>
Next, an abnormality prediction method by the abnormality prediction device 40 will be explained. The abnormality prediction method includes a data acquisition step and an abnormality prediction step. When the data acquisition process is started, the data acquisition unit 41 collects a set of first evaluation values R1, second evaluation values R2, and third evaluation values R3 acquired when molding a molded product to be predicted, and Get information. With the above, the data acquisition process is completed.

Next, when the abnormality prediction step is started, first, the abnormality prediction unit 42 inputs a set of the first evaluation value R1, the third evaluation value R3, and the molding information to the learned model Tm1. Get D1. Further, the abnormality prediction unit 42 obtains prediction information D2 by inputting a set of second evaluation value R2, third evaluation value R3, and shaping information to the learned model Tm2.

Next, the output unit 43 obtains first and second preliminary determination information PD1 and PD2 from the prediction information D1 and D2. Then, the output unit 43 acquires determination information based on the first and second preliminary determination information PD1 and PD2. Finally, the output unit 43 outputs the determination information and prediction information D1 and D2 to the display unit 60 and the control unit 271. With the above, the abnormality prediction process is completed.

<Functions and effects of the molding system>
The molding system 10 according to the present embodiment is configured such that the pressure of the molding material L1 in the cavity C1 from the holding pressure operation (time point X1) to the holding pressure release operation (time point X2) reaches the maximum value (first evaluation value R1). Based on this, prediction information for predicting the wear state of the gate 243b is acquired. Furthermore, the molding system 10 according to the present embodiment predicts the wear state of the gate 243b based on the value related to the change in pressure of the molding material L1 in the cavity C1 after the holding pressure release operation (second evaluation value R2). Get forecast information for.

With such a configuration, once the learned models Tm1 and Tm2 are generated, the operator does not need to inspect the mold part 24 one by one, and the various sensors (pressure sensor 25, It becomes possible to predict the wear state of the gate 243b from the information acquired by the pressure sensor 227, etc.).

Furthermore, the molding system 10 according to the present embodiment acquires prediction information D1 by inputting the first evaluation value R1 to a learned model Tm1 that has been subjected to machine learning on the correlation between the first evaluation value R1 and the wear information. do. Here, the explanatory variables of the learned model Tm1 include the first evaluation value R1, and the objective variables of the learned model Tm1 include the inner diameter of the gate 243b or the amount of wear of the gate 243b. With this configuration, the wear state of the gate 243b can be predicted more accurately even if there are variations in the molding conditions.

Furthermore, the molding system 10 according to the present embodiment acquires the prediction information D2 by inputting the second evaluation value R2 to the learned model Tm2, which has undergone machine learning on the correlation between the second evaluation value R2 and the wear information. do. Here, the explanatory variables of the learned model Tm2 include the second evaluation value R2, and the objective variables of the learned model Tm2 include the inner diameter of the gate 243b or the amount of wear of the gate 243b. With this configuration, the wear state of the gate 243b can be predicted more accurately even if there are variations in the molding conditions.

Moreover, the pressure sensor 25 according to this embodiment is provided on the surface of the cavity C1 that is close to the gate 243b. With this configuration, the pressure sensor 25 can more easily detect the pressure change caused by the backflow of the molding material L1 from the cavity C1 to the flow path 243 as shown in FIG. Therefore, a more rapid decrease in pressure after the holding pressure release operation can be detected more accurately. As a result, the second evaluation value R2 can be obtained more accurately.

Moreover, the first evaluation value R1 and the second evaluation value R2 also depend on the temperature of the molding material L1. The higher the temperature of the molding material L1 is, the lower the viscosity of the molding material L1 is, and the easier the molding material L1 is to flow. For this reason, for example, when the gate sealing is not completed, the higher the temperature of the molding material L1 is, the more the molding material L1 will flow back from the cavity C1 to the flow path 243, so the second evaluation value R2 will be negative. tends to increase in the direction of By including the third evaluation value R3 regarding the temperature of the molding material L1 as an explanatory variable, the molding system 10 according to the present embodiment can make the learned models Tm1 and Tm2 into models incorporating the above correlation. Therefore, the wear state of the gate 243b can be predicted more accurately.

<Second embodiment>
The molding system according to the first embodiment has been described above. However, the implementation of the present invention is not limited to this, and various modifications can be made. A molding system 11 according to a second embodiment of the present invention will be described below. In the following description, parts that are unchanged from the first embodiment are designated by the same reference numerals, and the description thereof will be omitted.

FIG. 13 is a block diagram schematically showing a molding system 11 according to the second embodiment. The molding system 11 includes a plurality of molding devices 20, an abnormality prediction device 40a, an input section 50, and a display section 60.

In this embodiment, the abnormality prediction device 40a of the molding system 11 acquires prediction information D3 for predicting the wear state of the gate 243b based on the first evaluation value R1 and the predetermined first reference value Rf1. That is, the molding system 11 is different from the molding system according to the first embodiment in that it predicts the wear state of the gate 243b by comparing the first evaluation value R1 and the first reference value Rf1 without using the learned model Tm1. It is different from 10.

Furthermore, in this embodiment, the abnormality prediction device 40a of the molding system 11 acquires prediction information D4 for predicting the wear state of the gate 243b based on the second evaluation value R2 and the predetermined second reference value Rf2. do. That is, the molding system 11 is different from the molding system according to the first embodiment in that it predicts the wear state of the gate 243b by comparing the second evaluation value R2 and the second reference value Rf2 without using the learned model Tm2. It is different from 10.

FIG. 14 is a block diagram showing the functional configuration of the abnormality prediction device 40a according to this embodiment. The abnormality prediction device 40a includes a data acquisition section 41, an abnormality prediction section 42a, an output section 43a, a molding information storage section 44, and a reference value storage section 46. Each of these units is realized by a computer device having a calculation unit such as a CPU and a storage unit such as an HDD. The data acquisition unit 41 acquires evaluation information similarly to the data acquisition unit 41 according to the first embodiment.

The reference value storage unit 46 stores a first reference value Rf1 and a second reference value Rf2. The first reference value Rf1 is a value generated based on the first evaluation value R1 (maximum value of pressure) obtained when a molded product is molded using the gate 243b whose wear state is normal. The first reference value Rf1 is, for example, an average value or a median value of a plurality of first evaluation values R1 obtained when a plurality of molded products are molded using a gate 243b whose wear state is normal, and a predetermined margin. This value takes into consideration The predetermined margin is determined by the maximum allowable wear amount (or the maximum allowable inner diameter) in the gate 243b.

Further, the second reference value Rf2 is a value generated based on the second evaluation value R2 (a value based on a change in pressure) obtained when a molded product is molded using the gate 243b whose wear state is normal. It is. The second reference value Rf2 is, for example, an average value or a median value of a plurality of second evaluation values R2 obtained when a plurality of molded products are molded using a gate 243b whose wear state is normal, and a predetermined margin. This value takes into consideration The predetermined margin is determined by the maximum allowable wear amount (or the maximum allowable inner diameter) in the gate 243b.

The abnormality prediction unit 42a acquires the first evaluation value R1 acquired by the data acquisition unit 41 and the first reference value Rf1 stored in the reference value storage unit 46. Then, the abnormality prediction unit 42a obtains prediction information D1 by comparing the first evaluation value R1 and the first reference value Rf1. The prediction information D1 is, for example, the difference or ratio between the first evaluation value R1 and the first reference value Rf1.

As mentioned above, when the gate 243b wears out and the flow path resistance of the flow path 243 decreases, the pressurization applied to the molding material from the screw 223 becomes difficult to attenuate. The value R1 tends to be larger than the first reference value Rf1 regarding the maximum pressure when the gate 243b is normal. Therefore, when the first evaluation value R1 becomes larger than the first reference value Rf1, it is predicted that wear has occurred in the gate 243b. Therefore, the prediction information D1 indicates whether or not wear has occurred in the gate 243b.

Further, the abnormality prediction unit 42a acquires the second evaluation value R2 acquired by the data acquisition unit 41 and the second reference value Rf2 stored in the reference value storage unit 46. Then, the abnormality prediction unit 42a obtains prediction information D2 by comparing the second evaluation value R2 and the second reference value Rf2. The prediction information D2 is, for example, the difference or ratio between the second evaluation value R2 and the second reference value Rf2.

As explained in FIGS. 7 to 9, when the gate 243b is worn out and the holding pressure release operation is performed in a state where the gate seal is not completed, the pressure inside the cavity C1 after the holding pressure release operation will drop more rapidly. The second evaluation value R2 regarding the change in pressure after the pressure holding release operation tends to be smaller (that is, larger in the negative direction) than the second reference value Rf2 regarding the change in pressure when the gate 243b is normal. Therefore, when the second evaluation value R2 becomes smaller than the second reference value Rf2, it is predicted that wear has occurred in the gate 243b. Therefore, the prediction information D2 indicates whether or not wear has occurred in the gate 243b.

The output unit 43a determines whether or not the wear state of the gate 243b is abnormal based on the prediction information D1 and D2 acquired by the abnormality prediction unit 42a. For example, the prediction information D1 is the difference (R1-Rf1) between the first evaluation value R1 and the first reference value Rf1, and the prediction information D2 is the difference (R2-Rf2) between the second evaluation value R2 and the second reference value Rf2. ), the output unit 43a determines that the wear state of the gate 243b is abnormal when the prediction information D1 is a positive value and the prediction information D2 is a negative value. Note that the output unit 43a determines that the wear state of the gate 243b is abnormal when either the prediction information D1 is a positive value or the prediction information D2 is a negative value. You may also do so.

The output unit 43a outputs determination information regarding the determination result of the wear state of the gate 243b and prediction information D1 and D2 to the display unit 60 and the control unit 271. The display unit 60 and the control unit 271 perform operations similar to those in the first embodiment based on the determination information and prediction information D1 and D2.

In addition, in this embodiment, the prediction information D1 and D2 obtained by the abnormality prediction part 42a may be displayed on the display part 60 as they are without providing the output part 43a. In this case, the operator may judge the state of the mold section 24 based on the prediction information D1 and D2 displayed on the display section 60.

According to the molding system 11 according to the present embodiment, the wear state of the gate 243b is determined by comparing the first evaluation value R1 and the first reference value Rf1 and comparing the second evaluation value R2 and the second reference value Rf2. can be easily predicted. Note that in this embodiment, the wear state of the gate 243b is predicted using both the first evaluation value R1 and the second evaluation value R2, but the wear state of the gate 243b is predicted using only one of the evaluation values. It's okay.

<Modification 1>
In the second embodiment described above, the first reference value Rf1 is based on the first evaluation value R1 (maximum value of pressure) obtained when a molded product is molded using the gate 243b whose wear state is normal. generated. However, the implementation of the present invention is not limited to this. For example, the first reference value Rf1 may be generated based on the maximum value of the pressure detected by the pressurization sensor 227 during the pressure holding step ST4. Further, the first reference value Rf1 may be generated based on the maximum value of the pressure detected by the pressure sensor 227a during the pressure holding step ST4.

The pressure detected by the pressurization sensor 227 and the pressure sensor 227a is the pressure of the molding material before passing through the flow path 243, and therefore is not affected by flow path resistance due to the flow path 243. When the gate 243b is normal, the flow path 243 has a flow path resistance higher than a predetermined value, so the maximum value of the pressure detected by the pressure sensor 25 during the pressure holding step ST4 is detected by the pressurization sensor 227 and the pressure sensor 227a, respectively. is lower than the maximum value of the pressure by a predetermined value Th1 or more. Furthermore, when the gate 243b is worn out, the flow path resistance of the flow path 243 becomes less than a predetermined value, and the maximum value of the pressure detected by the pressure sensor 25 during the pressure holding step ST4, and the pressure detected by the pressurization sensor 227 and the pressure sensor 227a, respectively. The difference from the maximum value of the pressure applied is less than the predetermined value Th1.

Therefore, the data acquisition unit 41 acquires the maximum value of the pressure detected by the pressurization sensor 227 and the pressure sensor 227a as the first reference value Rf1, and the abnormality prediction unit 42a acquires the first reference value Rf1 and the first evaluation value. The difference (Rf1-R1) from R1 is acquired as prediction information, and the output unit 43a determines that there is an abnormality in the wear state of the gate 243b when the difference (Rf1-R1) is smaller than a predetermined value Th1. may be configured. Note that the abnormality prediction unit 42a acquires the ratio between the first reference value Rf1 and the first evaluation value R1 as prediction information, and the output unit 43a determines whether or not the wear state of the gate 243b is abnormal based on the ratio. It may be configured to make a determination.

With this configuration, regardless of whether the wear state of the gate 243b is normal or not, both the first reference value Rf1 and the first evaluation value R1 are set when the molded product to be predicted is molded once. can be obtained. Therefore, for example, even if the molding conditions (pressurization, mold clamping force, type of molding material, etc.) are changed while using the same mold part 24, it is easy to change the molding conditions after the change. It becomes possible to obtain the reference value Rf1.

In addition, since the first reference value Rf1 and the first evaluation value R1 are values obtained by different pressure sensors, at least one of the first reference value Rf1 and the first evaluation value R1 includes the pressure sensors 227, 227a, A correction value for calibration between 25 and 25 may be added.

<Modification 2>
In the second embodiment described above, the output unit 43a determines the wear state of the gate 243b based on the prediction information D1 and D2 acquired in one molding. However, the output unit 43a may determine the wear state of the gate 243b based on information acquired through multiple molding operations.

FIG. 15 is a graph schematically explaining the abnormality determination method according to this modification. FIG. 15(a) is a graph showing a plurality of first evaluation values R1 for each number of times of molding. The vertical axis of the graph is pressure, and the horizontal axis is the number of moldings. For example, a plurality of molded products to be predicted are formed by the molding device 20, and a plurality of first evaluation values R1 are obtained. In the example of FIG. 15(a), the first evaluation value R1 of the first molding is "Pta". Further, the first reference value Rf1 is a value Pth obtained by adding a margin Th2 to the first evaluation value R1 of the first molding (that is, when there is no wear on the gate 243b).

Molding is repeatedly performed in the molding device 20, and the data acquisition unit 41 acquires the first evaluation value R1 each time. Then, when a plurality of (for example, three) first evaluation values R1 that are consecutively acquired over a predetermined plurality of times (for example, three times) are all larger than the first reference value Rf1, the gate Prediction information D1 indicating that wear has occurred in 243b is acquired.

For example, since the first evaluation value R1 of the N1th time is less than the first reference value Rf1 (=Pth), it is not predicted that the gate 243b is worn out, and the output part 43a indicates that the gate 243b is normal. Prediction information D1 (for example, D1=0) indicating that there is something is acquired. Furthermore, although the first evaluation value R1 of the N2th time is greater than or equal to the first reference value Rf1, since the first evaluation value R1 of the N1th time is less than the first reference value Rf1 (=Pth), the gate 243b is worn out. It is not predicted that this has occurred, and the output unit 43a obtains prediction information D1 indicating that the gate 243b is normal. Similarly, during the N3-th molding, the output unit 43a acquires prediction information D1 indicating that the gate 243b is normal.

The first evaluation value R1 of the N4th time is greater than or equal to the first reference value Rf1, and the first evaluation value R1 of the N2 and N3rd times is also greater than or equal to the first reference value Rf1. The value R1 becomes larger than the first reference value Rf1. In this case, the output unit 43a acquires prediction information D1 (for example, D1=1) indicating that wear has occurred in the gate 243b.

FIG. 15(b) is a graph showing a plurality of second evaluation values R2 for each number of times of molding. The vertical axis of the graph is the slope of pressure, and the horizontal axis is the number of moldings. For example, a plurality of molded products to be predicted are formed by the molding device 20, and a plurality of second evaluation values R2 are obtained. In the example of FIG. 15(b), the second evaluation value R2 of the first molding is "dPva". Further, the second reference value Rf2 is a value dPth obtained by subtracting the margin Th3 from the second evaluation value R2 of the first molding (that is, when there is no wear on the gate 243b).

Molding is repeatedly performed in the molding device 20, and the data acquisition unit 41 acquires the second evaluation value R2 each time. For example, if the second evaluation value R2 of the N1th time is greater than or equal to the second reference value Rf2 (=dPth), and the second evaluation value R2 of the N2 to N4th times is less than the second reference value Rf2, the output unit 43a Until then, prediction information D2 (for example, D2=0) indicating that the gate 243b is normal is obtained, and at the N4th time, prediction information D2 (for example, D2=1) indicating that wear has occurred in the gate 243b is obtained. get.

That is, when a plurality of (for example, three) second evaluation values R2 that are consecutively acquired over a predetermined plurality of times (for example, three times) are all smaller than the second reference value Rf2, the gate 243b is configured to acquire prediction information D2 indicating that wear has occurred.

When a disturbance (for example, vibration) is suddenly applied to the molding device 20, the first evaluation value R1 and the second evaluation value R2 are obtained which are abnormal values only once, although there is no abnormality in the gate 243b. There is. As described above, by monitoring the first evaluation value R1 and the second evaluation value R2 multiple times, it is possible to exclude the influence of such disturbances and predict the wear state of the gate 243b more accurately. Become.

The embodiments disclosed above are illustrative in all respects and are not restrictive. That is, the molding system of the present invention is not limited to the illustrated form, but may take other forms within the scope of the present invention.

10, 11 Molding system 20 Molding device 21 Bed 22 Injection part 221 Hopper 222 Cylinder 223 Screw 225 Ball screw 226 Motor 227 Pressure sensor 227a Pressure sensor 228 Travel sensor 229 Heater 224 Nozzle 23 Mold clamping part 231 Fixed plate 231a Through hole 23 2 Movable platen 232a Through hole 233 Tie bar 234 Ball screw 235 Support plate 236 Force sensor 237 Motor 24 Mold section 241 Mold 242 Mold 243 Channel 243a First opening 243b Second opening (gate) 25 Pressure sensor 26 Temperature sensor 27 control panel 271 control unit 272 communication unit 30 learning device 31 training data acquisition unit 32 learning calculation unit 33 molding information storage unit 34 learned model storage unit 40, 40a abnormality prediction device 41 data acquisition unit 42, 42a abnormality prediction unit 43, 43a Output section 44 Molding information storage section 45 Learned model storage section 46 Reference value storage section 50 Input section 60 Display section C1 Cavity L1 Molding material F0, F1, F2 Graph line F Ps1 Pressure (set pressure)
X1 Time point (of holding pressure operation) X2 Time point (after holding pressure release operation) Pt1, Pt2 Maximum pressure d1 Inner diameter (when gate is normal) d2 Inner diameter (when gate is abnormal) S1 Solidified body G1 Gap R1 First evaluation value R2 Second evaluation value R3 Third evaluation value T1 Predetermined time Tm1, Tm2, Tm3 Learned model D1, D2, D3, D4 Prediction information PD1 First preliminary determination information PD2 Second preliminary determination information Rf1 First reference value Rf2 Second standard Value Th1 Predetermined value Th2, Th3 Margin

Claims (13)

A molding system comprising a molding device that molds a molded product and an abnormality prediction device that predicts abnormalities in the molding device,
The molding device includes:
a mold portion having a cavity formed therein and a gate opening toward the cavity;
A filling operation for filling the molding material in a molten state into the cavity via the gate, a holding operation for maintaining the pressure of the molding material filled in the cavity, and a release of the holding of the pressure of the molding material. an injection unit that performs a holding pressure release operation;
a pressure sensor that detects the pressure of the molding material in the cavity;
has
The abnormality prediction device includes:
a data acquisition unit that acquires a maximum value of the pressure detected by the pressure sensor between the pressure holding operation and the pressure holding release operation as a first evaluation value;
an abnormality prediction unit that acquires prediction information for predicting a wear state of the gate based on the first evaluation value;
A molding system with The abnormality prediction unit is configured to calculate the abnormality prediction value obtained during molding of the molded product to be predicted, rather than the first reference value generated based on the first evaluation value obtained when the wear state of the gate is normal. acquiring the prediction information indicating that wear has occurred in the gate when the first evaluation value becomes large;
The molding system according to claim 1. The abnormality prediction unit molds a plurality of the molded products to be predicted by the molding device, and the first evaluation values that are consecutively acquired over a predetermined plurality of times are all larger than the first reference value. obtaining the prediction information indicating that wear has occurred in the gate if
The molding system according to claim 2. The injection part is
a cylinder that stores the molding material in a molten state and connects to the mold section;
a screw inserted into the cylinder and pushing the molten molding material toward the mold section;
The molding device further includes a second pressure sensor that detects the pressure of the molding material in the cylinder or the pressure that the screw receives from the molding material,
The data acquisition unit acquires a first reference value based on the maximum value of the pressure detected by the second pressure sensor between the pressure holding operation and the pressure releasing operation,
The molding system according to claim 1, wherein the abnormality prediction unit acquires the prediction information based on a difference or ratio between the first evaluation value and the first reference value. The abnormality prediction unit acquires the prediction information by inputting the first evaluation value into a learned model that has been subjected to machine learning of the correlation between the first evaluation value and the wear state of the gate,
The explanatory variables of the trained model include the first evaluation value,
The objective variable of the learned model includes the inner diameter of the gate or the amount of wear of the gate,
The molding system according to claim 1. The data acquisition unit acquires a second evaluation value regarding a change in pressure of the molding material after the pressure release operation based on time series data of the pressure detected by the pressure sensor,
The abnormality prediction unit acquires one or more of the prediction information based on the first evaluation value and the second evaluation value.
The molding system according to any one of claims 1 to 5. a mold portion having a cavity formed therein and a gate opening toward the cavity;
A filling operation of filling the molding material in a molten state into the cavity via the gate, a holding operation of maintaining the pressure of the molding material filled in the cavity, and a release of the holding of the pressure of the molding material. an injection unit that performs a holding pressure release operation;
a pressure sensor that detects the pressure of the molding material in the cavity;
A molding device that molds a molded product;
an abnormality prediction device for predicting abnormalities in the molding device;
A molding system comprising a learning device,
The learning device generates a trained model,
The explanatory variables of the learned model include a first evaluation value obtained based on the maximum value of the pressure detected by the pressure sensor during the period from the pressure holding operation to the pressure releasing operation,
The objective variable of the learned model includes the inner diameter of the gate or the amount of wear of the gate.
molding system. An abnormality prediction device that predicts abnormalities in a molding device that molds a molded product,
The molding device includes:
a mold portion having a cavity formed therein and a gate opening toward the cavity;
A filling operation of filling the molding material in a molten state into the cavity via the gate, a holding operation of maintaining the pressure of the molding material filled in the cavity, and a release of the holding of the pressure of the molding material. an injection unit that performs a holding pressure release operation;
a pressure sensor that detects the pressure of the molding material in the cavity;
has
The abnormality prediction device includes:
a data acquisition unit that acquires a maximum value of the pressure detected by the pressure sensor between the pressure holding operation and the pressure releasing operation as a first evaluation value;
an abnormality prediction unit that acquires prediction information for predicting a wear state of the gate based on the first evaluation value;
An anomaly prediction device comprising: a mold portion having a cavity formed therein and a gate opening toward the cavity; a filling operation for filling the molding material in a molten state into the cavity via the gate; Abnormality prediction for predicting an abnormality in a molding device that molds a molded product, which includes an injection unit that performs a pressure holding operation that holds the pressure of the molding material and a holding pressure release operation that releases the holding of the pressure of the molding material. A method,
a data acquisition step of acquiring, as a first evaluation value, a maximum value of the pressure detected by a pressure sensor that detects the pressure of the molding material in the cavity between the pressure holding operation and the pressure releasing operation;
an abnormality prediction step of acquiring prediction information for predicting the wear state of the gate based on the first evaluation value;
An anomaly prediction method comprising: a mold portion having a cavity formed therein and a gate opening toward the cavity; a filling operation for filling the molding material in a molten state into the cavity via the gate; For predicting an abnormality in a molding device that molds a molded product, the injection unit performs a pressure holding operation that holds the pressure of the molding material and a holding pressure release operation that releases the holding of the pressure of the molding material. A program,
a data acquisition step of acquiring, as a first evaluation value, a maximum value of the pressure detected by a pressure sensor that detects the pressure of the molding material in the cavity between the pressure holding operation and the pressure releasing operation;
an abnormality prediction step of acquiring prediction information for predicting the wear state of the gate based on the first evaluation value;
A program that causes a computer device to execute. a mold portion having a cavity formed therein and a gate opening toward the cavity;
A filling operation for filling the molding material in a molten state into the cavity via the gate, a holding operation for maintaining the pressure of the molding material filled in the cavity, and a release of the holding of the pressure of the molding material. and an injection section that performs the holding pressure release operation .
A trained model for predicting abnormalities in a molding device that molds a molded product,
The trained model is
A description including a first evaluation value obtained based on a maximum value of pressure detected by a pressure sensor that detects the pressure of the molding material in the cavity between the pressure holding operation and the pressure releasing operation. variables are entered,
outputting a target variable including an inner diameter of the gate or an amount of wear of the gate based on the explanatory variable;
A trained model that makes computers work . The trained model is generated by a learning device,
The learning device is constituted by a computer device having a calculation section and a storage section,
The learning device includes a training data acquisition unit and a learning calculation unit,
The training data acquisition unit acquires the first evaluation value and the inner diameter of the gate or the amount of wear of the gate as training data,
The trained model is calculated by performing supervised machine learning in the learning calculation unit based on the plurality of sets of training data, so that the first evaluation value and the wear amount of the inner diameter of the gate or the wear amount of all gates are calculated. It is generated by modeling the correlation between
The trained model according to claim 11. The learned model according to claim 12, wherein the calculation unit is a CPU or a GPU, and the storage unit is an HDD or an SSD.

Images