JP7530687B2 - Information processing device, information processing system, and program - Google Patents

Description

The present invention relates to an information processing device, an information processing system and a program, and in particular to setting molding conditions for a molding machine.

Generally, in a method of molding a product by supplying molten resin (plastic) to a mold in a molding machine, the multiple molding conditions are adjusted little by little to find the conditions that will produce a good product.
This condition adjustment occurs when starting up production first thing in the morning (starting with looser conditions at first) or when there are changes in the surrounding environment.
For example, in injection molding, the state of the molten resin is very important, and since the molten resin is affected by the surrounding environment (such as temperature and humidity), even if it is molded under the same molding conditions, it may be a good product or a defective product.In addition, if there is variation in the physical properties of the resin depending on the resin lot, or if a mold used in a domestic factory is used in an overseas factory, it is necessary to adjust the molding conditions depending on the environment and the molding machine.

JP 2017-030152 A JP 2020-049929 A

However, in many cases, an experienced person must make fine adjustments to the molding conditions while observing the condition of the molded product each time, which can be difficult when an experienced person is not available, and it takes time to master the skilled techniques.

As a result, setting molding conditions largely depends on intuition and experience, and unless the worker is an experienced worker, it takes a lot of time to set them up. There were also problems with all the molding done to set the conditions being wasted, and variations in the conditions set by different workers also led to variations in the quality of the molded products.

Here, a technology has been disclosed in which the molding conditions to be adjusted and the amount of adjustment are calculated and output based on physical quantity data during molding using machine learning with AI (artificial intelligence) (see Patent Document 1).
In addition, the technology disclosed in the above-mentioned Patent Document 1 adjusts the conditions taking into account the quality of the molded product, but since it is necessary to proceed to the inspection process in order to provide feedback on the quality, a technology is disclosed in which the conditions are adjusted (specifically, the internal mold pressure is used) using a first model and a second model generated by machine learning using only the information up to the inspection process (see Patent Document 2).

However, even with the technology disclosed in Patent Document 2 to overcome the problems of Patent Document 1, if the first model is based on machine learning in addition to the second model, the processing becomes complicated. Meanwhile, as stated in paragraph 0029 of the specification of Patent Document 1, "machine learning is generally classified into various algorithms depending on the purpose and conditions, such as supervised learning and unsupervised learning. The purpose of this invention is to learn the operation condition setting work for a mold, and considering that in an injection molding system, there are parameters that cannot be directly measured in the injection environment, and it is difficult to explicitly indicate what action (adjustment of the operation conditions) is correct for the state of the molded product that is the result of injection, a reinforcement learning algorithm is adopted in which the machine learning device automatically learns the action to achieve the goal simply by giving a reward." It is presumed that the parameters during molding are complex and some are unknown, so it can be said that the first model is not necessarily perfect just because it is machine learning.

The objective of one aspect of the present invention is to provide technology that simplifies the AI processing procedures while assisting workers in setting molding conditions in a way that is easy for them to understand.

A first molding condition determination support system according to one aspect of the present invention comprises:
A molding condition determination support system including a molding machine that molds a molded product by supplying a molten material to a cavity of a mold, and a sensor and a display device connected thereto,
The sensor acquires data on the mold pressure waveform that serves as the reference for molding a specific molded product.
The information processing device generates a learning model by machine learning based on learning data as to, when there is a difference between a reference mold internal pressure waveform and a waveform for each shot, what kind of reference waveform is required to change the values of setting conditions related to a plurality of cut-out sections set for the mold internal pressure waveform in order to make the waveform for each shot approach the reference waveform,
The information processing device calculates optimal setting conditions and their values to be adjusted using the learning model so as to reduce the absolute value of the difference between a reference mold internal pressure waveform and the mold internal pressure waveforms obtained by the sensor for each of the multiple shots in the manufacturing stage, and at that time, determines the setting conditions for any of the sections one by one in a predetermined order,
The display device displays the setting conditions to be reset and their values on the screen in order according to the time series of molding of the cut-out section.

A second molding condition determination support system according to an aspect of the present invention comprises:
A molding condition determination support system including a molding machine that molds a molded product by supplying a molten material to a cavity of a mold, and a sensor and a display device connected thereto,
The sensor acquires data on a mold internal pressure waveform that serves as a reference during molding of the predetermined molded product,
the information processing device generates a learning model by machine learning based on learning data indicating, when there is a difference between the reference mold internal pressure waveform and the waveform for each shot acquired by the sensor, what kind of reference waveform requires how much the values of at least (1) injection speed, (2) VP switching position, (3) dwell pressure, and (4) dwell time setting conditions need to be changed in order for the waveform for each shot to approach the reference waveform;
The information processing device calculates the optimal setting conditions and their values to be adjusted using the learning model so as to reduce the absolute value of the difference between the reference mold internal pressure waveform and the mold internal pressure waveforms obtained by the sensor for each of the multiple shots in the manufacturing stage, and at that time, determines the (1) injection speed, the (2) VP switching position, the (3) dwell pressure, and the (4) dwell pressure time in this order;
The display device displays on the screen the setting conditions to be reset and their values in order of the molding time series of (1) injection speed, (2) VP switching position, (3) holding pressure, and (4) holding pressure time.

An information processing device according to a first aspect of the present invention is an information processing device that outputs information regarding setting conditions of a molding machine that molds a molded product by supplying a molten material to a cavity of a mold,
At least one storage device in which at least one reference waveform data serving as a reference is stored, the reference waveform data being waveform data detected by a sensor provided in the molding machine or the mold when a predetermined molded product is molded as a good product;
a molding condition providing unit that acquires reference waveform data by referring to the storage device, acquires at least one piece of waveform data for one shot during molding of the molded product from the sensor, and processes at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to input at least one value of the setting condition and the waveform data during molding at the value of the setting condition so as to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of the setting condition detected by the sensor as waveform data that is the same as or similar to the reference waveform data;
Equipped with.

An information processing device according to a second aspect of the present invention is an information processing device according to the first aspect, in which the molding condition providing unit outputs values of the setting conditions for a plurality of processes or sections by determining the setting conditions for each process or section in a predetermined order among the plurality of processes or sections included in one shot during molding of the molded product.

An information processing device according to a third aspect of the present invention is an information processing device according to the second aspect, wherein the molding condition providing unit outputs information for displaying the values of the output setting conditions for each process or section in order following the chronological order of the molding processes or sections.

An information processing device according to a fourth aspect of the present invention is an information processing device according to any one of the first to third aspects, wherein the sensor is a pressure sensor and/or a temperature sensor.

An information processing device according to a fifth aspect of the present invention is the information processing device according to any one of the first to fourth aspects, wherein the storage device stores a plurality of the reference waveform data;
The molding condition providing unit identifies multiple setting condition ranges that will produce good products from multiple setting conditions corresponding to the multiple reference waveform data, selects one setting condition range from the identified multiple setting condition ranges, and outputs the value of the setting condition based on the selected setting condition range.

An information processing device according to a sixth aspect of the present invention is an information processing device according to the fifth aspect, wherein the molding condition providing unit, when selecting one setting condition range from a plurality of identified setting condition ranges, selects the setting condition range having the widest setting condition range.

An information processing device according to a seventh aspect of the present invention is an information processing device according to any one of the first to sixth aspects, wherein the storage device stores a machine learning model for each type of resin or each property of the resin, and the molding condition providing unit selects a machine learning model corresponding to the type of resin or each physical property of the resin input by a user by referring to the storage device, and outputs a value of the at least one setting condition using the selected machine learning model.

An information processing device according to an eighth aspect of the present invention is an information processing device according to any one of the first to seventh aspects, wherein the storage device stores a machine learning model in which the type of resin or the physical properties of the resin are included as input parameters, and the molding condition providing unit outputs a value of the at least one setting condition by inputting the type of resin or the characteristics of the resin into the machine learning model based on the type of resin or the physical properties of the resin input by a user.

An information processing device according to a ninth aspect of the present invention is an information processing system that outputs information regarding setting conditions of a molding machine that molds a molded product by supplying a molten material to a cavity of a mold,
At least one storage device in which at least one reference waveform data serving as a reference, which is waveform data detected by a sensor provided on the molding machine or the mold when a specified molded product is molded as a good product, and at least one reference operation data serving as a reference, which is operation data of a controller that controls the molding machine when a good product is molded, are stored;
a molding condition providing unit that uses as input at least one value of a setting condition and a first difference and a second difference when the value of the setting condition is used, and processes the value of at least one setting condition and the first difference and the second difference when the value of the setting condition is used using a machine learning model that learns to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of a setting condition for which the sensor detects waveform data that is the same as or similar to the reference waveform data;
Equipped with.

An information processing device according to a tenth aspect of the present invention is the information processing device according to the ninth aspect, wherein the storage device stores a machine learning model for each physical property of a resin;
The molding condition providing unit estimates physical properties of the resin of the molding machine using characteristic values of a motor provided in the molding machine and/or characteristic values of the motor stored in the controller, and outputs a value of the at least one setting condition using a machine learning model corresponding to the estimated physical properties of the resin.

An information processing device according to an eleventh aspect of the present invention is the information processing device according to the ninth aspect, wherein the storage device stores a machine learning model including a physical property of a resin as an input parameter;
The molding condition providing unit estimates physical properties of the resin of the molding machine using characteristic values of a motor provided in the molding machine and/or characteristic values of the motor stored in the controller, inputs the estimated physical properties of the resin into a machine learning model, and outputs a value of the at least one setting condition.

An information processing device according to a twelfth aspect of the present invention is an information processing device according to any one of the ninth to eleventh aspects, wherein the molding condition providing unit outputs a value of at least one set condition to the molding machine so as to feed it back to the molding machine.

An information processing device according to a thirteenth aspect of the present invention is an information processing device according to any one of the first to twelfth aspects, wherein the machine learning model is Bayesian optimization.

An information processing device according to a 14th aspect of the present invention is an information processing device according to any one of the first to 13th aspects, wherein the at least one setting condition is at least one of injection speed, VP switching position, holding pressure, and holding pressure time.

An information processing system according to a fifteenth aspect of the present invention is an information processing system that outputs information regarding setting conditions of a molding machine that molds a molded product by supplying a molten material to a cavity of a mold,
At least one storage device in which at least one reference waveform data serving as a reference is stored, the reference waveform data being waveform data detected by a sensor provided in the molding machine or the mold when a predetermined molded product is molded as a good product;
a molding condition providing unit that acquires reference waveform data by referring to the storage device, acquires at least one piece of waveform data for one shot during molding of the molded product from the sensor, and processes at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to input at least one value of the setting condition and the waveform data during molding at the value of the setting condition so as to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of the setting condition detected by the sensor as waveform data that is the same as or similar to the reference waveform data;
Equipped with.

An information processing system according to a sixteenth aspect of the present invention is an information processing system that outputs information regarding setting conditions of a molding machine that molds a molded product by supplying a molten material to a cavity of a mold,
At least one storage device in which at least one reference waveform data serving as a reference, which is waveform data detected by a sensor provided on the molding machine or the mold when a specified molded product is molded as a good product, and at least one reference operation data serving as a reference, which is operation data of a controller that controls the molding machine when a good product is molded, are stored;
a molding condition providing unit that uses as input at least one value of a setting condition and a first difference and a second difference when the value of the setting condition is used, and processes the value of at least one setting condition and the first difference and the second difference when the value of the setting condition is used using a machine learning model that learns to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of a setting condition for which the sensor detects waveform data that is the same as or similar to the reference waveform data;
Equipped with.

A program according to a seventeenth aspect of the present invention is a program for causing a computer to refer to at least one storage device in which at least one piece of reference waveform data, which is waveform data detected by a sensor provided in a molding machine or a mold when a predetermined molded product is molded as a good product, is stored, the program comprising:
a step of acquiring reference waveform data by referring to the storage device, acquiring at least one piece of waveform data for one shot during molding of the molded product from the sensor, and processing at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to input at least one value of the setting condition and the waveform data during molding at the value of the setting condition so as to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of the setting condition detected by the sensor as waveform data that is identical or similar to the reference waveform data;
This is a program for executing the above.

A program according to an eighteenth aspect of the present invention is a program for causing a computer to refer to at least one storage device in which at least one reference waveform data serving as a reference, which is waveform data detected by a sensor provided on a molding machine or a mold when a specified molded product is molded as a good product, and at least one reference operation data serving as a reference, which is operation data of a controller that controls the molding machine when a good product is molded,
a step of using a machine learning model that learns to reduce a difference between waveform data during molding and the reference waveform data and processes the value of at least one setting condition and the first difference and the second difference when the value of the setting condition is used as input, thereby outputting at least one value of the setting condition for which the sensor detects waveform data identical or similar to the reference waveform data;
This is a program for executing the above.

According to one aspect of the present invention, a technology can be provided that simplifies the AI processing procedures while assisting workers in setting molding conditions in a manner that is easy for them to understand.

1 is an external view of a molding machine of a molding condition determination support system according to an embodiment of the present invention. 1 is a schematic configuration diagram of a part of a molding machine of a molding condition determination support system according to an embodiment of the present invention. 1 is a diagram showing an example of a device configuration of a molding condition determination support system according to an embodiment of the present invention; FIG. 2 is a diagram showing an internal mold pressure waveform over time during one shot for molding one molded product in an embodiment of the present invention. 3 is a diagram illustrating a process flow of the molding condition determination support system according to the embodiment of the present invention. FIG. FIG. 2 is a diagram showing an example of a display screen of a display device of the molding condition determination support system according to the embodiment of the present invention. 1 is a functional block diagram of an information processing device of a molding condition determination support system according to an embodiment of the present invention. 4 is a flowchart showing a process flow of an information processing device of the molding condition determination support system according to the embodiment of the present invention. FIG. 13 is a diagram showing an example of a device configuration of a molding condition determination support system according to a second embodiment. FIG. 11 is a functional block diagram of an information processing device of a molding condition determination support system according to a second embodiment. FIG. 11 is a schematic diagram showing the positions of sensors in a modified example according to the second embodiment. FIG. 13 is a diagram showing an example of a device configuration of a molding condition determination support system according to a third embodiment. FIG. 13 is a functional block diagram of an information processing device of the molding condition determination support system according to the third embodiment. FIG. 2 is a diagram showing multiple examples of mold internal pressure waveforms over time during one shot for molding one molded product. FIG. 13 is a diagram showing an example of a device configuration of a molding condition determination support system according to a fourth embodiment. FIG. 13 is a functional block diagram of an information processing device of a molding condition determination support system according to a fourth embodiment. FIG. 2 is a schematic diagram showing an example of a machine learning model stored in a storage unit. FIG. 13 is an external view of a molding machine of a molding condition determination support system according to a fifth embodiment. FIG. 13 is a diagram showing an example of a device configuration of a molding condition determination support system according to a fifth embodiment. FIG. 13 is a functional block diagram of an information processing device of a molding condition determination support system according to a fifth embodiment. FIG. 13 is a schematic configuration diagram of a molding condition determination support system according to a sixth embodiment.

Each embodiment will be described below with reference to the drawings. However, more detailed explanations than necessary may be omitted. For example, detailed explanations of matters that are already well known or duplicate explanations of substantially identical configurations may be omitted. This is to avoid making the following explanation unnecessarily redundant and to make it easier for those skilled in the art to understand.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a molding condition determination support system according to an embodiment of the present invention will be described in detail with reference to the drawings.
The molding condition determination support system is applied to a method of molding a molded product by supplying molten material to a mold of a molding machine. Molding methods to which the system can be applied include, for example, injection molding of resin or rubber, blow molding, extrusion molding, metal casting such as die casting, etc. In the following, the system will be mainly described using injection molding as an example of the application.

First Embodiment
FIG. 1 is an external view of a molding machine 1 of a molding condition determination support system 100 according to an embodiment of the present invention.
The molding machine 1 is an injection molding machine that produces various plastic products by pouring molten resin (plastic) into a mold, cooling and solidifying it, and removing it. It is roughly divided into an injection unit 10 that uses heat to melt the resin material and inject it into the mold, a mold clamping unit 20 that opens and closes the mold, and a controller 30 that controls the molding conditions (e.g., the speed, pressure, temperature, etc. when injecting the resin into the mold).

FIG. 2 is a schematic diagram of a part of a known general molding machine 1 such as that disclosed in the above-mentioned Patent Documents 1 and 2.
The injection unit 10 includes a hopper, a heating cylinder, a screw, a nozzle, a heater, a drive unit, etc. The hopper is an inlet for pellets (granular molding material). The heating cylinder is provided so as to be movable in the axial direction relative to the bed. The screw is disposed inside the heating cylinder and is provided so as to be rotatable and movable in the axial direction. The nozzle is an injection port provided at the tip of the heating cylinder, and supplies the molten resin inside the heating cylinder to the cavity of the mold by the axial movement of the screw.

The clamping unit 20 performs opening and closing operations of the attached mold, and prevents the mold from opening due to the pressure of the molten resin injected into the mold cavity (molded product portion) when the mold is clamped. The clamping unit 20 includes a fixed platen, a movable platen, tie bars, a drive device, and the like. A fixed first mold is fixed to the fixed platen. The fixed platen can abut against the nozzle and guides the resin injected from the nozzle to the mold cavity. The cavity is an area formed between the first mold and the second mold and corresponds to the product shape. A movable second mold is fixed to the movable platen and can approach and move away from the fixed platen. The tie bars support the movement of the movable platen. The drive device is, for example, a cylinder device, and moves the movable platen. The first mold has a supply path between the nozzle and the cavity. A three-plate mold or the like can be similarly applied.
Ejector pins (extrusion pins) 22 fixed to an ejector plate 21 are pushed out when the mold is opened, so that the molded product (workpiece) can be removed from the mold after molding.
A plurality of sensors 23 for acquiring data inside the mold are provided at positions corresponding (opposite) to the positions of the ejector pins 22 in such a manner as to be embedded in the ejector plate 21 (see also FIG. 5). The sensors 23 are pressure sensors for detecting the pressure inside the mold received from the molten resin, and are, for example, load cells. Note that the measurement mechanism via the ejector plate 21 and the ejector pins 22 is not essential, and the mounting position of the sensors 23 is not particularly limited as long as it is possible to acquire data inside the mold, and may be any position, whether inside or outside the mold.

The controller 30 controls the drive device of the injection unit 10 and the drive device of the mold clamping unit 20 based on command values (parameters) related to molding conditions.
The molding machine 1 may be composed of a plurality of devices equipped with a control device, a sensor, etc. Examples of such devices include a mold clamping force control device, a control device for driving the mold opening and closing, screw rotation and forward and backward movement, injection unit forward and backward movement, etc., a nozzle cylinder, a hopper lower temperature control device, a pressure control device, etc. In relation to the mold, for example, there are a mold temperature control device, a hot runner temperature and nozzle opening and closing control device, a mold vibration device, etc. Examples of peripheral devices include a molded product removal device, an insert product insertion device, a nesting insertion device, a molded product burr removal device, a runner cutting device, a molded product weight meter, a molded product strength tester, a molded product optical inspection device, a molded product photography device and image processing device, a molded product transport robot, etc., all of which are controlled by the control device. Among these devices, there may be a control device equipped with a sensor and performing feedback control or feedforward control by a closed loop, and there may also be a device having a function of only outputting data.

As disclosed in the above-mentioned Patent Documents 1 and 2, the following general injection molding method is known.
In the metering process, the resin is melted by the shear friction heat generated by the heater and the rotation of the screw, and is stored between the tip of the heating cylinder and the nozzle. As the amount of stored resin increases, the screw retreats, and the amount of stored resin is measured from the retreated position.
In the mold clamping step, the movable platen is moved to align the first mold with the second mold, thereby performing mold clamping. Furthermore, the nozzle is connected to the fixed platen.
In the injection filling step, with the rotation of the screw stopped, the screw is moved toward the nozzle at a predetermined injection speed and pushing force, thereby injecting and filling the resin into the cavity of the mold at high pressure.
In the dwelling process, the filled resin is further pushed into the cavity, and a dwelling process is performed in which a predetermined dwelling pressure is applied to the resin in the cavity for a predetermined time. Specifically, a constant pushing force (dwelling pressure) is applied to the screw to apply a predetermined dwelling pressure to the resin. The pressure generated in the resin in the cavity varies depending on the position of the cavity.
In the cooling process, the pushing of the resin is stopped, the holding pressure is reduced, and the mold is further cooled to solidify the resin in the cavity.
In the demolding and removal process, the first and second molds are separated from each other, and the ejector pins (extrusion pins) 22 protrude when the molds are opened, allowing the molded product to be removed (removed) from the molds after molding.

Figure 3 is a diagram showing an example of the equipment configuration of the molding condition determination support system 100 in this embodiment.

A mold equipped with a sensor (load cell) 23 for acquiring data inside the mold such as mold pressure is installed in the mold clamping unit 20 of the molding machine 1. The sensor 23 transmits sensing data to an information processing device 4 via an instrumentation amplifier 2, an A/D converter 3, etc., which serve as a data acquisition unit.
The sensor 23 and the information processing device 4 may be connected wirelessly or the like. Here, the sensor (load cell) 23 is used, but any type or data communication method may be used as long as the sensor can acquire data inside the cavity such as mold pressure, and the installation position may be arbitrary. For example, when the mold is manufactured, if a sensor is embedded in the mold itself or the sensor is provided in such a manner that it touches the cavity, data inside the cavity such as mold pressure can be acquired.
The information processing device 4 is, for example, an industrial PC (personal computer) and is equipped with a CPU (Central Processing Unit) which is an arithmetic device serving as a control unit and calculation unit, known input/output means such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and is equipped with a specified storage device as a built-in or external HDD or the like as a storage unit.
FIG. 4 shows an example of an in-mold pressure waveform, which is the in-mold pressure from the molten resin detected by the sensor 23, over time during one shot for molding one molded product.
The pressure data inside the mold during one shot is displayed as a pressure waveform with the horizontal axis (time) and the vertical axis (pressure).
The mold cavity pressure waveform can be analyzed by dividing it into multiple (here, three) sections along the horizontal axis of time. The first section a is analyzed as being related to the above-mentioned (1) injection speed and (2) VP switching position, the second section b is analyzed as being related to the above-mentioned (3) dwell pressure, and the third section c is analyzed as being related to the above-mentioned (4) dwell pressure time.
Here, (1) the injection speed refers to how fast the screw is advanced toward the nozzle. It may be set in stages. (2) the VP switching position refers to the position where the control method is switched from V (injection speed) to P (holding pressure) in relation to the advancement of the screw. The amount of resin to be injected is determined by the position of the screw, and the position where the injection starts is called the metering position, and the position where the injection ends is called the VP switching position. The image of the VP switching position is, for example, a position where about 95% of the resin is filled. If the value of the VP switching position is reduced, the short circuit of the molded product is gradually improved, and conversely, if the VP switching position is too small, the resin will enter more than the capacity and burrs will occur in the molded product. If the value of the VP switching position is reduced, more resin will enter, and the mold pressure will also become higher (larger). If the value of the VP switching position is changed, it will be affected by changes in the vertical axis (mold pressure) and horizontal axis (time) of the first section a. (3) Holding pressure refers to the pressure applied after the resin has been filled into the cavity of the mold until the gate is sealed (the gate solidifies) with a certain force to prevent the resin from flowing back from the gate. The unit is MPa or kgf/cm2. For example, 40 MPa is set for 2 seconds after the end of injection. Note that this may be set in stages. (4) Holding time refers to the time during which holding pressure is maintained.

Returning to FIG. 3, the display device 5 for the molding conditions (changed values) calculated using the AI of the information processing device 4 may be, for example, a monitor, tablet, etc., and may be supported by a predetermined stand or fixed to the molding machine 1 by a predetermined jig. The display device 5 is connected to the information processing device 4 by wire or wirelessly, etc.

FIG. 5 is a diagram for explaining the flow of processing of the molding condition determination support system 100 in this embodiment.
The mold pressure waveform for one shot for a given molded product is measured and obtained by a load cell 23 (pressure sensor) installed on the back of the ejector pin 22 of the mold, and this is carried out in two major ways. One is to obtain a reference waveform as a premise, and the other is to obtain the mold pressure waveform for each shot in the actual manufacturing stage. These two ways will be described in detail below.

First, to obtain the reference waveform as the above premise, that is, to obtain the reference waveform of the mold internal pressure in one shot for a specified molded product, the mold internal pressure is measured by the load cell 23 (pressure sensor) installed on the back of the ejector pin 22 of the mold, and the information processing device 4 obtains the reference waveform data. That is, molding is performed under conditions that produce good products (in the operation condition setting work to calculate the optimal operation conditions of the molding machine 1, the operation conditions that become the approximate standard based on experience are set and molding work is performed, while the operator refers to the process monitoring data and the molded product weight measurement value, etc., and visually checks the molded product to confirm the molding state, while performing several test injections, fine-tuning various operation conditions, molding conditions, and parameters, and adjusting them to the optimal operation conditions, etc.), and the mold internal pressure at that time is stored as a reference waveform in a table of a database in a specified memory unit 58 of the information processing device 4. This is because the pressure when the product is good and the pressure when the product is defective (strength, dimensions, surface condition, etc.) are different, and the aim is to minimize defective products by using the mold internal pressure waveform. For example, if there is little resin in the cavity, the pressure will be low, and if there is a lot of resin, the pressure will be high. In addition, differences in the tendency of the resin to cool due to variations in the resin temperature and mold temperature will also appear in the pressure. Note that the above database table or the like may store a reference waveform linked to each type of molded product.

As a further premise, the information processing device 4 generates a learning model by machine learning of AI, that is, the above-mentioned reference waveform for a predetermined molded product and the mold internal pressure waveform at each shot, as well as the change amount of each parameter and the resulting mold internal pressure waveform, etc. are used as learning data (teaching data) in advance, and a learning model is generated by learning which parameters among (1) injection speed, (2) VP switching position, (3) holding pressure, and (4) holding pressure time should be changed by how much (value, width) in order to effectively and efficiently correct the waveform so as to approach the reference waveform (ideally to match it). However, this learning model is updated every time new learning data (teaching data) is obtained regarding the difference between the above-mentioned reference waveform and the mold internal pressure waveform at each shot. Note that parameters such as temperature may also be considered as (5), etc.

In order to obtain the mold internal pressure waveform for each shot during the above-mentioned actual manufacturing stage, the mold internal pressure waveform for each shot for a specified molded product is measured by a load cell 23 (pressure sensor) installed on the back of the ejector pin 22 of the mold, and the information processing device 4 obtains the mold internal pressure waveform data.
Next, the information processing device 4 uses an AI-equipped molding condition setting support application or the like to compare the mold pressure waveform for each shot with a reference waveform and determine whether the difference exceeds a predetermined value.If it does, it determines that the difference is large (the deviation is large) and therefore the conditions need to be reset.
When the information processing device 4 judges that the above conditions need to be reset, the AI (artificial intelligence) calculates molding conditions that minimize the difference between the most recent mold pressure waveform and the reference waveform (matching the target reference waveform) using the above-generated learning model. That is, the most recent mold pressure waveform is compared with the reference waveform, and based on which part of the waveform has a deviation (there are various deviation patterns such as inclination), the molding conditions to be adjusted and their values are determined from among (1) injection speed, (2) VP switching position, (3) dwell pressure, and (4) dwell pressure time (molding condition optimization). Note that, when calculating based on the learning model, the order of adjustment of the molding conditions is fixed. That is, when adjusting the molding conditions, instead of searching all the conditions at once, adjustments are made in this order (sequence) of (1) injection speed, (2) VP switching position, (3) dwell pressure, and (4) dwell pressure time. The conditions (1) to (4) are not completely independent of each other, and some of them have mutually influencing relationships. Therefore, by determining the conditions one by one, the optimization process can be performed in fewer iterations than searching for all the conditions at once, and the adjustment can be performed in a procedure that is easy for the operator to understand (the adjustment values of each of the conditions (1) to (4) in chronological order can be easily visually confirmed). Here, the conditions are determined in order starting from (1), but it is also possible to determine any condition first, rather than in the order of (1) to (4). For example, the order may be (3), (4), (1), and (2). Note that parameters such as temperature may also be considered as (5) and the like. Of course, as a result of the calculation, the values to be adjusted among the molding conditions may become the same as the original values, and as a result, adjustment is not required.
Finally, the information processing device 4 outputs the molding conditions calculated by the AI, displays them on the display device 5, and instructs the worker to change them (recommends that you change these parameters), and the worker inputs the conditions into the controller 30 of the molding machine 1 by pressing a button or other operation, and resumes molding of the specified molded product.
FIG. 6 is a diagram showing an example of a display screen of the display device 5. The recommended values of the injection conditions (optimal conditions to be changed to these values) are arranged from top to bottom in the order of the molding time series of (1) injection speed, (2) VP switching position, (3) holding pressure, and (4) holding pressure time, so that it is intuitively easy to understand (easy to get used to), and the minimum items are displayed so that the setting change work is not complicated. The convenience of the operator is improved by outputting the conditions to be set next on the screen as a procedure that is easy for the operator to understand. As described above, any condition may be determined first. For example, in the case of the order of (3), (4), (1), and (2), the order may be displayed from the top. Also, FIG. 6 shows an example in which the number of sensors is six and the number of graphs (the most recent mold pressure waveform and the reference waveform) corresponding to the number of sensors are displayed.

FIG. 7 is a functional block diagram of the information processing device 4 of the molding condition determination support system 100 in this embodiment.
7, the information processing device 4 includes a communication unit 41, an acquisition unit 42, a calculation unit 43, a learning data storage unit 44, a model generation unit 45, a molding condition providing unit 46, an output unit 47, and a storage unit 48. Each unit may be realized by a known input/output I/F, a CPU, or a HDD. Some functions of the information processing device 4 may be realized as predetermined software, and the software may be clouded.

The communication unit 41 has the function of communicating with each device connected to a specified network.

The acquisition unit 42 has a function of acquiring the mold internal pressure data actually measured from the sensor 23 in chronological order via the communication unit 41, the instrumentation amplifier 2, the A/D converter 3, etc. As described above, there are two main situations: one is to acquire a reference waveform as a premise, and the other is to acquire the mold internal pressure waveform for each shot during the actual manufacturing stage.

The calculation unit 43 has a function of calculating the deviation between the reference waveform (reference waveform data) stored in the memory unit 48 and the waveform acquired by the acquisition unit 42 for each shot, and calculating how the waveform of the next shot would change if the parameters of the setting conditions (1) to (4) above were changed based on the amount of deviation (distance, etc.). Depending on various patterns in which the mold cavity pressure waveform deviates from the reference waveform, it calculates how close the waveform effectively and efficiently comes to the reference waveform by changing which parameter and by how much (value, range) among (1) injection speed, (2) VP switching position, (3) holding pressure, and (4) holding pressure time.

The learning data storage unit 44 associates data on how the waveform of the next shot changes when the parameters of the setting conditions (1) to (4) above are changed based on the deviation between the reference waveform and the waveform for each shot calculated by the calculation unit 43, and stores the data as learning data (teacher data) in the memory unit 48. Note that data on materials such as resin may also be associated and stored.

The model generation unit 45 generates a learning model based on the learning data by an AI (artificial intelligence) module, which describes the relationship between the change in the parameters of the setting conditions (1) to (4) above and the change in the parameters of the setting conditions (1) to (4) above, in the case of a reference waveform, and the change in the parameters of the setting conditions (1) to (4), in the case of a reference waveform, and stores the model in the storage unit 48. Specifically, the model generation unit 45 performs machine learning by regression analysis or the like to determine the relationship between the change in the parameters of the setting conditions (1) to (4) in the case of a reference waveform, and how much the waveform of each shot approaches the reference waveform. Since the regression method can be a known method such as linear regression, polynomial regression, or logistic regression, a detailed description will be omitted. The model generation unit 45 may perform machine learning by deep learning using a multilayered neural network based on a large amount of learning data, or may generate a learning model by other known machine learning or the like. In addition, parameters such as temperature may also be considered as (5).

The molding condition providing unit 46 has a function of comparing the mold internal pressure waveform for each shot with a reference waveform, judging whether the difference exceeds a predetermined value, and if so, judging that the difference is large (deviation is large) and therefore that the conditions need to be reset. If it is determined that the conditions need to be reset, it has a function of calculating, based on a learning model, how much the parameters of the setting conditions (1) to (4) above should be changed, and providing the molding conditions that minimize the difference between the most recent mold internal pressure waveform and the reference waveform (to match the target reference waveform).
As described above, the order of adjustment of molding conditions is fixed when performing calculation based on the learning model. That is, when adjusting molding conditions, instead of searching all conditions at once, adjustments are performed in this order (sequence) of (1) injection speed, (2) VP switching position, (3) holding pressure, and (4) holding pressure time. Since the conditions (1) to (4) are not completely independent of each other and have some mutually influencing relationships, by determining the conditions one by one, optimization processing can be performed in fewer iterations than searching for each condition at once, and adjustments can be performed in a procedure that is easy for the operator to understand (the adjustment values of each condition (1) to (4) in chronological order can be easily visually confirmed). Here, the conditions are determined in order from (1), but any condition may be determined first, not just in the order from (1) to (4) (the information processing device 4 holds information on this order set at some timing and uses it when performing calculation based on the learning model). For example, the order may be (3), (4), (1), and (2). The Bayesian optimization algorithm may be applied as the main algorithm of the AI used in calculating molding conditions. This is an optimization method that derives the next action to be taken (point to be explored) using the concepts of exploitation (using data/information obtained in the past) and exploration (going to obtain data that has not yet been obtained). In addition, parameters such as temperature may also be considered as (5).

The output unit 47 has a function of outputting various data such as image data and audio data by, for example, a display mechanism such as a display or an audio output mechanism such as a speaker. In particular, it outputs the change values and change amounts of the parameters of the setting conditions (1) to (4) calculated by the AI so that they can be displayed on the display device 5.

The memory unit 48 has a function of storing programs and various data for operating the information processing device 4. The memory unit 48 may be realized as a memory unit encrypted by blockchain. The database has predetermined tables etc. in which various data are linked and stored. In the tables etc. of the predetermined database, reference waveform data is stored in association with each type of molded product.

FIG. 8 is a flowchart showing the flow of main processes of the information processing device 4 of the molding condition determination support system in this embodiment.
First, the acquisition unit 42 of the information processing device 4 acquires a reference waveform from the mold internal pressure data actually measured by the sensor 23 via the communication unit 41, the instrumentation amplifier 2, the A/D converter 3, etc. (step S801).

Next, the model generating unit 45 generates a learning model of the relationship of how much the parameters of the setting conditions (1) to (4) above need to be changed in order for the waveform of each shot to approach the reference waveform in the case of a given reference waveform, based on the learning data (stored in the memory unit 48 by the learning data storage unit 44) to which data on how much the waveform of the next shot changes when the parameters of the setting conditions (1) to (4) above need to be changed is linked based on the deviation between the reference waveform and the waveform of each shot in the case of a given reference waveform calculated by the calculation unit 43 (step S802).
Next, the acquisition unit 42 acquires the mold internal pressure waveform for each shot during the actual manufacturing stage from the mold internal pressure data actually measured from the sensor 23 via the communication unit 41, the instrumentation amplifier 2, the A/D converter 3, etc. (step S803).

Next, the molding condition providing unit 46 compares the mold pressure waveform for each shot with the reference waveform and determines whether the difference is smaller than a predetermined value (step S804). If it is not smaller (step S804/No), it determines that the difference is large (deviation is large) and therefore the conditions need to be reset.
Next, the learning model generated in S802 is updated by adding information on the difference between the most recent mold pressure waveform and the reference waveform as learning data (step S805), and based on the updated learning model, it is calculated how much the parameters of the setting conditions (1) to (4) should be changed (step S806). As described above, when making calculations based on the learning model, the order of adjustment of the molding conditions is fixed in advance.

Then, the output unit 47 outputs the change values and change amounts of the parameters for each of the above setting conditions (1) to (4) calculated by the molding condition providing unit 46 to the display device 5 in a displayable manner (step S807). After that, molding is resumed under the conditions reset according to the output results, and the acquisition of the mold internal pressure waveform for each shot (step S803) and the determination of whether the difference between the mold internal pressure waveform and the reference waveform is smaller than a predetermined value (step S804) are repeated until the difference becomes smaller than the predetermined value.

According to the present embodiment described above, the processing procedure by the AI installed in the information processing device 4 is simplified, while the change values of the parameters of the molding conditions are displayed on the display device 5 along the time series of molding so that the worker can easily understand, thereby assisting in setting the molding conditions.

[Modification 1]
The information processing device and the result display unit (monitor) may be integrated together. In other words, the data output to the display device may be displayed on a predetermined display unit (such as a liquid crystal display) in the information processing device alone and used without being output to the display device.

[Modification 2]
The molding machine 10 may be configured to be integrated with the information processing device 4. This broadly includes a configuration in which the information processing device 4 is physically incorporated in the molding machine 10, or only its functions are realized in the molding machine 10. In this case, the display device 5 may also be integrated, or the display device 5 may not be provided (in which case parameter changes are fed back by automatic control).

Second Embodiment
Next, a second embodiment will be described. The first embodiment uses a pressure sensor as the sensor, whereas the second embodiment uses a temperature sensor as the sensor. In the second embodiment, a reference temperature waveform when a non-defective product is produced is obtained, and molding conditions related to temperature are adjusted so as to approach the reference temperature waveform.

Figure 9 is a diagram showing an example of the equipment configuration of a molding condition determination support system in the second embodiment. As shown in Figure 9, the molding condition determination support system 100b in the second embodiment is different from the first embodiment in Figure 3 in that the sensor 23 is changed to a sensor 23b. The sensor 23b detects the temperature of the mold. The sensor 23b may be embedded inside the mold, or the sensor 23b may be an ejector pin type temperature sensor in the shape of an ejector pin (ejector pin) 22.

Figure 10 is a functional block diagram of an information processing device of a molding condition determination support system in the second embodiment. In the information processing device of the second embodiment, the same elements as in the first embodiment are given the same symbols, and detailed descriptions thereof will be omitted as appropriate. The functions of the acquisition unit 42, calculation unit 43b, learning data storage unit 44b, model generation unit 45b, molding condition providing unit 46, and output unit 47 may be realized by the processor 40 reading and executing a program from the memory unit 48. The processing of each unit is the same as in the first embodiment, so details thereof will be omitted as appropriate.

The memory unit 48 is a storage device, and at least one reference waveform data is stored in the memory unit 48. The reference waveform data is waveform data detected by a sensor 23b provided on the molding machine or the mold when a specified molded product is molded as a good product. Here, the waveform data is data that represents a time series change in temperature.

The molding condition providing unit 46 acquires reference waveform data by referring to the memory unit 48, and acquires at least one piece of sensor waveform data for one shot during molding of the molded product from the sensor 23b provided on the molding machine or mold.
The molding condition providing unit 46 uses a machine learning model that receives as input the value of at least one setting condition and the waveform data during molding at the value of the setting condition, and learns to reduce the difference between the waveform data during molding and the reference waveform data, and processes the value of at least one setting condition and the waveform data during molding at the value of the setting condition, thereby outputting at least one setting condition value for which sensor 23b detects waveform data that is identical or similar to the reference waveform data.
Here, the machine learning model is, for example, a machine learning model (specifically, for example, Bayesian optimization) that aims to optimize a black box function that inputs at least one set condition value and waveform data during molding at the set condition value, and outputs an evaluation value based on the difference between the waveform data during molding and the reference waveform data. The set conditions are, for example, (1) injection speed, (2) VP switching position, (3) dwell pressure, and (4) dwell time.

In addition, the molding condition providing unit 46 may output values of the setting conditions for multiple processes (or sections) by, for example, determining the setting conditions for each process (or section) in a predetermined order among the multiple processes (or sections) included in one shot when molding a molded product.
Here, the process corresponds to the above-mentioned section. For example, the first section a in FIG. 4 is the above-mentioned injection filling process section, the second section b is the above-mentioned pressure holding process section, and the third section c is the above-mentioned cooling process section.

Also, similar to Figure 6 of the first embodiment, the molding condition providing unit 46 may, for example, output information for displaying the values of the setting conditions for each output process (or section) in order following the chronological order of the molding processes (or sections).

The temperature measured by sensor 23b is not limited to the temperature of the mold, and may be the temperature of a heater, as shown in Fig. 11. Fig. 11 is a schematic diagram showing the position of a sensor in a modified example according to the second embodiment. As shown in Fig. 11, heater 12 is provided so as to wrap around the outer surface of nozzle 11, and sensor 23b may be provided so as to be in contact with heater 12, for example.

In addition, a pressure sensor and a temperature sensor may be used together, in which case the molding conditions may be adjusted so that the pressure and temperature waveforms of one shot approach the reference pressure and temperature waveforms, respectively.

Third Embodiment
Next, the third embodiment will be described. In the third embodiment, the process of outputting the set condition value is changed compared to the first embodiment. FIG. 12 is a diagram showing an example of the device configuration of the molding condition determination support system in the third embodiment. As shown in FIG. 12, in the molding condition determination support system 100c in the third embodiment, the information processing device 4 is changed to an information processing device 4c compared to the first embodiment in FIG. 3.

Figure 13 is a functional block diagram of an information processing device of the molding condition determination support system in the third embodiment. As shown in Figure 13, compared to the information processing device 4 of the first embodiment, the information processing device 4c has the molding condition providing unit 46 changed to a molding condition providing unit 46c.

Figure 14 shows several examples of mold internal pressure waveforms over time during one shot to form a molded product. In Figure 14, the vertical axis is mold internal pressure and the horizontal axis is elapsed time. As Figure 14 shows, waveforms that result in a passable product, waveforms when there is a burr, waveforms when there are flow marks, waveforms when there is a short shot (a phenomenon in which the molding material does not completely fill the cavity), and waveforms when there is poor mold release are shown. As Figure 14 shows, there are several waveforms that result in a passable molded product.

Therefore, the information processing device 4c may obtain multiple pieces of reference waveform data that produce good products and store them in the memory unit 48. In this way, multiple pieces of reference waveform data may be stored.

In this case, the calculation unit 43 selects one of the multiple reference waveforms (reference waveform data) stored in the memory unit 48, calculates the deviation between the selected reference waveform data and the waveform for each shot acquired by the acquisition unit 42, and, based on the amount of deviation (distance, etc.), calculates how the waveform of the next shot will change if the parameters of the setting conditions (1) to (4) above are changed.

In addition to or instead of this, the molding condition providing unit 46c may identify multiple setting condition ranges that produce good products from multiple setting conditions corresponding to multiple reference waveform data. Specifically, for example, multiple setting condition ranges may be identified by selecting two setting conditions from the setting conditions when each of the multiple reference waveform data was obtained and extracting the range between the two selected setting conditions as the setting condition range.

The molding condition providing unit 46c may select one of the specified multiple setting condition ranges. Specifically, for example, when selecting one of the specified multiple setting condition ranges, the molding condition providing unit 46c may select the widest setting condition range.
Then, the molding condition providing unit 46c may output the value of the set condition based on the selected set condition range. Specifically, for example, the molding condition providing unit 46c may select and output the set condition from the selected set condition range, or may output the set condition obtained by statistically processing the set condition range (for example, calculating the average or median). Here, the set condition is, for example, (1) injection speed, (2) VP switching position, (3) dwell pressure, and (4) dwell pressure time.

Fourth Embodiment
Next, a fourth embodiment will be described. The way in which the waveform of the sensor changes when conditions are changed varies depending on the type of resin (e.g., polypropylene, polystyrene, etc.). In light of this, in the fourth embodiment, a machine learning model is provided for each type of resin or each characteristic of the resin (e.g., the viscosity of the resin).

Figure 15 is a diagram showing an example of the equipment configuration of a molding condition determination support system in the fourth embodiment. In the molding condition determination support system 100d in the fourth embodiment, compared to the first embodiment in Figure 3, the information processing device 4 is changed to an information processing device 4d.

Figure 16 is a functional block diagram of an information processing device of a molding condition determination support system in the fourth embodiment. Compared to the information processing device 4d of the first embodiment in Figure 7, the model generation unit 45 has been changed to a model generation unit 45d, the molding condition provision unit 46 has been changed to a molding condition provision unit 46d, and an input unit 49 has been added. The input unit 49 accepts input from the user (e.g., the type of resin or the characteristics of the resin).

17 is a schematic diagram showing an example of a machine learning model stored in the storage unit 10. As shown in FIG 17, machine learning models 48-1, ..., 48-N (N is an integer of 2 or more) are stored.
The model generating unit 45d may generate a machine learning model for each type of resin or each characteristic of the resin (e.g., the viscosity of the resin) and store the model in the storage unit 48 as shown in FIG. 17. The storage unit 48 may store a machine learning model for each type of resin or each characteristic of the resin (e.g., the viscosity of the resin). In this case, the acquiring unit 42 may acquire the type of resin or the characteristic of the resin input by the user. The molding condition providing unit 46d may select a machine learning model according to the type of resin or the physical property of the resin input by the user with reference to the storage unit 48, and output the value of at least one setting condition using the selected machine learning model.

Alternatively, the model generation unit 45d may generate a machine learning model in which the type of resin or the physical properties of the resin are included in the input parameters. More specifically, the model generation unit 45d may generate a machine learning model that learns to reduce the difference between the waveform data during molding and the reference waveform data by inputting the type of resin or the physical properties of the resin in addition to the value of at least one set condition and the waveform data during molding at the value of the set condition.

The memory unit 48 may store a machine learning model in which the type of resin or the physical properties of the resin are included in the input parameters. In this case, the acquisition unit 42 may acquire the type of resin or the properties of the resin input by the user. The molding condition providing unit 46d may output the value of at least one setting condition by inputting the type of resin or the properties of the resin input by the user into the machine learning model.

In each of the above embodiments, the data is stored in one storage unit, but the data may be distributed and stored in multiple storage units. In each of the above embodiments, the data is executed by one processor, but the data may be executed by multiple processors.

As described above, the information processing device according to the first to fourth embodiments is an information processing device that outputs information related to the setting conditions of a molding machine that molds a molded product by supplying molten material to a cavity of a mold (e.g., a metal die).The information processing device according to the first to fourth embodiments includes at least one storage device that stores at least one piece of reference waveform data that serves as a reference and is detected by a sensor provided in the molding machine (e.g., a heater) or the mold when a predetermined molded product is molded as a good product.
The information processing device according to the first to fourth embodiments further includes at least one processor that acquires reference waveform data by referring to the storage device, acquires at least one waveform data for one shot during molding of the molded product from the sensor, and processes at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to reduce the difference between the waveform data during molding and the reference waveform data by inputting the value of at least one setting condition and the waveform data during molding at the value of the setting condition, thereby outputting at least one value of the setting condition that the sensor detects as waveform data that is the same or similar to the reference waveform data. This simplifies the processing procedure by AI and makes it easier for an operator to acquire the setting conditions for a non-defective product.

Fifth embodiment
Next, the fifth embodiment will be described. In the fifth embodiment, the operation data of the molding machine for each molding shot (for example, motor current, motor torque, etc.) is monitored, and the information of the operation data is taken into account to adjust the conditions so that the waveform of the sensor during the molding shot approaches the reference waveform of the sensor. Here, the set conditions are values specified to the molding machine side to operate the molding machine. On the other hand, the operation data are values that represent the operation of the molding machine when the molding machine is operated under the set conditions, and the values change over time. For example, the VP switching position is a set condition because the position is specified by a value to the molding machine side, but the torque value is operation data because the magnitude changes over time when the molding machine operates under the set conditions.
Fig. 18 is an external view of a molding machine of a molding condition determination support system according to the fifth embodiment. As shown in Fig. 18, the molding machine 1 is provided with a motor 13 for driving a screw and a current sensor 14 for detecting a current of the motor 13.
Fig. 19 is a diagram showing an example of the device configuration of the molding condition determination support system in the fifth embodiment. In the molding condition determination support system 100e in the fifth embodiment, the information processing device 4 is changed to an information processing device 4e, compared with the first embodiment in Fig. 3.

20 is a functional block diagram of an information processing device of the molding condition determination support system in the fifth embodiment. The information processing device 4e includes a communication unit 41, an acquisition unit 42, a storage unit 48, a state estimator 51, and a condition adjustment unit 52. The communication unit 41 and the acquisition unit 42 are the same as those in the first embodiment, and therefore the description thereof will be omitted.
The storage unit 48 stores, for example, a learning model database (DB) 481. The storage unit 48 stores at least one piece of reference waveform data serving as a reference, which is waveform data detected by a sensor provided on the molding machine 1 or a mold when a specified molded product is molded as a non-defective product, and at least one piece of reference operation data serving as a reference, which is operation data of the controller 30 that controls the molding machine 1 when a non-defective product is molded.

For example, the state estimator 51 acquires the current value of the motor 13 detected by the current sensor 14 (hereinafter also referred to as the motor current value) as data related to the extrusion torque of the molding machine 1. Alternatively, the state estimator 51 may acquire the data related to the extrusion torque of the molding machine 1 from parameters input to the molding machine 1. Alternatively, the state estimator 51 may acquire from the controller 30 the motor current value included in the molding machine data received by the controller 30 from the molding machine 1 as data related to the extrusion torque of the molding machine 1.

The state estimator 51 obtains a parameter related to the viscosity of the resin to be poured into the mold (hereinafter, referred to as a viscosity parameter) from the data related to the extrusion torque.
In this way, the state estimator 51 estimates the physical properties of the resin in the molding machine 1 using the characteristic values (e.g., current or torque) of the motor 13 provided in the molding machine 1 and/or the characteristic values (e.g., current or torque) of the motor 13 stored in the controller 30.

The condition adjustment unit 52 includes a condition adjustment AI 521. Here, the learning model database 481 of the storage unit 48 stores, for example, a machine learning model (specifically, parameters of the machine learning model) for each physical property of the resin.
The condition adjustment AI 521 can record and read out a machine learning model corresponding to the physical properties of the resin (for example, a viscosity parameter) from the learning model database 481. Here, as an example, a machine learning model is recorded for each viscosity parameter.

The condition adjustment unit 52, for example, determines molding conditions based on data from the sensor 23 that detects pressure and the viscosity parameters output from the state estimator 51 to approach the sensor's reference waveform.

Specifically, for example, the condition adjustment unit 52 generates a first difference, which is the difference between the waveform data during molding and the reference waveform data. The condition adjustment unit 52 also generates a second difference, which is the difference between the operation data of the molding machine and the reference operation data. These first and second differences are then input to the condition adjustment AI 521.

The condition adjustment AI 521 uses as input the value of at least one set condition and the first difference and the second difference when the value of the set condition is used, and processes the value of at least one set condition and the first difference and the second difference when the value of the set condition is used, using a machine learning model that learns to reduce the difference between the waveform data at the time of molding and the reference waveform data, thereby outputting at least one value of a set condition at which the sensor 23 detects waveform data that is identical or similar to the reference waveform data.

Specifically, for example, the condition adjustment AI 521 uses a machine learning model (e.g., Bayesian optimization) aimed at optimizing a black box function that outputs an evaluation value based on a first difference and a second difference, and processes the value of at least one setting condition and the first difference and the second difference when the value of the setting condition is set, so as to optimize the evaluation value, thereby outputting the value of at least one setting condition at which the sensor detects waveform data that is identical or similar to the reference waveform data.

At that time, for example, the condition adjustment unit 52 may estimate the physical properties of the resin in the molding machine 1 using a characteristic value (e.g., current or torque) of a motor provided in the molding machine 1 and/or a characteristic value (e.g., current or torque) of a motor stored in the controller 30, and output the value of at least one setting condition using a machine learning model corresponding to the estimated physical properties of the resin.

Alternatively, the memory unit 48 may store a machine learning model including the physical properties of the resin as an input parameter. In this case, the condition adjustment unit 52 may estimate the physical properties of the resin of the molding machine 1 using a characteristic value (e.g., current or torque) of the motor 13 provided in the molding machine 1 and/or a characteristic value (e.g., current or torque) of the motor 13 stored in the controller 30, input the estimated physical properties of the resin to the machine learning model, and output the value of at least one setting condition.

Instead of the user manually inputting the value of the setting condition determined by the condition adjustment AI 521 to the molding machine 1, the condition adjustment unit 52 may directly output the value of at least one of the output setting conditions to the molding machine 1 so as to feed it back to the molding machine 1. This allows the setting conditions for a non-defective product to be set in the molding machine 1 without human intervention.

In the fifth embodiment, the data is stored in one storage unit, but the data may be distributed and stored in multiple storage units. In addition, in each of the above embodiments, the data is executed by one processor, but the data may be executed by multiple processors.

As described above, the information processing device in the fifth embodiment is an information processing device that outputs information about the setting conditions of a molding machine that molds a molded product by supplying molten material to a cavity of a mold (e.g., a metal mold). The information processing device in the fifth embodiment includes at least one storage device in which at least one reference waveform data that is a reference and is waveform data detected by a sensor provided on the molding machine or the mold when a specified molded product is molded as a good product, and at least one reference operation data that is a reference and is operation data of a controller that controls the molding machine when a good product is molded, are stored. The information processing device in the fifth embodiment further includes at least one processor that outputs at least one value of a setting condition that the sensor detects waveform data that is the same as or similar to the reference waveform data by processing the value of at least one setting condition and the first difference and the second difference when the value of the setting condition is used as input and learning to reduce the difference between the waveform data during molding and the reference waveform data. This simplifies the processing procedure by AI while facilitating the acquisition of good product setting conditions by an operator.

In addition, in the first to fifth embodiments, the at least one setting condition output is, for example, at least one of the injection speed, VP switching position, holding pressure, and holding pressure time.

Sixth embodiment
Next, a sixth embodiment will be described. The sixth embodiment is different in that an information processing system having the functions of any one of the first to fifth information processing devices is connected to a terminal device 6 on which a molding machine 1 is installed via a communication network.

Figure 21 is a schematic diagram of a molding condition determination support system in the sixth embodiment. As shown in Figure 21, compared to the first embodiment, the molding condition determination support system 100f is provided with a terminal device 6 connected to the output of the A/D converter 3, and the information processing device 4 is changed to an information processing system 4f. The information processing system 4f has the functions of any of the first to fifth information processing devices, and is composed of, for example, one or more information devices. The information processing system 4f is capable of communicating with the terminal device 6 via a communication circuit network CN. The information processing system 4f can receive various data (for example, sensor waveform data, operating data, setting conditions, etc.) from the terminal device 6. The information processing system 4f also transmits the values of the setting conditions obtained by processing to the terminal device 6.

The terminal device 6 is connected to the display device 5 and controls the display device to display information including the values of the setting conditions received from the information processing system 4f. The information processing system 4f may be constructed remotely (e.g., in the cloud). This makes it possible, for example, to construct the information processing system 4f remotely, to determine the values of the setting conditions for molding good products for multiple molding machines 1 installed in multiple locations, and to transmit the values of the setting conditions to the terminal devices 6 used by users of the multiple molding machines 1.

Note that the above-described embodiments are preferred embodiments of the present invention, and various modifications can be made within the scope of the present invention. At least a part of the information processing devices 4, 4b, 4c, 4d, 4e or the information processing system 4f described in the above-described embodiments may be configured with hardware or software. For example, a program for realizing at least a part of the functions of each device/system may be read into each device/system and executed to perform processing for realizing at least a part of the functions of each device/system. Furthermore, the program may be transmitted to another computer system by transmission waves via a computer-readable recording medium such as a CD-ROM or a magneto-optical disk, or via a transmission medium such as the Internet or a telephone line. Here, the recording medium is not limited to a removable one such as a magnetic disk or an optical disk, but may be a fixed recording medium such as a hard disk device or a memory. In addition, some systems may be realized through human actions.

Furthermore, the information processing devices 4, 4b, 4c, 4d, and 4e may be realized as an information processing system having one or more information devices. When the information processing devices 4, 4b, 4c, 4d, and 4e or the information processing system 4f have multiple information devices, at least one of them may be a computer, and the computer may execute a predetermined program to realize the function as at least one means of the information processing devices 4, 4b, 4c, 4d, and 4e or the information processing system 4f.

As mentioned above, the present invention is not limited to the above-mentioned embodiment as it is, and in the implementation stage, the components can be modified and embodied without departing from the gist of the invention. Furthermore, various inventions can be formed by appropriate combinations of multiple components disclosed in the above-mentioned embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, components from different embodiments may be appropriately combined.

REFERENCE SIGNS LIST 1 Molding machine 2 Instrumentation amplifier 3 A/D converter 4, 4b, 4c, 4d, 4e Information processing device 4f Information processing system 5 Display device 6 Terminal device 10 Injection unit 20 Mold clamping unit 21 Ejector plate 22 Ejector pin 23 Sensor 30 Controller 40 Display device 41 Cable 100, 100b, 100c, 100d, 100e, 100f Molding condition determination support system

Claims (11)

An information processing system that outputs information regarding setting conditions of a molding machine that molds a molded product by supplying molten material to a cavity of a mold,
At least one storage device in which at least one reference waveform data serving as a reference is stored, the reference waveform data being waveform data detected by a sensor provided in the molding machine or the mold when a predetermined molded product is molded as a good product;
a molding condition providing unit that acquires reference waveform data by referring to the storage device, acquires at least one piece of waveform data for one shot during molding of the molded product from the sensor, and processes at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to input at least one value of the setting condition and the waveform data during molding at the value of the setting condition so as to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of the setting condition detected by the sensor as waveform data that is the same as or similar to the reference waveform data;
An information processing system comprising: The information processing system according to claim 1, wherein the at least one setting condition is at least one of the injection speed, VP switching position, dwell pressure, and dwell pressure time. The storage device stores a plurality of the reference waveform data,
The information processing system according to claim 1, wherein the molding condition providing unit identifies multiple setting condition ranges that will produce good products from multiple setting conditions corresponding to the multiple reference waveform data, selects one setting condition range from the identified multiple setting condition ranges, and outputs the value of the setting condition based on the selected setting condition range. The information processing system according to any one of claims 1 to 3, wherein the storage device stores a machine learning model for each type of resin or each characteristic of the resin. The information processing system according to claim 4 , wherein the molding condition providing unit selects a machine learning model corresponding to the type of resin or physical properties of the resin input by a user by referring to the storage device, and outputs a value of the at least one setting condition using the selected machine learning model. The information processing system of claim 4, wherein the molding condition providing unit estimates physical properties of the resin of the molding machine using characteristic values of a motor provided in the molding machine and/or characteristic values of the motor stored in a controller that controls the molding machine, and outputs a value of the at least one setting condition using a machine learning model corresponding to the estimated physical properties of the resin. The information processing system according to claim 1, wherein the storage device stores a machine learning model in which the type of resin or the physical properties of the resin are included as input parameters. The information processing system according to claim 7 , wherein the molding condition providing unit outputs a value of the at least one set condition by inputting the type of resin or the characteristics of the resin into the machine learning model based on the type of resin or the physical properties of the resin input by a user. The information processing system of claim 7, wherein the molding condition providing unit estimates physical properties of the resin of the molding machine using characteristic values of a motor provided in the molding machine and/or characteristic values of the motor stored in a controller that controls the molding machine, inputs the estimated physical properties of the resin into a machine learning model, and outputs a value of the at least one setting condition. The information processing system according to claim 1, which is composed of one or more information processing devices. A computer capable of referencing at least one storage device storing at least one reference waveform data serving as a reference, the reference waveform data being detected by a sensor provided in a molding machine or a mold when a specified molded product is molded as a good product,
a step of acquiring reference waveform data by referring to the storage device, acquiring at least one piece of waveform data for one shot during molding of the molded product from the sensor, and processing at least one value of the setting condition and the waveform data during molding at the value of the setting condition using a machine learning model that learns to input at least one value of the setting condition and the waveform data during molding at the value of the setting condition so as to reduce the difference between the waveform data during molding and the reference waveform data, thereby outputting at least one value of the setting condition detected by the sensor as waveform data that is identical or similar to the reference waveform data;
A program for executing.

Images