Molding condition determination auxiliary device and resin state estimation device

Abstract

The molding condition determination auxiliary device and the resin state inference device include: a quality inference unit that infers the quality of a molded product through machine learning based on detection data; a quality change storage unit that accumulates the inferred quality of the molded product and stores the quality changes of multiple accumulated molded products; a trend evaluation unit that evaluates the quality change trend relative to a specified quality benchmark based on the quality change; a relationship storage unit that stores the relationship between the quality change trend and the correction amount of the molding condition used to return to the quality benchmark; and a correction condition determination unit that determines the correction amount of the molding condition based on the quality change trend and the relationship.

Description

Technical Field

[0001] The present disclosure relates to a resin state estimation device and a molding condition determination assisting device. Background Art

[0002] In a method of molding a molded product by supplying a molding material or a molten material obtained by melting a resin into a mold cavity of a molding machine, such as injection molding, when a defective product is produced, the operator needs to correct the molding conditions. Correcting the molding conditions requires skilled techniques. For an unskilled person, it is difficult to determine which molding condition should be changed and to what extent.

[0003] Therefore, in recent years, research related to artificial intelligence has been promoted. For example, Japanese Patent Publication No. 2020-49843 and Japanese Patent Publication No. 2020-49929 describe the correction amount of molding conditions determined by machine learning. In the technology described in Japanese Patent Publication No. 2020-49843, the relationship between the type of quality of the molded product and the type of molding condition is obtained in advance by machine learning, so that when a defect occurs in a certain quality type, it is output which molding condition needs to be corrected. In the technology described in Japanese Patent Publication No. 2020-49929, the correction amount of the molding condition is determined by machine learning based on the detection data detected by the sensor installed in the molding machine during molding.

[0004] In the above-mentioned related art, even if molding is performed under the same molding conditions, the quality of the molded product is sometimes different due to changes in the ambient temperature and the like. For example, even in one day, the ambient temperature is different in the morning and at noon. As the ambient temperature rises from morning to noon, the quality of the molded product sometimes changes. In addition, the same is true when the ambient temperature drops from noon to evening. In addition, with the change of seasons, the quality of the molded product sometimes gradually changes. In addition, by changing the production batch of the raw materials of the molding material, the quality of the molded product sometimes changes before and after the change. Therefore, in the case where the quality of the molded product changes due to external factors as described above, it is desired to correct the molding conditions in a manner that brings the quality of the molded product close to the quality benchmark.

[0005] In injection molding, it is known that the quality of the molded product varies depending on the melt state of the resin in the cavity. For example, the quality of the finished molded product varies depending on whether the resin in the cavity has high fluidity or low fluidity.

[0006] The molten state of the resin in the cavity is of course affected by the control parameters used for control in the injection molding machine, but it is also considered to be affected by other factors, such as the structure and function of the components constituting the injection molding machine, the ambient temperature, etc. That is, as long as the molten state of the resin in the cavity can be grasped, the appropriate correction amount of the molding conditions can be determined. However, it is not easy to grasp the molten state of the resin in the cavity.

[0007] Furthermore, even if molding is performed under the same molding conditions, the quality of the molded product may be different due to slight differences in the components of the raw materials of the molding material. Therefore, when the quality of the molded product changes due to external factors or slight differences in the components contained in the raw materials of the molding material as described above, it is desirable to correct the molding conditions so that the quality of the molded product approaches the quality standard. Summary of the invention

[0008] In view of the above-mentioned related technologies, the present disclosure provides a molding condition determination auxiliary device, which can modify the molding conditions when the quality of the molded product changes due to external factors so that the quality of the molded product approaches the quality benchmark.

[0009] In addition, the present disclosure provides a resin state estimation device capable of estimating the molten state of the resin in the cavity. Furthermore, the present disclosure provides a molding condition determination auxiliary device capable of correcting the molding conditions using the resin state estimation device so that the quality of the molded product approaches the quality standard.

[0010] Furthermore, the present disclosure provides a molding condition determination auxiliary device, which can correct the molding conditions so that the quality of the molded product approaches the quality standard when the quality of the molded product changes due to external factors or slight differences in the components contained in the raw materials of the molding material.

[0011] (1. Molding conditions determine auxiliary devices)

[0012] According to one embodiment of the present disclosure, a molding condition determination auxiliary device is a device used for determining molding conditions of a molding product in a molding method in which a molten material formed by melting a molding material is supplied to a cavity of a mold of a molding machine to mold a molding product. The auxiliary device includes: a detection data acquisition unit that acquires detection data detected by a sensor installed in the molding machine during molding; a quality estimation unit that estimates the quality of the molding product by machine learning based on the detection data; a quality transition storage unit that accumulates the estimated quality of the molding product and stores quality transitions of the accumulated plurality of molding products; a trend evaluation unit that evaluates a quality change trend relative to a predetermined quality benchmark based on the quality transition; a relationship storage unit that stores a relationship between the quality change trend and a correction amount of the molding condition for returning the quality to the quality benchmark; and a correction condition determination unit that determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit and the relationship stored in the relationship storage unit.

[0013] The quality transition storage unit accumulates the quality of the molded product inferred by machine learning and stores the quality transition. The quality transition refers to information that arranges the qualities of multiple molded products in the order of molding. Therefore, the trend evaluation unit can evaluate the quality change trend based on the quality transition of multiple continuous molded products.

[0014] In particular, the trend evaluation unit evaluates the quality change trend relative to the predetermined quality standard. For example, the trend evaluation unit can evaluate the state where the quality continues to deviate from the predetermined quality standard, the quality changes within the quality allowable range including the predetermined quality standard, etc. as the quality change trend.

[0015] Furthermore, the relationship between the quality change trend and the correction amount of the molding conditions is pre-stored in the relationship storage unit. This relationship can be set based on the experience of a skilled person, the output result of machine learning, the experimental result, etc. Furthermore, the correction condition determination unit determines the correction amount of the molding conditions based on the newly evaluated quality change trend and the relationship stored in the relationship storage unit. Here, the relationship stored in the relationship storage unit is related to the correction amount of the molding conditions used to return the quality to the prescribed quality standard. Therefore, when the molding conditions of the molding machine are corrected according to the correction amount of the molding conditions determined by the correction condition determination unit, the quality of the molded product to be molded next can be brought close to the prescribed quality standard.

[0016] That is, even if the quality of the molded product changes due to external factors such as ambient temperature, the molding conditions can be corrected by understanding the quality change trend so that the quality of the molded product can reach the specified quality standard. Therefore, not only skilled people, but also unskilled people can correct the molding conditions to improve the quality of the molded product.

[0017] (2-1. Resin state estimation device)

[0018] According to other embodiments of the present disclosure, a resin state inference device infers a melting state of a resin in a mold cavity of an injection molding machine, wherein the resin state inference device comprises: a detection data acquisition unit, which acquires detection data detected by a sensor installed on the injection molding machine during molding; a feature quantity generation unit, which generates a feature quantity group composed of a plurality of feature quantities related to the detection data based on the detection data; and a control parameter acquisition unit, which acquires a control parameter value group composed of a plurality of control parameter values ​​used for control in the injection molding machine; an identification parameter value calculation unit, which defines the melting state of the resin as represented by a feature quantity group and a control parameter value group, and calculates a resin state identification parameter value representing the melting state of the resin corresponding to each feature quantity based on the feature quantity group and the control parameter value group; and a group acquisition unit, which, when the melting state of the resin is defined as being classified into a plurality of groups, applies a multivariate analysis with the resin state identification parameter value set as an explanatory variable based on the resin state identification parameter value, thereby acquiring a group of the melting state of the resin.

[0019] The detection data detected by the sensor installed in the injection molding machine during molding is considered to be affected by the control parameters of the injection molding machine and the melting state of the resin in the cavity. In other words, the melting state of the resin is defined as represented by the feature value group generated by the detection data and the control parameter value group.

[0020] Using this definition, the identification parameter value calculation unit calculates the resin state identification parameter value representing the melting state of the resin corresponding to each feature value based on the feature value group of the detection data and the control parameter value group. That is, the resin state identification parameter values ​​are generated in the same number as the number of feature value types.

[0021] Furthermore, when the melting state of the resin is defined as being classified into a plurality of groups, the group acquisition unit applies a multivariate analysis using the resin state identification parameter value as an explanatory variable based on the resin state identification parameter value, thereby acquiring the group of the melting state of the resin. Here, the group of the melting state of the resin does not need to be clearly defined, but, for example, the degree of fluidity can be used as one of the factors for classification.

[0022] That is, the resin state estimation device performs calculation processing using the detection data and the control parameters, thereby being able to classify the groups of the melting states of the resin in the cavity during the molding of the molded product, for example, using the degree of resin fluidity as one of the factors. In this way, the melting states of the resin in the cavity can be grouped, and the correction amount of the molding condition corresponding to the group can be determined.

[0023] (2-2. Molding conditions determine auxiliary devices)

[0024] According to other embodiments of the present disclosure, a molding condition determination auxiliary device is used for a molding method for molding a molded product by supplying a molten material formed by melting a resin into a cavity of a mold of an injection molding machine, and determining the molding conditions of the molded product, wherein the device comprises: the above-mentioned resin state inference device; a quality inference unit, which infers the quality of the molded product through machine learning based on detection data; a quality transition storage unit, which accumulates the inferred quality of the molded product and stores the accumulated quality transitions of multiple molded products; a trend evaluation unit, which evaluates the quality change trend relative to a specified quality benchmark based on the quality transition; a relationship storage unit, which stores the relationship between the quality change trend and the correction amount of the molding condition used to return the quality to the quality benchmark in a corresponding relationship with a group of the molten state of the resin; and a correction condition determination unit, which determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit, the group of the molten state of the resin acquired by the group acquisition unit, and the relationship stored in the relationship storage unit.

[0025] That is, the correction amount of the molding conditions is determined using the group of the melting states of the resins acquired by the resin state estimation device. Thus, the correction amount of the molding conditions can be easily determined appropriately.

[0026] (3. Forming decision auxiliary device)

[0027] According to another aspect of the present disclosure, a molding condition determination assisting device is used in a molding method for molding a molded product by supplying a molten material obtained by melting a molding material into a cavity of a mold of a molding machine, and determines molding conditions of the molded product. The auxiliary device comprises: a detection data acquisition unit, which acquires detection data detected by a sensor installed on the above-mentioned molding machine during molding; a quality inference unit, which infers the quality of the above-mentioned molded product through machine learning based on the above-mentioned detection data; a quality transition storage unit, which accumulates the inferred quality of the above-mentioned molded product and stores the quality transition of the accumulated multiple above-mentioned molded products; a trend evaluation unit, which evaluates the quality change trend relative to a specified quality benchmark based on the above-mentioned quality transition; a molten state inference unit, which infers the molten state of the above-mentioned molten material in the above-mentioned cavity based on the above-mentioned detection data; a relationship storage unit, which establishes a corresponding relationship between the relationship between the above-mentioned quality change trend and the correction amount of the molding conditions used to return the above-mentioned quality to the above-mentioned quality benchmark and the above-mentioned molten state and stores it; and a correction condition determination unit, which determines the correction amount of the above-mentioned molding conditions based on the above-mentioned quality change trend evaluated by the above-mentioned trend evaluation unit, the above-mentioned molten state evaluated by the above-mentioned molten state inference unit, and the above-mentioned relationship stored in the above-mentioned relationship storage unit.

[0028] The quality transition storage unit accumulates the quality of the molded product inferred by machine learning and stores the quality transition. The quality transition refers to information that arranges the qualities of multiple molded products in the order of molding. Therefore, the trend evaluation unit can evaluate the quality change trend based on the quality transition of multiple continuous molded products.

[0029] In particular, the trend evaluation unit evaluates the quality change trend relative to the predetermined quality standard. For example, the trend evaluation unit can evaluate the state where the quality continues to deviate from the predetermined quality standard, the quality changes within the quality allowable range including the predetermined quality standard, etc. as the quality change trend.

[0030] Furthermore, the molten state inference unit infers the molten state of the molten material in the cavity based on the detection data. Here, the molten state depends on the components contained in the raw materials of the molding material. For example, the deviation of the components contained in the raw materials of the molding material includes the amount of water, the length of the reinforcing fiber, the proportion of the reinforcing fiber, the molecular weight of the main component, etc. Moreover, the molten state in the cavity affects the detection data during molding. Therefore, the molten state inference unit can infer the molten state by using the detection data depending on the molten state during actual molding.

[0031] The relationship between the quality change trend and the correction amount of the molding condition is stored in the relationship storage unit in advance in a corresponding relationship with the melting state of the molten material in the cavity. That is, in the relationship storage unit, the relationship between the quality change trend and the correction amount of the molding condition is stored for each type of the melting state of the molten material. This relationship can be set based on the experience of a skilled person, the output result of machine learning, the experimental result, etc.

[0032] The correction condition determination unit determines the correction amount of the molding condition based on the newly evaluated quality change trend, the newly inferred melting state of the molten material in the cavity, and the relationship stored in the relationship storage unit. Here, the relationship stored in the relationship storage unit is related to the correction amount of the molding condition for returning the quality to the specified quality standard. In particular, the correction amount of the molding condition is determined according to the melting state of the molten material in the cavity. Therefore, when the molding condition of the molding machine is corrected according to the correction amount of the molding condition determined by the correction condition determination unit, the quality of the molded product to be molded next can be brought close to the specified quality standard.

[0033] That is, even if the quality of the molded product changes due to external factors such as ambient temperature or slight differences in the components contained in the raw materials of the molding material, the molding conditions can be corrected by understanding the quality change trend and further understanding the melting state of the molten material in the cavity so that the quality of the molded product can be adjusted to the specified quality standard. Therefore, not only skilled people, but even unskilled people can correct the molding conditions so that the quality of the molded product is good. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a diagram showing the overall configuration of a molding machine system according to the first embodiment.

[0035] Figure 2 yes Figure 1 An enlarged view of the mold of the molding machine is shown.

[0036] Figure 3 yes Figure 2 A cross-sectional view of the mold along line III-III.

[0037] Figure 4 This is a functional block diagram showing a molding condition determination auxiliary device.

[0038] Figure 5 It is a graph showing the detection data.

[0039] Figure 6 This is a graph showing the quality transition.

[0040] Figure 7 It is a graph showing the quality change trend.

[0041] Figure 8 This is a diagram showing the relationship between the degree of quality deviation and the correction amount.

[0042] Fig. 9 This is a diagram for explaining a first example in the learning phase of the relationship between the degree of quality deviation and the correction amount.

[0043] Fig.10 This is a diagram for explaining a second example in the learning phase of the relationship between the degree of quality deviation and the correction amount.

[0044] Fig.11 This is a diagram showing the timing of the processing of the molding machine and the processing of the molding condition determination auxiliary device. This is a diagram showing the passage of time from top to bottom.

[0045] Fig.12 It is a diagram showing a configuration example of a molding machine system.

[0046] Fig.13 It is a diagram showing the overall configuration of a molding machine system of a first example according to the second embodiment.

[0047] Fig.14 This is a diagram showing the relationship among control parameters, the melting state of the resin, and detection data.

[0048] Fig.15 Schematically represents the relationship between the resin state identification parameter value.

[0049] Fig.16 It is a graph showing the detection data.

[0050] Fig.17 It is a schematic diagram of a group showing the molten state of resin.

[0051] Fig.18 It is a diagram showing the functional block structure of the resin state estimation device.

[0052] Fig.19 It is a diagram showing the overall configuration of a molding machine system of a second example according to the second embodiment.

[0053] Fig. 20 It is a diagram showing the functional block structure of the molding condition determination auxiliary device.

[0054] Fig.21 This is a graph showing the quality transition.

[0055] Fig. 22 It is a graph showing the quality change trend.

[0056] Fig.23 It is a diagram showing a group of molten states of resins.

[0057] Fig.24 This is a diagram showing the relationship between the degree of quality deviation and the correction amount.

[0058] Fig.25 It is a diagram showing a configuration example of a molding machine system.

[0059] Fig.26 It is a diagram showing the overall configuration of a molding machine system according to a third embodiment.

[0060] Fig. 27 yes Fig.26 An enlarged view of the mold of the molding machine is shown.

[0061] Fig.28 yes Fig. 27 A cross-sectional view of the mold along line III-III.

[0062] Fig.29 This is a functional block diagram showing a molding condition determination assisting device.

[0063] Fig.30 It is a graph showing the detection data.

[0064] Fig.31 This is a graph showing the quality transition.

[0065] Fig.32 It is a graph showing the quality change trend.

[0066] Fig.33 It is a diagram showing the types of molten states of the molten material.

[0067] Fig.34 This is a diagram showing the relationship between the degree of quality deviation and the correction amount.

[0068] Fig.35 This is a diagram for explaining the first example in the learning phase of the relationship between the degree of quality deviation and the correction amount.

[0069] Fig.36 This is a diagram for explaining a second example in the learning phase of the relationship between the degree of quality deviation and the correction amount.

[0070] Fig.37 This is a diagram showing the timing of the processing of the molding machine and the processing of the molding condition determination auxiliary device. This is a diagram showing the passage of time from top to bottom.

[0071] Fig.38 It is a diagram showing a configuration example of a molding machine system. DETAILED DESCRIPTION

[0072] The molding condition determination assisting device will be described below as a first embodiment. The present disclosure is not limited to the first embodiment, and various design changes can be made within the scope of the present disclosure.

[0073] (1. Applicable objects)

[0074] The molding condition determination auxiliary device is applicable to a molding method in which a molded product is molded by supplying a molten material formed by melting a molding material to a mold cavity of a molding machine. The molding machine of the applicable object can be, for example, an injection molding machine that performs injection molding of a resin or rubber as a molding material. In addition, other molding machines of the applicable object can also be, for example, a blow molding machine and a compression molding machine. In addition, regarding the resin as the molding material, examples can be given of thermoplastic resins such as monomeric polyamides and reinforced resins obtained by adding fillers to the base material of the thermoplastic resin. As fillers, micrometer-sized or nanometer-sized fillers can be cited. As fillers, for example, glass fibers, carbon fibers, etc. can be cited.

[0075] (2. Molding machine system 1)

[0076] Reference Figure 1 A molding machine system 1 including a molding condition determination auxiliary device is described. Figure 1 As shown, the molding machine system 1 includes a molding machine 2 and a molding condition determination assisting device 3 .

[0077] The molding machine 2 is an injection molding machine, a blow molding machine, or a compression molding machine. In this example, the molding machine 2 is an injection molding machine. The molding machine 2 molds, for example, a resin molded product. The molding condition determination auxiliary device 3 is a device for determining the molding conditions in the molding machine 2. In particular, in this example, the molding condition determination auxiliary device 3 determines a correction amount of the molding conditions for improving the quality of the molded product when the molded product is molded according to the applied molding conditions.

[0078] The molding condition determination auxiliary device 3 may be a device separate from the molding machine 2 or may be a device assembled to the molding machine 2. In addition, the molding condition determination auxiliary device 3 may be partially assembled to the molding machine 2, and the remaining part may be separated from the molding machine 2. In the case where all or part of the molding condition determination auxiliary device 3 is separated from the molding machine 2, the separated part may be configured to be connected to only one molding machine 2 or to be connected to a plurality of molding machines 2. In the latter case, the separated part of the molding condition determination auxiliary device 3 and the plurality of molding machines 2 constitute the same network, and become a structure capable of communicating with each other.

[0079] (3. Molding machine 2)

[0080] (3-1. Structure of Molding Machine 2)

[0081] Reference Figure 1The structure of an injection molding machine as an example of the molding machine 2 will be described. The molding machine 2 mainly includes a base 20 , an injection device 30 , a mold 40 , a mold clamping device 50 , and a control device 60 .

[0082] The injection device 30 is arranged on the base 20. The injection device 30 is a device that melts the molding material and applies pressure to the molten material to supply the molten material to the cavity C1 of the mold 40. The injection device 30 mainly includes a hopper 31, a heating cylinder 32, a screw 33, a nozzle 34, a heater 35, a driving device 36, and a sensor 37 for the injection device.

[0083] The hopper 31 is an inlet for pellets (granular molding material) as a raw material of the molding material. The heating cylinder 32 heats and melts the pellets put into the hopper 31 and pressurizes the formed molten material. In addition, the heating cylinder 32 is configured to be movable in the axial direction of the heating cylinder 32 relative to the base 20. The screw 33 is arranged inside the heating cylinder 32 and is configured to be rotatable and movable in the axial direction. The nozzle 34 is an injection port provided at the front end of the heating cylinder 32, and the molten material inside the heating cylinder 32 is supplied to the mold 40 by the axial movement of the screw 33.

[0084] The heater 35 is, for example, provided outside the heating cylinder 32 to heat the pellets inside the heating cylinder 32. The drive device 36 moves the heating cylinder 32 in the axial direction, rotates the screw 33 and moves it in the axial direction. The injection molding device sensor 37 is generally referred to as a sensor for obtaining the storage amount of the molten material, the holding pressure, the holding time, the injection speed, the state of the drive device 36, etc. However, the injection molding device sensor 37 is not limited to the above, and various information can also be obtained.

[0085] The mold 40 is a metal mold including a first mold 41 as a fixed side and a second mold 42 as a movable side. The mold 40 forms a cavity C1 between the first mold 41 and the second mold 42 by closing the first mold 41 and the second mold 42. The first mold 41 includes a supply path 43 (sprue, runner, gate) for guiding the molten material supplied from the nozzle 34 to the cavity C1. In addition, the mold 40 includes a pressure sensor 44 and a temperature sensor 45. The pressure sensor 44 detects the pressure applied to the molten material in the supply path 43. The temperature sensor 45 directly detects the temperature of the molten material in the supply path 43.

[0086] The mold clamping device 50 is disposed on the base 20 facing the injection device 30. The mold clamping device 50 opens and closes the attached mold 40 and prevents the mold 40 from opening due to the pressure of the molten material injected into the cavity C1 while the mold 40 is being tightened.

[0087] The mold clamping device 50 includes a fixed plate 51, a movable plate 52, a tie rod 53, a driving device 54, and a sensor 55 for the mold clamping device. The first mold 41 is fixed to the fixed plate 51. The fixed plate 51 can abut against the nozzle 34 of the injection device 30 to guide the molten material injected from the nozzle 34 to the mold 40. The second mold 42 is fixed to the movable plate 52. The movable plate 52 can approach and separate from the fixed plate 51. The tie rod 53 supports the movement of the movable plate 52. The driving device 54 is composed of, for example, a cylinder device, and moves the movable plate 52. The sensors 55 for the mold clamping device are generally referred to as sensors that obtain the mold clamping force, the metal mold temperature, the state of the driving device 54, etc.

[0088] The control device 60 controls the drive device 36 of the injection device 30 and the drive device 54 of the mold clamping device 50. For example, the control device 60 obtains various information from the injection device sensor 37 and the mold clamping device sensor 55 to control the drive device 36 of the injection device 30 and the drive device 54 of the mold clamping device 50 to perform actions corresponding to the action instruction data.

[0089] (3-2. Molding method)

[0090] The molding method of the molded product using the molding machine 2 is described. In the molding method using the molding machine 2, the metering process, the mold closing process, the injection filling process, the pressure holding process, the cooling process, and the demolding process are performed in sequence in one cycle. That is, in the molding of the next molded product, the above processes are performed again in sequence. Here, the metering process and the mold closing process constitute the starting preparation process, the injection filling process, the pressure holding process, and the cooling process constitute the molding process, and the demolding process constitutes the ending process. In addition, the initial stage of the demolding process (soon after the mold is opened) can also be included in the molding process, and the later stage can be used as the ending process.

[0091] In the metering process, the pellets are melted by the heating of the heater 35 and the shear friction heat accompanying the rotation of the screw 33, and the molten material is accumulated between the tip of the screw 33 and the nozzle 34 in the heating cylinder 32. Since the screw 33 retreats as the accumulated amount of the molten material increases, the accumulated amount of the molten material is metered according to the retreated position of the screw 33.

[0092] In the mold clamping process after the metering process, the movable platen 52 is moved to align the first mold 41 and the second mold 42 to perform mold clamping. In addition, the heating cylinder 32 is moved in the axial direction to approach the mold clamping device 50, and the nozzle 34 is connected to the fixed platen 51 of the mold clamping device 50. Next, in the injection filling process, the screw 33 is moved toward the nozzle 34 by using a predetermined thrust while the rotation of the screw 33 is stopped, so that the molten material is injection-filled into the mold 40 at a high pressure. When the molten material is filled into the cavity C1, it is transferred to the pressure holding process.

[0093] In the pressure holding process, the molten material is further pressed into the cavity C1 while the cavity C1 is filled with the molten material, and a pressure holding process is performed in which a predetermined pressure (pressure holding pressure) is applied to the molten material in the cavity C1 for a predetermined time. Specifically, a predetermined pressure holding pressure is applied to the molten material by applying a certain thrust to the screw 33.

[0094] After the predetermined holding pressure is applied for a predetermined time, the process proceeds to the cooling process. In the cooling process, the pushing of the molten material is stopped and the holding pressure is reduced to cool the mold 40. By cooling the mold 40, the molten material supplied to the mold 40 is solidified. Finally, in the demolding and removing process, the second mold 42 is separated from the first mold 41, and the molded product is removed.

[0095] (3-3. Mold 40)

[0096] Reference Figure 2 as well as Figure 3 The detailed structure of the mold 40 will be described. The mold 40 is a so-called multiple-acquisition metal mold, and a plurality of cavities C1 are formed in the mold 40. However, in order to simplify the drawings, Figure 2 as well as Figure 3 Only one cavity C1 is shown in the figure. In addition, in this example, the molded product molded by the molding machine 2 is a retainer for a constant velocity joint. Therefore, the molded product is annular, and the cavity C1 is formed into an annular shape that imitates the shape of the retainer. In addition, the shape of the molded product and the cavity C1 can also be a shape other than an annular shape, such as a C shape, a rectangular frame shape, etc.

[0097] The supply path 43 includes a nozzle 43a, a runner 43b, and a gate 43c. The nozzle 43a is a passage for supplying molten material from the nozzle 34. The runner 43b is a passage branching from the nozzle 43a, and the molten material supplied to the nozzle 43a flows into the runner 43b. The gate 43c is a passage for introducing the molten material flowing into the runner 43b into the cavity C1, and the flow path cross-sectional area of ​​the gate 43c is smaller than that of the flow path cross-sectional area of ​​the runner 43b. In the mold 40, the same number of runners 43b and gates 43c as the number of cavities C1 are formed, and the molten material supplied to the nozzle 43a is supplied to each cavity C1 via the runner 43b and the gate 43c.

[0098] Furthermore, when the cavity C1 is annular and the first mold 41 has one gate 43c, the inflow path of the molten material in the cavity C1 is a path that flows from the gate 43c along the annular circumferential direction of the cavity C1. That is, in the cavity C1, the molten material first flows into the vicinity of the gate 43c and finally flows into the farthest distance from the gate 43c.

[0099] In addition, the mold 40 is provided with a pressure sensor 44, and the pressure sensor 44 detects the pressure received by the molten material in the supply path 43. In this example, a plurality of pressure sensors 44 are provided. For example, the pressure sensor 44 is provided near the farthest position from the gate 43c and near the gate 43c in the cavity C1. In addition, the pressure sensor 44 may also be provided at the injection port 43a and the runner 43b. The pressure sensor 44 may be a contact sensor or a non-contact sensor.

[0100] The mold 40 is provided with a temperature sensor 45 for detecting the temperature of the molten material in the supply path 43. The temperature sensor 45 may be provided in the cavity C1, or in the sprue 43a or runner 43b, similarly to the pressure sensor 44. A plurality of temperature sensors 45 may be provided.

[0101] (4. Molding conditions determine the structure of auxiliary device 3)

[0102] Reference Figure 4-Figure 8 The structure of the molding condition determination auxiliary device 3 is described. The molding condition determination auxiliary device 3 includes, for example: an operation processing device including a processor, a storage device, an interface, etc.; an input device that can be connected to the interface of the operation processing device; and an output device that can be connected to the interface of the operation processing device. The output device may also include a display device, for example. In addition, the operation processing device, the input device, and the output device may also constitute a unit without an interface. In addition, a part of the operation processing device and a part of the storage device may also be applied to a physical server or a cloud server.

[0103] like Figure 4As shown, the molding condition determination support device 3 includes a detection data acquisition unit 101 , a quality estimation unit 102 , a quality transition storage unit 103 , a trend evaluation unit 104 , a relationship storage unit 105 , and a correction condition determination unit 106 .

[0104] The detection data acquisition unit 101 acquires detection data detected during molding by the sensors 44 and 45 installed in the molding machine 2. That is, the type of detection data acquired by the detection data acquisition unit 101 is at least one of the pressure received by the mold 40 from the molten material in the supply path 43 and the temperature of the molten material in the supply path 43.

[0105] The detection data based on the pressure sensor 44 is, for example, Figure 5 The data shown in Figure 5 In the figure, time T1 is the start time of filling, time T2 is the end time of filling and the start time of pressure holding, time T3 is the end time of pressure holding and the start time of cooling, and time T4 is the end time of cooling and the mold opening time. That is, the period between T1 and T2 is the injection filling process, the period between T2 and T3 is the pressure holding process, and the period between T3 and T4 is the cooling process. In addition, Figure 5 In the embodiment, the maximum pressure in the pressure holding process is set as the maximum pressure holding pressure Pmax, the pressure integral value in the pressure holding process is set as the pressure holding area Sa, and the pressure integral value in the cooling process is set as the cooling area Sb.

[0106] The quality inference unit 102 infers the quality of the molded product through machine learning based on the detection data acquired by the detection data acquisition unit 101. For example, the quality inference unit 102 infers a value in one or more quality categories in the molded product. The quality category of the molded product is at least one of the mass of the molded product, the size of the molded product, and the void volume in the molded product.

[0107] The quality estimation unit 102 generates a learned model that has learned the relationship between the detection data and the quality of the molded product in advance through machine learning. The learned model is generated for each quality type. Furthermore, the quality estimation unit 102 stores the learned model and uses the newly acquired detection data and the learned model to estimate the quality of the molded product.

[0108] The quality transition storage unit 103 accumulates the quality of the molded products estimated by the quality estimation unit 102, and stores the quality transition of the accumulated plurality of molded products. The quality transition is information that arranges the qualities of the plurality of molded products in the order of molding. The quality transition is, for example, Figure 6 Data as shown. Figure 6 This is an example of the quality of the molded product. Figure 6In the quality standard, Std is set, the upper limit of the quality allowable range is set as Thmax, and the lower limit of the quality allowable range is set as Thmin. That is, the upper limit Thmax and the lower limit Thmin mean that the product is qualified, and the deviation range means that the product is unqualified. Even if the product is qualified, it is ideally in the case of the prescribed quality standard Std.

[0109] exist Figure 6 In the initial stage of molding, the quality of most molded products shows a value close to the quality standard Std. However, as a sudden abnormality, there are molded products ( Figure 6 That is, when the molded products with sudden abnormalities are excluded, the quality of the molded products at the beginning of molding shows a value close to the quality standard Std ( Figure 6 A1). After that, if molding is continued, the quality of the molded product gradually deviates from the quality standard Std. Furthermore, the quality of the molded product shows a value of +N% higher than the quality standard Std ( Figure 6 A3).

[0110] The trend evaluation unit 104 evaluates the quality change trend relative to the quality standard Std based on the quality transition stored in the quality transition storage unit 103. Figure 7 As shown, the trend evaluation unit 104 evaluates the degree of deviation of the quality from the quality standard Std in each of the plurality of quality categories and the degree of dispersion of the quality of the plurality of molded products in each quality category as the quality change trend.

[0111] The trend evaluation unit 104 pre-sets the number of molded products with quality transitions for evaluating the quality change trend. That is, the trend evaluation unit 104 calculates the degree of deviation of the quality of the preset number of molded products. For example, the trend evaluation unit 104 calculates the degree (absolute value or relative value) of the average value of the quality of the number of molded products from the quality standard Std as the degree of deviation of the quality. Furthermore, the degree of deviation refers to the degree of quality deviation or stability of the preset number of molded products. In addition, the degree of deviation may also be expressed by values ​​such as standard deviation and variance.

[0112] like Figure 7 As shown, for example, the trend evaluation unit 104 evaluates the degree of deviation of quality as "+2.2%", which is a "stable" degree of deviation; evaluates the degree of deviation of size as "+0.3%", which is a "stable" degree of deviation; and evaluates the degree of deviation of void volume as "-0.5%", which is a "stable" degree of deviation.

[0113] In addition, Figure 6In the quality transition A1 shown in FIG. 1 , the trend evaluation unit 104 evaluates the quality change trend as being stable and located near the quality standard Std. Figure 6 In the quality transition A1 shown, the molded product A2, which indicates a quality change due to an unexpected abnormality, is excluded for evaluation. In this case, the trend evaluation unit 104 evaluates the degree of deviation from the quality standard Std as "0.1%" for the target quality type and evaluates the degree of deviation as "stable". Since the quality of the molded product returns to normal after the unexpected abnormality, the unexpected abnormality is not related to the molding conditions. Therefore, the evaluation is performed excluding the unexpected abnormality.

[0114] In addition, Figure 6 In the quality transition A3 shown, the trend evaluation unit 104 evaluates the quality change trend as stable and shows a value that deviates from the quality standard Std by about +N%. In this case, the trend evaluation unit 104 evaluates the degree of deviation from the quality standard Std as "+N%" for the quality type of the object, and evaluates the degree of deviation as "stable".

[0115] The relationship storage unit 105 stores the relationship between the quality change trend and the correction amount of the molding condition for returning the quality to the quality standard. Figure 8 As shown in FIG. 1 , the storage matrix shows the relationship between the level of deviation and the correction amount of the molding conditions for each quality type. For example, six levels are set as the level of deviation for each of quality, size, and void volume. In addition, the type of molding condition to be corrected is at least one of the injection speed, holding pressure, holding time, mold temperature during holding, cooling time, etc.

[0116] Here, the degree of relationship between the quality type and the type of molding condition, and the relationship between the degree of deviation of the quality type and the correction amount of the molding condition can be derived using machine learning. That is, the relationship stored in the relationship storage unit 105 can be generated by machine learning. Of course, the relationship can also be set based on experiments, past experience, etc. instead of machine learning.

[0117] The correction condition determination unit 106 determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit 104 and the relationship stored in the relationship storage unit 105. For example, the correction condition determination unit 106 determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit 104 and the relationship stored in the relationship storage unit 105. Figure 8 In the relationship represented by the matrix shown in FIG. 1 , the level closest to the degree of deviation evaluated by the trend evaluation unit 104 is determined, and the correction amount of the molding condition corresponding to the level is set as the correction amount of the determined molding condition. Figure 7 As shown, when the quality deviation is "+2.2%", Figure 8In the matrix shown, the level of deviation of quality "+2%" is selected.

[0118] Here, the correction condition determination unit 106 determines the correction amount of the molding condition for each of the plurality of quality types. Therefore, the correction condition determination unit 106 determines the final correction amount of the molding condition based on the plurality of correction amounts for the same type of molding condition. In this case, the correction condition determination unit 106 may, for example, use the value obtained by summing the plurality of correction amounts as the final correction amount, or may use the value obtained by further summing the values ​​multiplied by the weighting coefficients corresponding to the quality types as the final correction amount.

[0119] Furthermore, the correction condition determination unit 106 outputs the final correction amount to the control device 60 of the molding machine 2, and adds the correction amount to the molding conditions in the next molded product. Therefore, the control device 60 performs molding of the next molded product according to the corrected molding conditions. As a result, the quality of the molded product molded according to the corrected molding conditions can be made close to the quality standard Std.

[0120] (5. Determine the effect of auxiliary device 3 by molding conditions)

[0121] The effect of using the above-mentioned molding condition determination auxiliary device 3 is described. The quality transition storage unit 103 accumulates the quality of the molded product inferred by machine learning and stores the quality transition. The quality transition refers to information that arranges the qualities of multiple molded products in the order of molding. Therefore, the trend evaluation unit 104 can evaluate the quality change trend based on the quality transition of multiple continuous molded products.

[0122] In particular, the trend evaluation unit 104 evaluates the quality change trend relative to the predetermined quality standard Std. For example, the trend evaluation unit 104 can evaluate the state where the quality continues to deviate from the predetermined quality standard Std, the quality changes within the quality allowable range including the predetermined quality standard, etc.

[0123] Furthermore, the relationship between the quality change trend and the correction amount of the molding condition is pre-stored in the relationship storage unit 105. This relationship is set based on the experience of a skilled person, the output result of machine learning, the experimental result, etc. Furthermore, the correction condition determination unit 106 determines the correction amount of the molding condition based on the newly evaluated quality change trend and the relationship stored in the relationship storage unit 105.

[0124] Here, the relationship stored in the relationship storage unit 105 is related to the correction amount of the molding condition for returning the quality to the predetermined quality standard Std. Therefore, when the molding condition of the molding machine 2 is corrected according to the correction amount of the molding condition determined by the correction condition determination unit 106, the quality of the molded product to be molded next can be brought close to the predetermined quality standard Std.

[0125] That is, even if the quality of the molded product changes due to external factors such as ambient temperature, the molding conditions can be corrected by understanding the quality change trend so that the quality of the molded product can reach the specified quality standard Std. Therefore, not only skilled people, but even unskilled people can correct the molding conditions to improve the quality of the molded product.

[0126] (6. Learning phase in quality inference)

[0127] As described above, the quality estimation unit 102 estimates the quality by machine learning. A learned model is stored in the quality estimation unit 102. The learned model is generated in advance. The generation of the learned model, that is, an example of the learning stage will be described.

[0128] First, detection data for a plurality of molded products is obtained. Here, feature quantities of the detection data are extracted based on the detection data. For example, the maximum holding pressure Pmax in the holding process, the pressure deviation in the holding process, and the holding area Sa ( Figure 5 ), the actual holding time (the holding time obtained based on the pressure data), the speed of change of the pressure at the start of the holding process (the pressure differential value), etc. In addition, for the extraction of characteristic quantities, the cooling area Sb ( Figure 5 As shown in the figure), the speed of pressure change during the cooling process (pressure differential value), etc.

[0129] Furthermore, for the extraction of the characteristic quantity, the maximum temperature in the pressure holding process, the temperature deviation in the pressure holding process, the temperature area (temperature×time) in the pressure holding process, the cooling area (temperature×time) in the cooling process, the temperature change rate (differential value) in the cooling process, etc. are used based on the temperature data detected by the temperature sensor. In addition, since a plurality of pressure sensors 44 and temperature sensors 45 are provided in the mold 40, the above information is used for the extraction of the characteristic quantity for each sensor.

[0130] Then, the plurality of information described above is acquired for a plurality of molded products. Then, for each of the information described above, the maximum value, minimum value, average value, variance, etc. in the plurality of molded products are calculated and used as feature quantities.

[0131] On the other hand, the quality of a plurality of molded products is measured using an external measuring instrument or the like. For example, as quality, mass, size, void volume, etc. are measured. Furthermore, machine learning is performed using the feature quantity as an explanatory variable and the quality as a target variable, thereby generating a learned model. The learned model becomes a model that can input the feature quantity of the detection data and output the quality.

[0132] In the above description, a learning model is described in which the feature amount is used as an explanatory variable. However, in some cases, the learning model may not use the feature amount as an explanatory variable but use the detection data itself as an explanatory variable and output the quality as a target variable.

[0133] (7. Relationship-Determined Learning Stage)

[0134] The relationship between the degree of deviation of quality and the correction amount stored in the relationship storage unit 105 can be generated by, for example, using machine learning. Hereinafter, a case where machine learning is used in the generation of this relationship will be described.

[0135] (7-1. First example)

[0136] Reference Fig. 9 The first example of the learning phase is explained below. Fig. 9 As shown, the relationship between the quality type and the type of molding condition is directly generated by machine learning.

[0137] For example, by inputting the quality value in each quality category and the value of the molding condition, the degree of influence (contribution degree, influence degree) of the molding condition on the quality category can be obtained through machine learning. In addition, the degree of influence of the correction amount of the molding condition value on the quality value can also be obtained. As a result, the relationship information between the quality category and the type of molding condition can be obtained, and further the relationship information between the quality value and the correction amount of the molding condition can be obtained.

[0138] Based on the relationship information obtained through machine learning, people decide to use Figure 8 The relationship shown in the matrix is ​​shown. Of course, it can also be generated by machine learning Figure 8 The matrix shown.

[0139] (7-2. Second example)

[0140] Reference Fig.10 The second example of the learning phase is explained below. In the second example of the learning phase, Fig.10 As shown, the relationship between the quality type and the type of molding condition is indirectly generated by machine learning.

[0141] exist Fig. 9In the first example shown above, the relationship between the quality type and the type of molding condition is directly generated by machine learning. However, sometimes the relationship between the quality type and the type of molding condition cannot be directly obtained. Here, it is obvious that the molding conditions determine the state during molding, and the state during molding determines the quality. In other words, it can be said that the molding conditions and the quality are related by the state during molding. The state during molding refers to, for example, the maximum holding pressure Pmax, the holding area Sa, the cooling area Sb, etc.

[0142] Therefore, first, the quality value in each quality category and the characteristic quantity of the test data (the characteristic quantity obtained based on the maximum holding pressure, etc.) are input to obtain the degree of influence (contribution degree, influence degree) of the characteristic quantity of the test data on the quality category through machine learning. In addition, the degree of influence of the value of the characteristic quantity of the test data on the value of the quality can also be obtained. As a result, the relationship information between the quality category and the characteristic quantity of the test data can be obtained, and further the relationship information between the value of the quality and the value of the characteristic quantity of the test data can be obtained.

[0143] Next, the characteristic quantity of the detection data and the type of molding conditions are input to obtain the degree of influence (contribution degree, influence degree) of the type of molding conditions on the characteristic quantity of the detection data through machine learning. In addition, the degree of influence of the correction amount of the value of the molding condition on the value of the characteristic quantity of the detection data can also be obtained. As a result, the relationship information between the characteristic quantity of the detection data and the type of molding conditions can be obtained, and further the relationship information between the value of the characteristic quantity of the detection data and the correction amount of the value of the molding condition can be obtained.

[0144] In addition, the person determines the type of quality to be used based on the relationship information between the characteristic amount of the detection data and the relationship information between the characteristic amount of the detection data and the type of molding conditions. Figure 8 Of course, it can also be generated through machine learning Figure 8 The matrix shown.

[0145] (8. Motion timing of molding machine system 1)

[0146] (8-1. Basics)

[0147] Reference Fig.11 The timing of the operation of the molding machine system 1, in particular, the timing of the operation of the processing performed by the molding machine 2 and the processing performed by the molding condition determination auxiliary device will be described. First, the molding machine 2 continuously molds the molded products. For ease of description, the second molded product is molded after the first molded product.

[0148] like Fig.11As shown, the molding machine 2 performs a start preparation process (S11) for the first molded product. The start preparation process includes, for example, a metering process and a mold clamping process. Next, the molding machine 2 performs a molding process (S12) for the first molded product. The molding process includes, for example, an injection filling process, a pressure holding process, and a cooling process. Next, the molding machine 2 performs a finishing process (S13) for the first molded product. The finishing process includes, for example, a demolding and removing process.

[0149] However, when the molding condition determination support device 3 uses detection data detected by the pressure sensor 44 and the temperature sensor 45 at the beginning of the demolding process (soon after the mold is opened), the beginning of the demolding process may be included in the molding process.

[0150] After the first molded product is molded, the molding machine 2 starts the process related to the molding of the second molded product. First, the molding machine 2 performs the start preparation process for the second molded product (S21). Then, the molding machine 2 performs the molding process for the second molded product (S22). Then, the molding machine 2 performs the end processing process for the second molded product (S23).

[0151] On the other hand, the molding condition determination auxiliary device 3 is processed in parallel with the processing of the molding machine 2. Specifically, the primary processing step (S101) is processed in parallel with the detection of the data based on the pressure sensor 44 and the temperature sensor 45 when the first molded product is molded, and the secondary processing step (S102) is processed in parallel with the end processing step (S13) of the first molded product and the start preparation step (S21) of the second molded product. That is, the secondary processing step (S102) is executed in the preparation step of the molding machine 2 from the end of the molding step of the first molded product to the start of the molding step of the second molded product (the end processing step (S13) of the first molded product and the start preparation step (S21) of the second molded product).

[0152] The primary processing step (S101) includes at least processing performed by the detection data acquisition unit 101. The secondary processing step (S102) includes at least processing performed by the correction condition determination unit 106. The correction condition determination unit 106 determines the correction amount of the molding condition related to the second molded product. That is, the molding machine 2 makes the molding condition in the molding process (S22) of the second molded product a molding condition obtained by correcting it using the data in the molding process of the first molded product. In this way, the correction amount of the molding condition is determined within one cycle of molding of the molded product. Therefore, since it is possible to use the molding information not long ago in determining the correction amount of the molding condition, the correction amount of the molding condition can be made more accurately suitable for the current situation.

[0153] (8-2. First example)

[0154] The first example of the processing of the molding condition determination support device 3 is as follows. The primary processing step (S101) performs processing by the detection data acquisition unit 101. On the other hand, the secondary processing step (S102) performs processing by the quality estimation unit 102, the trend evaluation unit 104, and the correction condition determination unit 106.

[0155] (8-3. Second example)

[0156] The second example of the processing of the molding condition determination support device 3 is as follows. The primary processing step (S101) performs processing by the detection data acquisition unit 101 and processing by the quality estimation unit 102. On the other hand, the secondary processing step (S102) performs processing by the trend evaluation unit 104 and processing by the correction condition determination unit 106.

[0157] (9. Configuration Example of Molding Machine System 1)

[0158] (9-1. First example)

[0159] Reference Fig.12 The structure of the first example of the molding machine system 1 is described. Fig.12 As shown, the molding machine system 1 includes a plurality of molding machines 2, 2, edge computers 4, 4 integrally formed with the molding machines 2, 2, and a server 5 forming the same network with the plurality of molding machines 2, 2. In addition, the edge computers 4, 4 may constitute a part of the molding machines 2, 2, or may be separately formed from the molding machines 2, 2.

[0160] The edge computers 4 and 4 and the server 5 constitute the molding condition determination support device 3. The edge computers 4 and 4 include a detection data acquisition unit 101. The server 5 includes a quality estimation unit 102, a quality transition storage unit 103, a trend evaluation unit 104, a relationship storage unit 105, and a correction condition determination unit 106.

[0161] That is, the server 5 receives the detection data acquired by the detection data acquisition unit 101 from the edge computer 4 that is separate from the molding machine 2 or the edge computer 4 built into the molding machine 2. The server 5 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 2.

[0162] In this case, the server 5 can accumulate information related to a plurality of molding machines 2. Furthermore, by providing the server 5 with a processor capable of high-speed processing, the processing of the quality estimation unit 102, the processing of the trend evaluation unit 104, and the processing of the correction condition determination unit 106 can be high-speed processing. On the other hand, since it is not necessary to make each edge computer 4 high-specification, it is possible to achieve cost reduction.

[0163] (9-2. Second example)

[0164] The second example of the molding machine system 1 is similar to the first example. The molding machine system 1 includes a plurality of molding machines 2, 2, edge computers 4, 4 respectively connected to the molding machines 2, 2, and a server 5 constituting the same network as the plurality of molding machines 2, 2.

[0165] The edge computers 4, 4 include a detection data acquisition unit 101 and a quality estimation unit 102. The server 5 includes a quality transition storage unit 103, a trend evaluation unit 104, a relationship storage unit 105, and a correction condition determination unit 106. That is, the server 5 receives the quality of the molded product estimated by the quality estimation unit 102 from the edge computer 4 that is separate from the molding machine 2 or the edge computer 4 built into the molding machine 2. The server 5 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 2.

[0166] (9-3. The third example)

[0167] The third example of the molding machine system 1 is that the server 5 has all the functions. In this case, the edge computers 4, 4 are not required. The detection data acquisition unit 101 in the server 5 receives the detection data detected by the sensors 44, 45 from the molding machine 2. In addition, the correction condition determination unit 106 in the server 5 sends the correction amount of the molding condition to the molding machine 2.

[0168] The following describes a resin state estimation device and a molding condition determination support device as a second embodiment. The present disclosure is not limited to the second embodiment, and various design changes can be made within the scope of the present disclosure.

[0169] (1. Applicable objects)

[0170] The resin state estimation device and the molding condition determination auxiliary device are suitable for a molding method in which a molten material (resin) is supplied to the cavity of the mold of the injection molding machine to form a molded product. As for the resin used as the molding material, examples include thermoplastic resins such as monomeric polyamides and reinforced resins obtained by adding fillers to the base material of the thermoplastic resin. As fillers, micrometer-sized or nanometer-sized fillers can be cited. As fillers, for example, glass fibers, carbon fibers, etc. can be cited.

[0171] (2. Molding machine system 501A of the first example)

[0172] Reference Fig.13 A first example of a molding machine system 501A including a resin state estimation device 503 will be described. Fig.13 As shown, the molding machine system 501A includes an injection molding machine 502 (hereinafter referred to as “molding machine”) and a resin state estimation device 503 .

[0173] The molding machine 502 molds a resin molded product. The resin state estimation device 503 estimates the melting state of the resin in the cavity C2 of the mold 540 of the molding machine 502. The estimated melting state of the resin is used to determine molding conditions in the molding machine 502, for example.

[0174] The resin state estimation device 503 may be a device separate from the molding machine 502 or a device assembled to the molding machine 502. In addition, a part of the resin state estimation device 503 may be assembled to the molding machine 502, and the remaining part may be separated from the molding machine 502. In the case where all or a part of the resin state estimation device 503 is separated from the molding machine 502, the separated part may be a structure connected to only one molding machine 502 or a structure connected to a plurality of molding machines 502. In the latter case, the separated part of the resin state estimation device 503 and the plurality of molding machines 502 constitute the same network, and become a structure capable of communicating with each other.

[0175] (3. Molding machine 502)

[0176] (3-1. Structure of Molding Machine 502)

[0177] Reference Fig.13 The structure of the molding machine 502 will be described. The molding machine 502 mainly includes a base 520 , an injection device 530 , a mold 540 , a mold clamping device 550 , and a control device 560 .

[0178] The injection device 530 is arranged on the base 520. The injection device 530 is a device that melts the molding material (resin) and applies pressure to the molten material to supply the molten material to the cavity C2 of the mold 540. The injection device 530 mainly includes a hopper 531, a heating cylinder 532, a screw 533, a nozzle 534, a heater 535, a driving device 536, and a sensor 537 for the injection device.

[0179] The hopper 531 is an inlet for pellets (granular molding material) which are the raw materials of the molding material. The heating cylinder 532 heats and melts the pellets put into the hopper 531 and pressurizes the formed molten material. In addition, the heating cylinder 532 is configured to be movable in the axial direction of the heating cylinder 532 relative to the base 520. The screw 533 is arranged inside the heating cylinder 532 and is configured to be rotatable and movable in the axial direction. The nozzle 534 is an injection port provided at the front end of the heating cylinder 532, and the molten material inside the heating cylinder 532 is supplied to the mold 540 by the axial movement of the screw 533.

[0180] The heater 535 is, for example, provided outside the heating cylinder 532 to heat the pellets inside the heating cylinder 532. The driving device 536 moves the heating cylinder 532 in the axial direction, rotates the screw 533 and moves it in the axial direction. The injection molding device sensor 537 is generally referred to as a sensor for obtaining the storage amount of the molten material, the holding pressure, the holding time, the injection speed, the state of the driving device 536, etc. However, the injection molding device sensor 537 is not limited to the above, and various information can also be obtained.

[0181] The mold 540 is a metal mold including a first mold 541 as a fixed side and a second mold 542 as a movable side. The mold 540 forms a cavity C2 between the first mold 541 and the second mold 542 by closing the first mold 541 and the second mold 542. The first mold 541 includes a supply path 543 (injection gate, runner, gate) for guiding the molten material supplied from the nozzle 534 to the cavity C2. In addition, the mold 540 includes a pressure sensor 544 and a temperature sensor 545. The pressure sensor 544 detects the pressure applied to the molten material in the supply path 543. The temperature sensor 545 directly detects the temperature of the molten material in the supply path 543.

[0182] The mold clamping device 550 is disposed on the base 520 to face the injection device 530. The mold clamping device 550 opens and closes the attached mold 540 and prevents the mold 540 from opening due to the pressure of the molten material injected into the cavity C2 while the mold 540 is being tightened.

[0183] The mold clamping device 550 includes a fixed plate 551, a movable plate 552, a tie rod 553, a driving device 554, and a sensor 555 for the mold clamping device. A first mold 541 is fixed to the fixed plate 551. The fixed plate 551 can abut against the nozzle 534 of the injection device 530 to guide the molten material injected from the nozzle 534 to the mold 540. A second mold 542 is fixed to the movable plate 552. The movable plate 552 can approach and separate from the fixed plate 551. The tie rod 553 supports the movement of the movable plate 552. The driving device 554 is composed of, for example, a cylinder device, and moves the movable plate 552. The sensor 555 for the mold clamping device is generally referred to as a sensor for obtaining the mold clamping force, the metal mold temperature, the state of the driving device 554, and the like.

[0184] The control device 560 controls the drive device 536 of the injection device 530 and the drive device 554 of the mold clamping device 550. For example, the control device 560 obtains various information from the injection device sensor 537 and the mold clamping device sensor 555 to control the drive device 536 of the injection device 530 and the drive device 554 of the mold clamping device 550 to perform actions corresponding to the action instruction data.

[0185] (3-2. Molding method)

[0186] The molding method of the molded product using the molding machine 502 is described. In the molding method using the molding machine 502, in one cycle, the metering process, the mold closing process, the injection filling process, the pressure holding process, the cooling process, and the demolding process are performed in sequence. That is, in the molding of the next molded product, the above processes are performed again in sequence. Here, the metering process and the mold closing process constitute the starting preparation process, the injection filling process, the pressure holding process, and the cooling process constitute the molding process, and the demolding process constitutes the ending process. In addition, the initial stage of the demolding process (soon after the mold is opened) can also be included in the molding process, and the later stage can be used as the ending process.

[0187] In the metering process, the pellets are melted by the heating of the heater 535 and the shear friction heat accompanying the rotation of the screw 533, and the molten material is accumulated between the tip of the screw 533 and the nozzle 534 in the heating cylinder 532. Since the screw 533 retreats as the accumulated amount of the molten material increases, the accumulated amount of the molten material is measured according to the retreated position of the screw 533.

[0188] In the mold clamping process after the metering process, the movable platen 552 is moved to align the second mold 542 with the first mold 541, and the molds are clamped. In addition, the heating cylinder 532 is moved in the axial direction to approach the mold clamping device 550, and the nozzle 534 is connected to the fixed platen 551 of the mold clamping device 550. Next, in the injection filling process, the screw 533 is moved toward the nozzle 534 by using a predetermined thrust while the rotation of the screw 533 is stopped, so that the molten material is injection-filled into the mold 540 at a relatively high pressure. When the molten material is filled into the cavity C2, the process is transferred to the pressure holding process.

[0189] In the pressure holding process, the molten material is further pressed into the cavity C2 while the cavity C2 is filled with the molten material, and a pressure holding process is performed in which a predetermined pressure (pressure holding pressure) is applied to the molten material in the cavity C2 for a predetermined time. Specifically, a predetermined pressure holding pressure is applied to the molten material by applying a certain thrust to the screw 533.

[0190] After the predetermined holding pressure is applied for a predetermined time, the process proceeds to the cooling process. In the cooling process, the injection of the molten material is stopped and the holding pressure is reduced to cool the mold 540. By cooling the mold 540, the molten material supplied to the mold 540 is solidified. Finally, in the demolding and removing process, the second mold 542 is separated from the first mold 541, and the molded product is removed.

[0191] (4. Basics of Estimating the Molten State of Resin)

[0192] Reference Figure 14-17 The basics of estimating the molten state of the resin in the cavity C2 will be described. In this example, the molten state of the resin is defined as being classified into a plurality of groups, and the groups for estimating the molten state of the resin are used as the estimation of the molten state of the resin.

[0193] Fig.14 The control parameters and the melting state of the resin affect the detection data during molding. Here, the melting state of the resin depends on the components contained in the raw materials of the molding material. For example, the deviation of the components contained in the raw materials of the molding material includes the water content, the length of the reinforcing fiber, the proportion of the reinforcing fiber, the molecular weight of the main component, etc. Moreover, as mentioned above, the melting state of the resin in the cavity C2 affects the detection data during molding.

[0194] Specifically, the detection data detected by the sensors 544 and 545 installed in the molding machine 502 during molding are considered to be affected by the control parameters used for control in the molding machine 502 and the melting state of the resin in the cavity C2. In other words, the melting state of the resin affects the detection data and the control parameters.

[0195] Regarding the above relationship, the relationship with the value is expressed as Fig.15 . Here, the information as the value related to the detection data during molding is a feature quantity group [F] composed of multiple feature quantities related to the detection data. The feature quantity refers to the statistics (maximum value, minimum value, average value, variance, maximum value of differential, minimum value of differential, integral value, etc.) in the detection data of the pressure sensor 544 in each molding process (injection filling process, pressure holding process, cooling process, etc.), and the above statistics in the detection data of the temperature sensor 545 in each molding process. When multiple pressure sensors 544 and temperature sensors 545 are provided, the statistics in each sensor 544, 545 are used as feature quantities. In this way, multiple feature quantities can be obtained, and the above multiple feature quantities are collectively referred to as a feature quantity group [F].

[0196] For example, using Fig.16 The detection data of the pressure sensor 544 shown in FIG. 5 is used to describe a part of the feature quantity. In the detection data of the pressure sensor 544, as shown in FIG. Fig.16 As shown, time T1 is the start time of filling, time T2 is the end time of filling and the start time of pressure holding, time T3 is the end time of pressure holding and the start time of cooling, and time T4 is the end time of cooling and the mold opening time. That is, the period between T1 and T2 is the injection filling process, the period between T2 and T3 is the pressure holding process, and the period between T3 and T4 is the cooling process. Fig.16In the figure, the maximum pressure in the pressure holding process is Pmax, the pressure holding area as the pressure integral value in the pressure holding process is Sa, and the cooling area as the pressure integral value in the cooling process is Sb. Moreover, the maximum value Pmax, the pressure holding area Sa, and the cooling area Sb are part of the characteristic quantities.

[0197] In addition, the information of the value related to the control parameter is a plurality of control parameter values ​​used for control in the molding machine 502, and is set as a control parameter value group [A] composed of a plurality of control parameter values. In addition, the unit of each control parameter value can also be adjusted to become a predetermined value of the number of digits of the control parameter value. The types of control parameter values ​​are, for example, injection speed, holding pressure, holding time, mold temperature during holding pressure, cooling time, nozzle temperature, injection pressure (nozzle pressure), etc.

[0198] In addition, information as a value related to the molten state of the resin is a resin state identification parameter value indicating the molten state of the resin. Here, the resin state identification parameter value is not a value itself having meaning, but an indicator for grouping the molten state of the resin. Moreover, the resin state identification parameter value is a value corresponding to each feature quantity. That is, the resin state identification parameter value exists in the same number as the number of types of feature quantities. Therefore, a plurality of resin state identification parameter values ​​are set as a resin state identification parameter value group [P].

[0199] Moreover, if Fig.15 As shown, the resin state identification parameter value group [P] is represented by the feature value group [F] of the detection data during molding and the control parameter value group [A]. In detail, the resin state identification parameter value group [P] is defined as the value obtained by dividing the feature value group [F] of the detection data during molding by the control parameter value group [A].

[0200] Here, the content expressed as a numerical expression is equation (1). As shown in the condition of equation (1), the resin state identification parameter value group [P], the feature value group [F], and the control parameter value group [A] are matrices represented by the resin state identification parameter values ​​P1-Pm, the feature values ​​F1-Fm, and the control parameter values ​​A(F1)-A(Fm), respectively.

[0201]

[0202] in,

[0203]

[0204] The control parameter value A(F1)-A(Fm) is expressed by equation (2). The control parameter value A(Fj) is the infinite product of a plurality of control parameter values ​​Ak. The control parameter value A(Fj) is applied to either of the two types described below.

[0205] A(Fj)=ΠAk…(2)

[0206] The first control parameter value A(Fj) is the infinite product of all control parameter values ​​Ak. For example, A(F1) and A(F2) are as shown in equations (3) and (4).

[0207] A(F1)=A1×A2×A3×…×Am…(3)

[0208] A(F2)=A1×A2×A3×…×Am…(4)

[0209] The second control parameter value A(Fj) is an infinite product of a portion of the control parameter values ​​Ak. For example, A(F1) and A(F2) are as shown in equations (5) and (6).

[0210] A(F1)=A3×A4×A9×…×Am-1…(5)

[0211] A(F2)=A1×A2×A4×…×Am-2…(6)

[0212] Among the second control parameter values ​​A(Fj), a part of the control parameter values ​​Ak is one or more control parameter values ​​having a high degree of influence on the feature quantity Fj.

[0213] Next, if a resin state identification parameter value group for each detection data is obtained using a plurality of detection data, the resin state identification parameter value group is used to perform multivariate analysis, thereby grouping the melting state of the resin. In other words, multivariate analysis is performed using a plurality of resin state identification parameter values ​​as explanatory variables.

[0214] For ease of explanation, it is assumed that the resin state identification parameter values ​​are P1 and P2. In this case, if Fig.17 As shown in FIG. 1 , the two-dimensional coordinate system is represented by using P1 and P2 as explanatory variables. In this two-dimensional coordinate system, points P1 and P2 of each detection data are plotted.

[0215] Fig.17 The resin state identification parameter values ​​obtained using the detection data and control parameter values ​​acquired in the learning phase are plotted, and the group of the molten state of the resin is divided into two groups, for example, G1 and G2. In this case, the group of the molten state of the resin obtained using the newly acquired detection data and control parameter values ​​is classified into either G1 or G2.

[0216] In particular, the resin state identification parameter value can be set as an explanatory variable, the group of the molten state of the resin can be set as a target variable, and cluster analysis can be applied as a multivariate analysis. In this case, a learning model can be generated by performing machine learning of cluster analysis using a training data set including the explanatory variable and the target variable. Furthermore, the group of the molten state of the resin can be determined using the generated learning model.

[0217] Here, in the cluster analysis, the number of groups of the molten state of the resin is preset. That is, the learned model is generated in a manner classified into the preset number of groups. In addition, although the number of groups will be described later, when the molding conditions are corrected by the correction amount of the molding conditions set according to each group of the molten state of the resin, it can be set to a number that can obtain the desired quality of the molded product.

[0218] (5. Structure of Resin State Estimation Device 503)

[0219] Reference Fig.18 The structure of the resin state inference device 503 is described. The resin state inference device 503 is a device for obtaining the above-mentioned group of molten states of the resin. The resin state inference device 503 includes, for example: an operation processing device including a processor, a storage device, an interface, etc.; an input device that can be connected to the interface of the operation processing device; and an output device that can be connected to the interface of the operation processing device. The output device may also include a display device, for example. In addition, the operation processing device, the input device, and the output device may also constitute a unit without an interface. In addition, a part of the operation processing device and a part of the storage device may also be applied to a physical server or a cloud server.

[0220] like Fig.18 As shown, the resin state estimation device 503 includes a detection data acquisition unit 601, a feature value generation unit 602, a control parameter acquisition unit 603, an identification parameter value calculation unit 604, a learned model generation unit 605, a learned model storage unit 606, and a group acquisition unit 607.

[0221] The detection data acquisition unit 601 acquires detection data detected by the sensors 544 and 545 installed in the molding machine 502 during molding. For example, the detection data detected by the pressure sensor 544 is Fig.16 Although not shown in the figure, detection data detected by the temperature sensor 545 can also be acquired as time series data.

[0222] The feature quantity generating unit 602 generates a feature quantity group [F] composed of multiple feature quantities related to the detection data based on the detection data (refer to formula (1)). That is, multiple feature quantities are generated for each detection data. The feature quantity is a statistic (maximum value, minimum value, average value, variance, maximum value of differential, minimum value of differential, integral value, etc.) in the detection data of each sensor 544, 545 in each molding process (injection filling process, pressure holding process, cooling process, etc.).

[0223] The control parameter acquisition unit 603 acquires a control parameter value group [A] (see formula (1)) composed of a plurality of control parameter values ​​used for control in the molding machine 502. The types of control parameter values ​​include, for example, injection speed, holding pressure, holding time, mold temperature during holding pressure, cooling time, nozzle temperature, injection pressure (nozzle pressure), etc.

[0224] The identification parameter value calculation unit 604 calculates the resin state identification parameter values ​​P1-Pm representing the molten state of the resin corresponding to each feature value based on the feature value group [F] and the control parameter value group [A]. The identification parameter value calculation unit 604 calculates the resin state identification parameter values ​​P1-Pm according to the above-mentioned equations (1) and (2).

[0225] The identification parameter value calculation unit 604 may calculate the resin state identification parameter value Pj using all the control parameter values ​​Ak in each control parameter A(F1)-A(Fm) as in the above equations (3) and (4). In this case, the resin state identification parameter value Pj is the infinite product of all the control parameter values ​​Ak.

[0226] The identification parameter value calculation unit 604 may also calculate the resin state identification parameter value Pj corresponding to the feature value Fj using the feature value Fj and one or more control parameter values ​​Ak having a high degree of influence on the feature value Fj, as in the above-mentioned equations (5) and (6). In this case, the resin state identification parameter value Pj is an infinite product of a part of the control parameter values ​​Ak.

[0227] For example, one or more control parameter values ​​Ak having a high degree of influence on the feature quantity Fj can also be extracted by using the feature quantity Fj of the object and the control parameter value group [A] and using machine learning. For example, the influence coefficient obtained by machine learning can be used to extract a predetermined number of control parameter values ​​Ak from the control parameters having a high influence coefficient. The influence coefficient is, for example, a lasso coefficient obtained by lasso regression, a ridge coefficient obtained by ridge regression, and the like.

[0228] The learned model generation unit 605 generates a learned model of machine learning as a learning stage. When the melting state of the resin is defined as being classified into a plurality of groups G1, G2, ..., the learned model generation unit 605 sets the resin state identification parameter values ​​P1-Pm as explanatory variables, sets the groups G1, G2, ... of the melting state of the resin as target variables, and applies cluster analysis as multivariate analysis.

[0229] The learned model generation unit 605 performs machine learning of cluster analysis using the training data set including the above-mentioned explanatory variables and target variables, thereby generating a learned model. The generated learned model is stored in the learned model storage unit 606. That is, when the resin state identification parameter value is input, the learned model outputs a group.

[0230] The group acquisition unit 607 acquires the groups G1, G2, ... of the molten states of the resin based on the resin state identification parameter values ​​P1-Pm. In this example, the group acquisition unit 607 uses a learned model generated by applying a cluster analysis as a multivariate analysis using the resin state identification parameter values ​​P1-Pm as explanatory variables. That is, the group acquisition unit 607 acquires the groups G1, G2, ... of the molten states of the resin as outputs of the learned model when the resin state identification parameter values ​​P1-Pm are input as an application stage (also called an inference stage) of machine learning.

[0231] As described above, the resin state estimation device 503 determines a set of melting states of the resin in the cavity C2 when the target molded product is molded, using the detection data of the target molded product when being molded and the control parameter values ​​used for molding the target molded product.

[0232] (6. Effect of using the resin state estimation device 503)

[0233] As described above, the detection data detected by the sensors 544 and 545 installed in the molding machine 502 during molding are considered to be affected by the control parameters of the molding machine 502 and the melting state of the resin in the cavity C2. In other words, the melting state of the resin is defined as represented by the feature value group [F] generated by the detection data and the control parameter value group [A].

[0234] Using this definition, the identification parameter value calculation unit 604 calculates the resin state identification parameter values ​​P1-Pm representing the molten state of the resin corresponding to each feature value F1-Fm based on the feature value group [F] of the detection data and the control parameter value group [A]. That is, the resin state identification parameter values ​​P1-Pm are generated in the same number as the number of types of feature values ​​F1-Fm.

[0235] Furthermore, when the melting state of the resin is defined as being classified into a plurality of groups G1, G2, ..., the group acquisition unit 607 applies multivariate analysis using the resin state identification parameter values ​​P1-Pm as explanatory variables based on the resin state identification parameter values ​​P1-Pm, thereby acquiring the groups G1, G2, ... of the melting state of the resin. Here, the groups G1, G2, ... of the melting state of the resin do not need to be clearly defined, but can be classified using, for example, the degree of fluidity as one of the factors.

[0236] That is, the resin state estimation device 503 performs calculation processing using the detection data and the control parameters, thereby being able to classify the groups G1, G2, ... of the melting states of the resin in the cavity C2 of the molded product during molding, for example, using the degree of resin fluidity as one of the factors. In this way, the melting states of the resin in the cavity C2 can be grouped, and the correction amount of the molding condition corresponding to the groups G1, G2, ... can be determined.

[0237] (7. Molding machine system 501B of the second example)

[0238] Reference Fig.19 A second example of a molding machine system 501B including a molding condition determination assisting device 504 will be described. Fig.19 As shown, the molding machine system 501B includes a molding machine 502 and a molding condition determination assisting device 504 .

[0239] The molding condition determination auxiliary device 504 is a device for determining the molding conditions in the molding machine 502. In particular, in this example, the molding condition determination auxiliary device 504 determines the correction amount of the molding conditions for improving the quality of the molded product when the molded product is molded according to the applied molding conditions. In addition, the molding condition determination auxiliary device 504 includes the above-mentioned resin state estimation device 503, and performs processing using the group of molten states of the resin obtained by the resin state estimation device 503.

[0240] (8. Molding conditions determine the structure of the auxiliary device 504)

[0241] Reference Figure 20-24 The structure of the molding condition determination auxiliary device 504 is described. The molding condition determination auxiliary device 504 includes, for example: an operation processing device including a processor, a storage device, an interface, etc.; an input device that can be connected to the interface of the operation processing device; and an output device that can be connected to the interface of the operation processing device. The output device may also include a display device, for example. In addition, the operation processing device, the input device, and the output device may also constitute a unit without an interface. In addition, a part of the operation processing device and a part of the storage device may also be applied to a physical server or a cloud server.

[0242] like Fig. 20 As shown, the molding condition determination support device 504 includes a resin state estimation device 503, a quality estimation unit 701, a quality transition storage unit 702, a trend evaluation unit 703, a relationship storage unit 704, and a correction condition determination unit 705. The resin state estimation device 503 has the same structure as the resin state estimation device 503 described in the molding machine system 501A of the first example.

[0243] The quality inference unit 701 infers the quality of the molded product through machine learning based on the detection data acquired by the detection data acquisition unit 601. For example, the quality inference unit 701 infers the value of one or more quality categories in the molded product. The quality category of the molded product is at least one of the mass of the molded product, the size of the molded product, and the void volume in the molded product.

[0244] The quality estimation unit 701 generates a learned model that has learned the relationship between the inspection data and the quality of the molded product in advance through machine learning. The learned model is generated for each quality type. Furthermore, the quality estimation unit 701 stores the learned model and uses the newly acquired inspection data and the learned model to estimate the quality of the molded product.

[0245] The quality transition storage unit 702 accumulates the quality of the molded products estimated by the quality estimation unit 701, and stores the quality transition of the accumulated plurality of molded products. The quality transition is information that arranges the qualities of the plurality of molded products in the order of molding. The quality transition is, for example, Fig.21 Data as shown. Fig.21 This is an example of the quality of the molded product. Fig.21 In the quality standard, Std is set, the upper limit of the quality allowable range is set as Thmax, and the lower limit of the quality allowable range is set as Thmin. That is, the upper limit Thmax and the lower limit Thmin mean that the product is qualified, and the deviation range means that the product is unqualified. Even if the product is qualified, it is ideally in the case of the prescribed quality standard Std.

[0246] exist Fig.21 In the initial stage of molding, the quality of most molded products shows a value close to the quality standard Std. However, as a sudden abnormality, there are molded products that show a value exceeding the upper limit value Thmax. ( Fig.21 That is, when the molded products with sudden abnormalities are excluded, the quality of the molded products at the beginning of molding shows a value close to the quality standard Std ( Fig.21 After that, if molding continues, the quality of the molded product gradually deviates from the quality standard Std. Furthermore, the quality of the molded product shows a value of +N% higher than the quality standard Std ( Fig.21 D3).

[0247] The trend evaluation unit 703 evaluates the quality change trend relative to the quality standard Std based on the quality transition stored in the quality transition storage unit 702. Fig. 22 As shown, the trend evaluation unit 703 evaluates the degree of deviation of the quality from the quality standard Std in each of the plurality of quality categories and the degree of dispersion of the quality of the plurality of molded products in each quality category as the quality change trend.

[0248] The trend evaluation unit 703 pre-sets the number of molded products with quality transitions for evaluating the quality change trend. That is, the trend evaluation unit 703 calculates the degree of deviation of the quality of the preset number of molded products. For example, the trend evaluation unit 703 calculates the degree (absolute value or relative value) of the deviation of the average quality of the number of molded products from the quality standard Std as the degree of deviation of the quality. Furthermore, the degree of deviation refers to the degree of quality deviation or stability of the preset number of molded products. In addition, the degree of deviation may also be expressed by values ​​such as standard deviation and variance.

[0249] like Fig. 22 As shown, for example, the trend evaluation unit 703 evaluates the degree of deviation of quality as "+2.2%", which is a "stable" degree of deviation; evaluates the degree of deviation of size as "+0.3%", which is a "stable" degree of deviation; and evaluates the degree of deviation of void volume as "-0.5%", which is a "stable" degree of deviation.

[0250] In addition, Fig.21 In the quality transition D1 shown in FIG. 1 , the trend evaluation unit 703 evaluates the quality change trend as being stable and located near the quality standard Std. Fig.21 In the quality transition D1 shown, the molded product D2, which indicates a quality change due to an unexpected abnormality, is excluded for evaluation. In this case, the trend evaluation unit 703 evaluates the degree of deviation from the quality standard Std as "0.1%" for the target quality type and evaluates the degree of deviation as "stable". Since the quality of the molded product returns to normal after the unexpected abnormality, the unexpected abnormality is not related to the molding conditions. Therefore, the evaluation is performed excluding the unexpected abnormality.

[0251] In addition, Fig.21 In the quality transition D3 shown, the trend evaluation unit 703 evaluates the quality change trend as stable and shows a value that deviates from the quality standard Std by about +N%. In this case, the trend evaluation unit 703 evaluates the degree of deviation from the quality standard Std as "+N%" for the quality type of the object, and evaluates the degree of deviation as "stable".

[0252] The group acquisition unit 607 in the resin state estimation device 503 acquires the group of the molten state of the resin in the cavity C2 as described above. Here, in the group acquisition unit 607, the group of the molten state of the resin is, for example, Fig.23 As shown, it is classified into four types, namely, Type-A, Type-B, Type-C, and Type-D.

[0253] The relationship storage unit 704 stores the relationship between the quality change trend and the correction amount of the molding condition for returning the quality to the quality standard and the molten state of the resin in the cavity C2 in a corresponding relationship. Fig.24 As shown, the relationship between the quality change trend and the correction amount of the molding conditions is stored according to each group of the molten state of the resin. In detail, the relationship storage unit 704 stores a matrix showing the relationship between the level of deviation and the correction amount of the molding conditions according to each quality type in each group of the molten state of the resin. For example, six levels are set as the level of deviation for each of the quality, size, and void volume. In addition, the type of molding condition to be corrected is at least one of the injection speed, holding pressure, holding time, mold temperature during holding, cooling time, etc.

[0254] Here, the degree of relationship between the quality type and the type of molding condition, and the relationship between the degree of deviation of the quality type and the correction amount of the molding condition can be derived using machine learning. That is, the relationship stored in the relationship storage unit 704 can be generated by machine learning. Of course, the relationship can also be set based on experiments, past experience, etc. instead of machine learning.

[0255] The correction condition determination unit 705 determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit 703 , the group of molten states of the resin acquired by the group acquisition unit 607 , and the relationship stored in the relationship storage unit 704 .

[0256] The correction condition determination unit 705 first determines Fig.24 The correction condition determination unit 705 selects the group corresponding to the group of the molten state of the resin acquired by the group acquisition unit 607 from among the relationships shown in FIG. Fig.24 In the relationship expressed by the matrix shown in FIG. 1 , the level closest to the degree of deviation evaluated by the trend evaluation unit 703 is determined, and the correction amount of the molding condition corresponding to the level is set as the correction amount of the determined molding condition. Fig. 22 As shown, when the quality deviation is "+2.2%", Fig.24 In the matrix shown, the level of deviation of quality "+2%" is selected.

[0257] Here, the correction condition determination unit 705 determines the correction amount of the molding condition for each of the plurality of quality types. Therefore, the correction condition determination unit 705 determines the final correction amount of the molding condition based on the plurality of correction amounts for the same type of molding condition. In this case, the correction condition determination unit 705 may, for example, use the value obtained by summing the plurality of correction amounts as the final correction amount, or may use the value obtained by further summing the values ​​multiplied by the weighting coefficients corresponding to the quality types as the final correction amount.

[0258] Furthermore, the correction condition determination unit 705 outputs the final correction amount to the control device 560 of the molding machine 502, and adds the correction amount to the molding conditions in the next molded product. Therefore, the control device 560 performs molding of the next molded product according to the corrected molding conditions. As a result, the quality of the molded product molded according to the corrected molding conditions can be made close to the quality standard Std.

[0259] (9. Determining the effect of the auxiliary device 504 by molding conditions)

[0260] The effect of using the above-mentioned molding condition determination auxiliary device 504 is described. The quality transition storage unit 702 accumulates the quality of the molded product inferred by machine learning and stores the quality transition. The quality transition refers to information that arranges the qualities of multiple molded products in the order of molding. Therefore, the trend evaluation unit 703 can evaluate the quality change trend based on the quality transition of multiple continuous molded products.

[0261] In particular, the trend evaluation unit 703 evaluates the quality change trend relative to the predetermined quality standard Std. For example, the trend evaluation unit 703 can evaluate the state where the quality continues to deviate from the predetermined quality standard Std, the quality changes within the quality allowable range including the predetermined quality standard, etc. as the quality change trend.

[0262] The relationship between the quality change trend and the correction amount of the molding condition is stored in advance in correspondence with the group of the melting state of the resin in the cavity C2 in the relationship storage unit 704. That is, the relationship between the quality change trend and the correction amount of the molding condition is stored for each group of the melting state of the resin in the relationship storage unit 704. This relationship is set based on the experience of an expert, the output result of machine learning, the experimental result, etc.

[0263] The correction condition determination unit 705 determines the correction amount of the molding condition based on the newly evaluated quality change trend, the newly acquired group of the melting state of the resin in the cavity C2, and the relationship stored in the relationship storage unit 704. Here, the relationship stored in the relationship storage unit 704 is related to the correction amount of the molding condition for returning the quality to the specified quality standard Std. In particular, the correction amount of the molding condition is determined based on the group of the melting state of the resin in the cavity C2. Therefore, when the molding condition of the molding machine 502 is corrected according to the correction amount of the molding condition determined by the correction condition determination unit 705, the quality of the molded product to be molded next can be brought close to the specified quality standard Std.

[0264] That is, even if the quality of the molded product changes due to external factors such as ambient temperature or slight differences in the components contained in the raw materials of the molding material, the molding conditions can be corrected by understanding the quality change trend and further grouping the molten state of the resin in the cavity C2 so that the quality of the molded product can be made to meet the prescribed quality standard Std. Therefore, not only skilled people, but even unskilled people can correct the molding conditions to make the quality of the molded product good.

[0265] (10. How to set the number of groups)

[0266] The learned model generation unit 605 performs setting processing of the number of groups of the molten state of the resin in cooperation with other structures based on the generation of the learned model. Here, the number of groups can also be set to any number by the operator, but it can be set to an appropriate number by using the above-mentioned molding condition determination auxiliary device 504.

[0267] The learned model generation unit 605 acquires a group (either G1 or G2 ) of the resin melting state for the acquired detection data while setting the number of groups of the resin melting state to the initial setting number (eg, 2) (group acquisition step).

[0268] Furthermore, the relationship between the quality change trend and the correction amount of the molding condition when the number of groups is set to the initial setting number is set in correspondence with the group of the molten state of the resin (relationship setting step). Furthermore, the relationship is stored in the relationship storage unit 704. Next, the correction condition determination unit 705 determines the correction amount of the molding condition (correction amount of the control parameter value) based on the acquired group (either G1 or G2) and the relationship stored in the relationship storage unit 704 (correction amount determination step).

[0269] Next, the learned model generation unit 605 determines whether the quality of the molded product satisfies a predetermined range when the control parameter value is corrected based on the correction amount of the control parameter value corresponding to the acquired groups G1 and G2 of the molten state of the resin (determination step).

[0270] When the quality of the molded product does not meet the prescribed range, the number of groups is increased to repeatedly perform the above-mentioned group acquisition process, relationship setting process, correction amount determination process, and judgment process. That is, when the quality of the molded product does not meet the prescribed range, the learned model generation unit 605 increases the number of groups to repeatedly judge whether the acquisition of the group of the molten state of the resin and the quality of the molded product meet the prescribed range, and sets the number of groups when the quality of the molded product meets the prescribed range as the number of groups in the learned model.

[0271] By setting the number of groups in this way, the correction amount of the molding conditions can be determined according to the group of the molten state of the resin. In other words, the correction amount of the molding conditions can be determined appropriately.

[0272] (11. Configuration Example of Molding Machine System 501B)

[0273] (11-1. First example)

[0274] Reference Fig.25 The structure of the first example of the molding machine system 501B is described. Fig.25 As shown, the molding machine system 501B includes a plurality of molding machines 502, 502, edge computers 505, 505 integrally formed with the molding machines 502, 502, respectively, and a server 506 forming the same network with the plurality of molding machines 502, 502. The edge computers 505, 505 may constitute a part of the molding machines 502, 502, or may be constituted separately from the molding machines 502, 502.

[0275] Furthermore, the edge computers 505, 505 and the server 506 constitute a molding condition determination support device 504. The edge computers 505, 505 include a detection data acquisition unit 601. The server 506 includes the components of the resin state estimation device 503 except for the detection data acquisition unit 601, a quality estimation unit 701, a quality transition storage unit 702, a trend evaluation unit 703, a relationship storage unit 704, and a correction condition determination unit 705.

[0276] That is, the server 506 receives the detection data acquired by the detection data acquisition unit 601 from the edge computer 505 that is separate from the molding machine 502 or the edge computer 505 built into the molding machine 502. The server 506 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 502.

[0277] In this case, the server 506 can accumulate information related to the plurality of molding machines 502. Furthermore, by providing the server 506 with a processor capable of high-speed processing, the processing of the feature quantity generating unit 602, the processing of the identification parameter value calculating unit 604, the processing of the group acquiring unit 607, the processing of the quality estimating unit 701, the processing of the trend evaluating unit 703, and the processing of the correction condition determining unit 705 can be high-speed processing. On the other hand, since it is not necessary to make each edge computer 505 high-specification, it is possible to achieve cost reduction.

[0278] (11-2. Second example)

[0279] The second example of the molding machine system 501B is similar to the first example. The molding machine system 501B includes a plurality of molding machines 502, 502, edge computers 505, 505 respectively connected to the molding machines 502, 502, and a server 506 constituting the same network as the plurality of molding machines 502, 502.

[0280] The edge computers 505 and 505 include all the components of the resin state estimation device 503 and the quality estimation unit 701. The server 506 includes a quality transition storage unit 702, a trend evaluation unit 703, a relationship storage unit 704, and a correction condition determination unit 705. That is, the server 506 receives the group of molten states of the resin acquired by the group acquisition unit 607 of the resin state estimation device 503 and the quality of the molded product estimated by the quality estimation unit 701 from the edge computer 505 that is separate from the molding machine 502 or the edge computer 505 built into the molding machine 502. The server 506 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 502.

[0281] (111-3. The third example)

[0282] The third example of the molding machine system 501B is that the server 506 has all the functions of the molding condition determination auxiliary device 504. In this case, the edge computers 505 and 505 are not required. The detection data acquisition unit 601 in the server 506 receives the detection data detected by the sensors 544 and 545 from the molding machine 502. In addition, the correction condition determination unit 705 in the server 506 sends the correction amount of the molding condition to the molding machine 502.

[0283] The molding condition determination assisting device will be described below as a third embodiment. The present disclosure is not limited to the third embodiment, and various design changes can be made within the scope of the present disclosure.

[0284] (1. Applicable objects)

[0285] The molding condition determination auxiliary device is applicable to a molding method in which a molded product is molded by supplying a molten material formed by melting a molding material to a mold cavity of a molding machine. The molding machine of the applicable object can be, for example, an injection molding machine that performs injection molding of a resin or rubber as a molding material. In addition, other molding machines of the applicable object can also be, for example, a blow molding machine and a compression molding machine. In addition, regarding the resin as the molding material, examples can be given of thermoplastic resins such as monomeric polyamides and reinforced resins obtained by adding fillers to the base material of the thermoplastic resin. As fillers, micrometer-sized or nanometer-sized fillers can be cited. As fillers, for example, glass fibers, carbon fibers, etc. can be cited.

[0286] (2. Molding machine system 1001)

[0287] Reference Fig.26 The molding machine system 1001 including the molding condition determination auxiliary device is described. Fig.26 As shown, the molding machine system 1001 includes a molding machine 1002 and a molding condition determination assisting device 1003 .

[0288] The molding machine 1002 is an injection molding machine, a blow molding machine, or a compression molding machine. In this example, the molding machine 1002 is an injection molding machine. The molding machine 1002 molds, for example, a resin molded product. The molding condition determination auxiliary device 1003 is a device for determining the molding conditions in the molding machine 1002. In particular, in this example, the molding condition determination auxiliary device 1003 determines a correction amount of the molding conditions for improving the quality of the molded product when the molded product is molded according to the applied molding conditions.

[0289] The molding condition determination auxiliary device 1003 may be a device separate from the molding machine 1002 or may be a device assembled to the molding machine 1002. In addition, the molding condition determination auxiliary device 1003 may be partially assembled to the molding machine 1002, and the remaining part may be separated from the molding machine 1002. In the case where all or part of the molding condition determination auxiliary device 1003 is separated from the molding machine 1002, the separated part may be configured to be connected to only one molding machine 1002 or may be configured to be connected to a plurality of molding machines 1002. In the latter case, the separated part of the molding condition determination auxiliary device 1003 and the plurality of molding machines 1002 constitute the same network, and become a structure capable of communicating with each other.

[0290] (3. Molding machine 1002)

[0291] (3-1. Structure of Molding Machine 1002)

[0292] Reference Fig.26The structure of an injection molding machine will be described as an example of the molding machine 1002. The molding machine 1002 mainly includes a base 1020, an injection device 1030, a mold 1040, a mold clamping device 1050, and a control device 1060.

[0293] The injection molding device 1030 is arranged on the base 1020. The injection molding device 1030 is a device that melts the molding material and applies pressure to the molten material to supply the molten material to the cavity C3 of the mold 1040. The injection molding device 1030 mainly includes a hopper 1031, a heating cylinder 1032, a screw 1033, a nozzle 1034, a heater 1035, a driving device 1036, and a sensor 1037 for the injection molding device.

[0294] The hopper 1031 is an inlet for pellets (granular molding material) which are the raw materials of the molding material. The heating cylinder 1032 heats and melts the pellets put into the hopper 1031 and pressurizes the formed molten material. In addition, the heating cylinder 1032 is configured to be movable in the axial direction of the heating cylinder 1032 relative to the base 1020. The screw 1033 is arranged inside the heating cylinder 1032 and is configured to be rotatable and movable in the axial direction. The nozzle 1034 is an injection port provided at the front end of the heating cylinder 1032, and the molten material inside the heating cylinder 1032 is supplied to the mold 1040 by the axial movement of the screw 1033.

[0295] The heater 1035 is, for example, provided outside the heating cylinder 1032 to heat the pellets inside the heating cylinder 1032. The driving device 1036 moves the heating cylinder 1032 in the axial direction, rotates the screw 1033 and moves the screw 1033 in the axial direction, etc. The injection molding device sensor 1037 is generally referred to as a sensor for obtaining the storage amount of the molten material, the holding pressure, the holding time, the injection speed, the state of the driving device 1036, etc. However, the injection molding device sensor 1037 is not limited to the above, and various information can also be obtained.

[0296] The mold 1040 is a metal mold including a first mold 1041 as a fixed side and a second mold 1042 as a movable side. The mold 1040 forms a cavity C3 between the first mold 1041 and the second mold 1042 by clamping the first mold 1041 and the second mold 1042. The first mold 1041 includes a supply path 1043 (sprue, runner, gate) for guiding the molten material supplied from the nozzle 1034 to the cavity C3. In addition, the mold 1040 includes a pressure sensor 1044 and a temperature sensor 1045. The pressure sensor 1044 detects the pressure received by the molten material in the supply path 1043. The temperature sensor 1045 directly detects the temperature of the molten material in the supply path 1043.

[0297] The mold clamping device 1050 is disposed on the base 1020 facing the injection device 1030. The mold clamping device 1050 opens and closes the attached mold 1040, and in a closed and secured state, prevents the mold 1040 from opening due to the pressure of the molten material injected into the cavity C3.

[0298] The mold clamping device 1050 includes a fixed plate 1051, a movable plate 1052, a tie rod 1053, a driving device 1054, and a sensor 1055 for the mold clamping device. The first mold 1041 is fixed to the fixed plate 1051. The fixed plate 1051 can abut against the nozzle 1034 of the injection device 1030 to guide the molten material injected from the nozzle 1034 to the mold 1040. The second mold 1042 is fixed to the movable plate 1052. The movable plate 1052 can approach and separate from the fixed plate 1051. The tie rod 1053 supports the movement of the movable plate 1052. The driving device 1054 is composed of, for example, a cylinder device, and moves the movable plate 1052. The sensor 1055 for the mold clamping device is generally referred to as a sensor for obtaining the mold clamping force, the metal mold temperature, the state of the driving device 1054, and the like.

[0299] The control device 1060 controls the drive device 1036 of the injection device 1030 and the drive device 1054 of the mold clamping device 1050. For example, the control device 1060 obtains various information from the injection device sensor 1037 and the mold clamping device sensor 1055 to control the drive device 1036 of the injection device 1030 and the drive device 1054 of the mold clamping device 1050 to perform actions corresponding to the action instruction data.

[0300] (3-2. Molding method)

[0301] The molding method of the molded product using the molding machine 1002 is described. In the molding method using the molding machine 1002, the metering process, the mold closing process, the injection filling process, the pressure holding process, the cooling process, and the demolding process are performed in sequence in one cycle. That is, in the molding of the next molded product, the above processes are performed again in sequence. Here, the metering process and the mold closing process constitute the starting preparation process, the injection filling process, the pressure holding process, and the cooling process constitute the molding process, and the demolding process constitutes the ending process. In addition, the initial stage of the demolding process (soon after the mold is opened) can also be included in the molding process, and the later stage can be used as the ending process.

[0302] In the metering process, the pellets are melted by the heating of the heater 1035 and the shear friction heat accompanying the rotation of the screw 1033, and the molten material is accumulated between the tip of the screw 1033 and the nozzle 1034 in the heating cylinder 1032. Since the screw 1033 retreats as the accumulated amount of the molten material increases, the accumulated amount of the molten material is measured from the retreated position of the screw 1033.

[0303] In the mold clamping process after the metering process, the movable platen 1052 is moved to align the second mold 1042 with the first mold 1041, and the molds are clamped. In addition, the heating cylinder 1032 is moved in the axial direction to approach the mold clamping device 1050, and the nozzle 1034 is connected to the fixed platen 1051 of the mold clamping device 1050. Next, in the injection filling process, the screw 1033 is moved toward the nozzle 1034 by using a predetermined thrust while the rotation of the screw 1033 is stopped, so that the molten material is injection-filled into the mold 1040 at a relatively high pressure. When the molten material is filled into the cavity C3, the process is transferred to the pressure holding process.

[0304] In the pressure holding process, the molten material is further pressed into the cavity C3 while the cavity C3 is filled with the molten material, and a pressure holding process is performed in which a predetermined pressure (pressure holding pressure) is applied to the molten material in the cavity C3 for a predetermined time. Specifically, a predetermined pressure holding pressure is applied to the molten material by applying a certain thrust to the screw 1033.

[0305] After the predetermined holding pressure is applied for a predetermined time, the process proceeds to the cooling process. In the cooling process, the pushing of the molten material is stopped and the holding pressure is reduced to cool the mold 1040. By cooling the mold 1040, the molten material supplied to the mold 1040 is solidified. Finally, in the demolding and removing process, the second mold 1042 is separated from the first mold 1041, and the molded product is removed.

[0306] (3-3. Mold 1040)

[0307] Reference Fig. 27 as well as Fig.28 The detailed structure of the mold 1040 is described below. The mold 1040 is a so-called multiple metal mold, and multiple cavities C3 are formed in the mold 1040. However, in order to simplify the drawings, Fig. 27 as well as Fig.28 Only one cavity C3 is shown in the figure. In addition, in this example, the molded product molded by the molding machine 1002 is a retainer for a constant velocity universal joint. Therefore, the molded product is annular, and the cavity C3 is formed into an annular shape that imitates the shape of the retainer. In addition, the shape of the molded product and the cavity C3 can also be a shape other than an annular shape, such as a C shape, a rectangular frame shape, etc.

[0308] The supply path 1043 includes a nozzle 1043a, a runner 1043b, and a gate 1043c. The nozzle 1043a is a passage for supplying molten material from the nozzle 1034. The runner 1043b is a passage branching from the nozzle 1043a, and the molten material supplied to the nozzle 1043a flows into the runner 1043b. The gate 1043c is a passage for introducing the molten material flowing into the runner 1043b into the cavity C3, and the flow path cross-sectional area of ​​the gate 1043c is smaller than that of the runner 1043b. In the mold 1040, the same number of runners 1043b and gates 1043c as the number of cavities C3 are formed, and the molten material supplied to the nozzle 1043a is supplied to each cavity C3 via the runner 1043b and the gate 1043c.

[0309] Furthermore, when the cavity C3 is annular and the first mold 1041 has one gate 1043c, the inflow path of the molten material in the cavity C3 is a path that flows from the gate 1043c along the annular circumferential direction of the cavity C3. That is, in the cavity C3, the molten material first flows into the vicinity of the gate 1043c and finally flows into the farthest distance from the gate 1043c.

[0310] In addition, a pressure sensor 1044 is provided in the mold 1040, and the pressure sensor 1044 detects the pressure received by the molten material in the supply path 1043. In this example, a plurality of pressure sensors 1044 are provided. For example, the pressure sensor 1044 is provided near the position farthest from the gate 1043c in the cavity C3, near the gate 1043c. In addition, the pressure sensor 1044 may also be provided at the injection port 1043a and the runner 1043b. The pressure sensor 1044 may be a contact sensor or a non-contact sensor.

[0311] In addition, the mold 1040 is provided with a temperature sensor 1045 for detecting the temperature of the molten material in the supply path 1043. The temperature sensor 1045 may be provided in the cavity C3, or may be provided in the nozzle 1043a or the runner 1043b, similarly to the pressure sensor 1044. In addition, a plurality of pressure sensors 1044 may be provided.

[0312] (4. Molding conditions determine the structure of the auxiliary device 1003)

[0313] Reference Figure 29-Figure 34The structure of the molding condition determination auxiliary device 1003 is described. The molding condition determination auxiliary device 1003 includes, for example: an operation processing device including a processor, a storage device, an interface, etc.; an input device that can be connected to the interface of the operation processing device; and an output device that can be connected to the interface of the operation processing device. The output device may also include a display device, for example. In addition, the operation processing device, the input device, and the output device may also constitute a unit without an interface. In addition, a part of the operation processing device and a part of the storage device may also be applied to a physical server or a cloud server.

[0314] like Fig.29 As shown, the molding condition determination support device 1003 includes a detection data acquisition unit 1101, a quality estimation unit 1102, a quality transition storage unit 1103, a trend evaluation unit 1104, a molten state estimation unit 1105, a relationship storage unit 1106, and a correction condition determination unit 1107.

[0315] The detection data acquisition unit 1101 acquires detection data detected during molding by sensors 1044 and 1045 installed in the molding machine 1002. That is, the type of detection data acquired by the detection data acquisition unit 1101 is at least one of the pressure received by the mold 1040 from the molten material in the supply path 1043 and the temperature of the molten material in the supply path 1043.

[0316] The detection data based on the pressure sensor 1044 is, for example, Fig.30 The data shown in Fig.30 In the figure, time T1 is the start time of filling, time T2 is the end time of filling and the start time of pressure holding, time T3 is the end time of pressure holding and the start time of cooling, and time T4 is the end time of cooling and the mold opening time. That is, the period between T1 and T2 is the injection filling process, the period between T2 and T3 is the pressure holding process, and the period between T3 and T4 is the cooling process. In addition, Fig.30 In the embodiment, the maximum pressure in the pressure holding process is set as the maximum pressure holding pressure Pmax, the pressure integral value in the pressure holding process is set as the pressure holding area Sa, and the pressure integral value in the cooling process is set as the cooling area Sb.

[0317] The quality inference unit 1102 infers the quality of the molded product through machine learning based on the detection data acquired by the detection data acquisition unit 1101. For example, the quality inference unit 1102 infers a value in one or more quality categories in the molded product. The quality category of the molded product is at least one of the mass of the molded product, the size of the molded product, and the void volume in the molded product.

[0318] The quality estimation unit 1102 generates a learned model that has learned the relationship between the inspection data and the quality of the molded product in advance through machine learning. The learned model is generated for each quality type. Furthermore, the quality estimation unit 1102 stores the learned model and uses the newly acquired inspection data and the learned model to estimate the quality of the molded product.

[0319] The quality transition storage unit 1103 accumulates the quality of the molded products estimated by the quality estimation unit 1102, and stores the quality transition of the accumulated plurality of molded products. The quality transition is information that arranges the qualities of the plurality of molded products in the order of molding. The quality transition is, for example, Fig.31 Data as shown. Fig.31 This is an example of the quality of the molded product. Fig.31 In the quality standard, Std is set, the upper limit of the quality allowable range is set as Thmax, and the lower limit of the quality allowable range is set as Thmin. That is, the upper limit Thmax and the lower limit Thmin mean that the product is qualified, and the deviation range means that the product is unqualified. Even if the product is qualified, it is ideally in the case of the prescribed quality standard Std.

[0320] exist Fig.31 In the initial stage of molding, the quality of most molded products shows a value close to the quality standard Std. However, as a sudden abnormality, there are molded products ( Fig.31 That is, when the molded products with sudden abnormalities are excluded, the quality of the molded products at the beginning of molding shows a value close to the quality standard Std ( Fig.31 A1). After that, if molding is continued, the quality of the molded product gradually deviates from the quality standard Std. Furthermore, the quality of the molded product shows a value of +N% higher than the quality standard Std ( Fig.31 A3).

[0321] The trend evaluation unit 1104 evaluates the quality change trend relative to the quality standard Std based on the quality transition stored in the quality transition storage unit 1103. Fig.32 As shown, the trend evaluation unit 1104 evaluates the degree of deviation of the quality from the quality standard Std in each of the plurality of quality categories and the degree of dispersion of the quality of the plurality of molded products in each quality category as the quality change trend.

[0322] The trend evaluation unit 1104 pre-sets the number of molded products with quality transitions for evaluating the quality change trend. That is, the trend evaluation unit 1104 calculates the degree of deviation of the quality of the preset number of molded products. For example, the trend evaluation unit 1104 calculates the degree (absolute value or relative value) of the deviation of the average value of the quality of the number of molded products from the quality standard Std as the degree of deviation of the quality. Furthermore, the degree of deviation refers to the degree of quality deviation or stability of the preset number of molded products. In addition, the degree of deviation may also be expressed by values ​​such as standard deviation and variance.

[0323] like Fig.32 As shown, for example, the trend evaluation unit 1104 evaluates the degree of deviation of quality as "+2.2%", which is a "stable" degree of deviation; evaluates the degree of deviation of size as "+0.3%", which is a "stable" degree of deviation; and evaluates the degree of deviation of void volume as "-0.5%", which is a "stable" degree of deviation.

[0324] In addition, Fig.31 In the quality transition A1 shown in FIG. 1 , the trend evaluation unit 1104 evaluates the quality change trend as being stable and located near the quality standard Std. Fig.31 In the quality transition A1 shown, the molded product A2, which indicates a quality change due to an unexpected abnormality, is excluded for evaluation. In this case, the trend evaluation unit 1104 evaluates the degree of deviation from the quality standard Std as "0.1%" for the target quality type and evaluates the degree of deviation as "stable". Since the quality of the molded product returns to normal after the unexpected abnormality, the unexpected abnormality is not related to the molding conditions. Therefore, the evaluation is performed excluding the unexpected abnormality.

[0325] In addition, Fig.31 In the quality transition A3 shown, the trend evaluation unit 1104 evaluates the quality change trend as stable and shows a value that deviates from the quality standard Std by about +N%. In this case, the trend evaluation unit 1104 evaluates the degree of deviation from the quality standard Std as "+N%" for the quality type of the object, and evaluates the degree of deviation as "stable".

[0326] The molten state estimation unit 1105 estimates the molten state of the molten material in the cavity C3 based on the detection data acquired by the detection data acquisition unit 1101. In particular, the molten state estimation unit 1105 estimates the degree of fluidity of the molten material as the molten state of the molten material.

[0327] Here, the melting state depends on the components contained in the raw material of the molding material. The components contained in the raw material of the molding material that affect the melting state include the amount of water, the length of the reinforcing fiber, the ratio of the reinforcing fiber, the molecular weight of the main component, etc. In addition, the melting state in the cavity C3 affects the detection data during molding. Therefore, the melting state estimation unit can estimate the melting state by using the detection data depending on the melting state during actual molding. The melting state is, for example, Fig.33 As shown, they are classified into four types, namely, Type-A, Type-B, Type-C, and Type-D.

[0328] The relationship storage unit 1106 stores the relationship between the quality change trend and the correction amount of the molding condition for returning the quality to the quality standard and the melting state of the molten material in the cavity C3 in a corresponding relationship. Fig.34 As shown, the relationship between the quality change trend and the correction amount of the molding conditions is stored according to the type of the melting state of each molten material. In detail, the relationship storage unit 1106 stores a matrix showing the relationship between the level of deviation and the correction amount of the molding conditions according to each quality type in each type of the melting state of the molten material. For example, for each of the quality, size, and void volume, six stages are set as the level of deviation. In addition, the type of molding condition to be corrected is at least one of the injection speed, holding pressure, holding time, mold temperature during holding pressure, cooling time, etc.

[0329] Here, the degree of relationship between the quality type and the type of molding condition, and the relationship between the degree of deviation of the quality type and the correction amount of the molding condition can be derived using machine learning. That is, the relationship stored in the relationship storage unit 1106 can be generated by machine learning. Of course, the relationship can also be set based on experiments, past experience, etc. instead of machine learning.

[0330] The correction condition determination unit 1107 determines the correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit 1104 , the melting state of the molten material estimated by the melting state estimation unit 1105 , and the relationship stored in the relationship storage unit 1106 .

[0331] The correction condition determination unit 1107 firstly Fig.34 The correction condition determination unit 1107 selects the molten state corresponding to the molten state estimated by the molten state estimation unit 1105 from the relationship shown in FIG. Fig.34 In the relationship represented by the matrix shown in FIG. 1 , the level closest to the degree of deviation evaluated by the trend evaluation unit 1104 is determined, and the correction amount of the molding condition corresponding to the level is set as the correction amount of the determined molding condition. Fig.32As shown, when the quality deviation is "+2.2%", Fig.34 In the matrix shown, the level of deviation of quality "+2%" is selected.

[0332] Here, the correction condition determination unit 1107 determines the correction amount of the molding condition for each of the plurality of quality types. Therefore, the correction condition determination unit 1107 determines the final correction amount for the molding condition based on the plurality of correction amounts for the same type of molding condition. In this case, the correction condition determination unit 1107 may, for example, use the value obtained by summing the plurality of correction amounts as the final correction amount, or may use the value obtained by further summing the values ​​multiplied by the weighting coefficients corresponding to the quality types as the final correction amount.

[0333] Furthermore, the correction condition determination unit 1107 outputs the final correction amount to the control device 1060 of the molding machine 1002, and adds the correction amount to the molding conditions in the next molded product. Therefore, the control device 1060 performs molding of the next molded product according to the corrected molding conditions. As a result, the quality of the molded product molded according to the corrected molding conditions can be made close to the quality standard Std.

[0334] (5. Determine the effect of auxiliary device 1003 by molding conditions)

[0335] The effect of using the above-mentioned molding condition determination auxiliary device 1003 is described. The quality transition storage unit 1103 accumulates the quality of the molded product inferred by machine learning and stores the quality transition. The quality transition refers to information that arranges the qualities of multiple molded products in the order of molding. Therefore, the trend evaluation unit 1104 can evaluate the quality change trend based on the quality transition of multiple continuous molded products.

[0336] In particular, the trend evaluation unit 1104 evaluates the quality change trend relative to the predetermined quality standard Std. For example, the trend evaluation unit 1104 can evaluate the state where the quality continues to deviate from the predetermined quality standard Std, the quality changes within the quality allowable range including the predetermined quality standard, etc. as the quality change trend.

[0337] Furthermore, the molten state inference unit 1105 infers the molten state of the molten material in the cavity C3 based on the detection data. Here, the molten state depends on the components contained in the raw material of the molding material. For example, the deviation of the components contained in the raw material of the molding material includes the amount of water, the length of the reinforcing fiber, the proportion of the reinforcing fiber, the molecular weight of the main component, etc. Moreover, the molten state in the cavity C3 affects the detection data during molding. Therefore, the molten state inference unit 1105 can infer the molten state by using the detection data depending on the molten state during actual molding.

[0338] The relationship between the quality change trend and the correction amount of the molding condition is stored in advance in the relationship storage unit 1106 in correspondence with the melting state of the molten material in the cavity C3. That is, the relationship between the quality change trend and the correction amount of the molding condition is stored in the relationship storage unit 1106 according to the type of the melting state of each molten material. This relationship is set based on the experience of a skilled person, the output result of machine learning, the experimental result, etc.

[0339] The correction condition determination unit 1107 determines the correction amount of the molding condition based on the newly evaluated quality change trend, the newly inferred melting state of the molten material in the cavity C3, and the relationship stored in the relationship storage unit 1106. Here, the relationship stored in the relationship storage unit 1106 is related to the correction amount of the molding condition for returning the quality to the specified quality standard Std. In particular, the correction amount of the molding condition is determined according to the melting state of the molten material in the cavity C3. Therefore, when the molding condition of the molding machine 1002 is corrected according to the correction amount of the molding condition determined by the correction condition determination unit 1107, the quality of the molded product to be molded next can be brought close to the specified quality standard Std.

[0340] That is, even if the quality of the molded product changes due to external factors such as ambient temperature or slight differences in the components contained in the raw materials of the molding material, the molding conditions can be corrected by understanding the quality change trend and further understanding the melting state of the molten material in the cavity C3 so that the quality of the molded product can be made to meet the prescribed quality standard Std. Therefore, not only skilled people, but even unskilled people can correct the molding conditions to make the quality of the molded product good.

[0341] (6. Learning phase in quality inference)

[0342] As described above, the quality estimation unit 1102 estimates the quality through machine learning. The quality estimation unit 1102 stores a learned model. The learned model is generated in advance. The generation of the learned model, that is, an example of the learning stage, will be described.

[0343] First, detection data for a plurality of molded products is obtained. Here, feature quantities of the detection data are extracted based on the detection data. For example, the feature quantities are extracted based on the pressure data detected by the pressure sensor 1044, the maximum holding pressure Pmax in the holding pressure process, the pressure deviation in the holding pressure process, the holding pressure area Sa ( Fig.30 ), the actual holding time (the holding time obtained based on the pressure data), the speed of change of the pressure at the start of the holding process (the pressure differential value), etc. In addition, for the extraction of characteristic quantities, the cooling area Sb ( Fig.30As shown in the figure), the speed of pressure change during the cooling process (pressure differential value), etc.

[0344] Furthermore, for the extraction of the characteristic quantity, the maximum temperature in the pressure holding process, the temperature deviation in the pressure holding process, the temperature area (temperature×time) in the pressure holding process, the cooling area (temperature×time) in the cooling process, the temperature change rate (differential value) in the cooling process, etc. are used based on the temperature data detected by the temperature sensor 1045. In addition, since the mold 1040 is provided with a plurality of pressure sensors 1044 and temperature sensors 1045, the above information is used for the extraction of the characteristic quantity for each sensor.

[0345] Furthermore, the above-mentioned plurality of information is acquired for a plurality of molded products. And, for each of the above-mentioned information, the maximum value, minimum value, average value, variance, etc. in the plurality of molded products are calculated and set as feature quantities.

[0346] On the other hand, the quality of a plurality of molded products is measured using an external measuring instrument or the like. For example, as quality, mass, size, void volume, etc. are measured. Furthermore, machine learning is performed using the feature quantity as an explanatory variable and the quality as a target variable, thereby generating a learned model. The learned model becomes a model that can input the feature quantity of the detection data and output the quality.

[0347] Furthermore, in the above description, a learning model using feature quantities as explanatory variables is described. However, a learning model may be used that uses detection data itself as explanatory variables instead of feature quantities and outputs quality as a target variable.

[0348] (7. Melting state of molten material in cavity C3)

[0349] (7-1. Relationship between molten state and quality)

[0350] When the fluidity of the molten material in the cavity C3 is high, the specific volume of the molten material in the cavity C3 can be increased. Therefore, the melting state of the molten material affects the quality of the molded product.

[0351] The fluidity of the molten material is affected by, for example, the amount of moisture absorbed by the particles of the raw material of the molding material. Hydrolysis occurs in the molten material due to the moisture of the particles, and as a result, the fluidity of the molten material is improved. Hydrolysis tends to occur more easily as the amount of moisture in the molding material increases. In addition, it is believed that the length and proportion of the reinforcing fibers contained in the particles of the raw material of the molding material also affect the fluidity of the molten material. In addition, it is believed that the molecular weight of the main component of the molding material also affects the fluidity of the molten material.

[0352] (7-2. Relationship between melting state and test data)

[0353] The time for filling the cavity C3 with the molten material (hereinafter referred to as "cavity filling time") varies depending on the melting state of the molten material in the cavity C3. The cavity filling time refers to the time required from the start of filling the cavity C3 with the molten material to the completion of filling.

[0354] For example, the start time of the cavity filling time can be the timing when the detection data of the sensor at the position closest to the gate 1043c among the pressure sensors 1044 in the cavity C3 rises. In addition, the detection data of the pressure sensor 1044 provided in the runner 1043b can also be used as the start time of the cavity filling time. The completion time of the cavity filling time can be the timing when the detection data of the sensor at the position farthest from the gate 1043c among the pressure sensors 1044 in the cavity C3 rises.

[0355] Furthermore, the melting state estimation unit 1105 can estimate the melting state of the molten material in the cavity C3 using the cavity filling time. However, the relationship between the cavity filling time and the melting state can be acquired by using machine learning.

[0356] At this time, the relationship between the cavity filling time and the molten state can also be grasped with higher accuracy by using machine learning based on the inspection report of the raw materials of the molding material. That is, the molten state inference unit can refer to the inspection report of the raw materials of the molding material based on the detection data (especially the cavity filling time) and infer the molten state of the molten material through machine learning. In addition, the inspection report of the raw materials of the molding material contains the molecular weight of the main component, the amount of water, the type of reinforcing fiber, the length of the reinforcing fiber, the ratio of the reinforcing fiber, etc.

[0357] (7-3. Other methods of estimating the molten state)

[0358] As described above, the melting state can be estimated using the cavity filling time. In addition, the melting state can also be estimated by the following method.

[0359] Here, it is obvious that the values ​​of the detection data based on the sensors such as the pressure sensor 1044 and the temperature sensor 1045 during molding are affected by the molding conditions of the molding machine 1002. In addition, it is considered that the values ​​of the detection data are also affected by the melting state of the molten material in the cavity C3 in addition to the molding conditions.

[0360] That is, the value of the detection data is affected by the molding conditions and the melting state of the molten material. This relationship can be considered to be, for example, that the value related to the molding conditions and the value related to the melting state of the molten material are multiplied to obtain the value of the detection data. If substitution is performed, the value related to the melting state of the molten material is the result obtained by dividing the value related to the molding conditions by the value of the detection data.

[0361] Therefore, characteristic quantities for a plurality of detection data are obtained and multiplied. Furthermore, values ​​of a plurality of molding conditions are obtained and multiplied. Furthermore, the result obtained by dividing the multiplied value of the plurality of molding conditions by the multiplied value of the characteristic quantities of the plurality of detection data is used as an index representing the melting state of the molten material in the cavity C3. Furthermore, by classifying the obtained index, it is possible to determine which category the melting state belongs to. In this way, the melting state of the molten material in the cavity C3 can be inferred.

[0362] (8. Relationship-Determined Learning Stage)

[0363] The relationship between the degree of deviation of quality and the correction amount stored in the relationship storage unit 1106 can be generated by, for example, using machine learning. Hereinafter, a case where machine learning is used in the generation of this relationship will be described.

[0364] (8-1. First example)

[0365] Reference Fig.35 The first example of the learning phase is explained below. Fig.35 As shown, the relationship between the quality type and the type of molding condition is directly generated by machine learning.

[0366] For example, by inputting the quality value in each quality category and the value of the molding condition, the degree of influence (contribution degree, influence degree) of the molding condition on the quality category can be obtained through machine learning. In addition, the degree of influence of the correction amount of the molding condition value on the quality value can also be obtained. As a result, the relationship information between the quality category and the type of molding condition can be obtained, and further the relationship information between the quality value and the correction amount of the molding condition can be obtained.

[0367] Based on the relationship information obtained through machine learning, people decide to use Fig.34 Of course, it can also be generated through machine learning Fig.34 The matrix shown.

[0368] (8-2. Second example)

[0369] Reference Fig.36 The second example of the learning phase is explained below. In the second example of the learning phase, Fig.36As shown, the relationship between the quality type and the type of molding condition is indirectly generated by machine learning.

[0370] exist Fig.35 In the first example shown above, the relationship between the quality type and the type of molding condition is directly generated by machine learning. However, sometimes the relationship between the quality type and the type of molding condition cannot be directly obtained. Here, it is obvious that the molding conditions determine the state during molding, and the state during molding determines the quality. In other words, it can also be said that the molding conditions and quality are related by virtue of the state during molding. The state during molding refers to, for example, the maximum holding pressure Pmax, the holding area Sa, the cooling area Sb, etc.

[0371] Therefore, first, the quality value in each quality category and the characteristic quantity of the test data (the characteristic quantity obtained based on the maximum holding pressure, etc.) are input to obtain the degree of influence (contribution, influence) of the characteristic quantity of the test data on the quality category through machine learning. In addition, the degree of influence of the value of the characteristic quantity of the test data on the value of the quality can also be obtained. As a result, the relationship information between the quality category and the characteristic quantity of the test data can be obtained, and further the relationship information between the value of the quality and the value of the characteristic quantity of the test data can be obtained.

[0372] Next, the characteristic quantity of the detection data and the type of molding conditions are input to obtain the degree of influence (contribution degree, influence degree) of the type of molding conditions on the characteristic quantity of the detection data through machine learning. In addition, the degree of influence of the correction amount of the value of the molding condition on the value of the characteristic quantity of the detection data can also be obtained. As a result, the relationship information between the characteristic quantity of the detection data and the type of molding conditions can be obtained, and further the relationship information between the value of the characteristic quantity of the detection data and the correction amount of the value of the molding condition can be obtained.

[0373] In addition, the person determines the type of quality to be used based on the relationship information between the characteristic amount of the detection data and the relationship information between the characteristic amount of the detection data and the type of molding conditions. Fig.34 Of course, it can also be generated through machine learning Fig.34 The matrix shown.

[0374] (9. Motion timing of molding machine system 1001)

[0375] (9-1. Basics)

[0376] Reference Fig.37The operation timing of the molding machine system 1001, in particular, the operation timing of the processing performed by the molding machine 1002 and the processing performed by the molding condition determination auxiliary device will be described. First, the molding machine 1002 continuously molds the molded products. For ease of description, the second molded product is molded after the first molded product.

[0377] like Fig.37 As shown, the molding machine 1002 performs a start preparation process (S1011) for the first molded product. The start preparation process includes, for example, a metering process and a mold clamping process. Next, the molding machine 1002 performs a molding process (S1012) for the first molded product. The molding process includes, for example, an injection filling process, a pressure holding process, and a cooling process. Next, the molding machine 1002 performs a finishing process (S1013) for the first molded product. The finishing process includes, for example, a demolding and removing process.

[0378] However, in the molding condition determination auxiliary device 1003, when using detection data detected by the pressure sensor 1044 and the temperature sensor 1045 at the beginning of the demolding process (soon after the mold is opened), the beginning of the demolding process can also be included in the molding process.

[0379] After the first molded product is molded, the molding machine 1002 starts the process related to the molding of the second molded product. First, the molding machine 1002 performs the start preparation process for the second molded product (S1021). Next, the molding machine 1002 performs the molding process for the second molded product (S1022). Next, the molding machine 1002 performs the end processing process for the second molded product (S1023).

[0380] On the other hand, the molding condition determination auxiliary device 1003 is processed in parallel with the processing of the molding machine 1002. Specifically, the primary processing step (S1101) is processed in parallel with the detection of the data based on the pressure sensor 1044 and the temperature sensor 1045 when the first molded product is molded, and the secondary processing step (S1102) is processed in parallel with the end processing step of the first molded product and the start preparation step of the second molded product. That is, the secondary processing step (S1102) is executed in the preparation step of the molding machine 1002 from the end of the molding step of the first molded product to the start of the molding step of the second molded product (the end processing step of the first molded product and the start preparation step of the second molded product).

[0381] The primary processing step (S1101) includes at least processing performed by the detection data acquisition unit 1101. The secondary processing step (S1102) includes at least processing performed by the correction condition determination unit 1107. The correction condition determination unit 1107 determines the correction amount of the molding condition related to the second molded product. That is, the molding machine 1002 makes the molding condition in the molding process (S1022) of the second molded product a molding condition obtained by correcting it using the data in the molding process of the first molded product. In this way, the correction amount of the molding condition is determined within one cycle of molding of the molded product. Therefore, since it is possible to use the molding information not long ago in determining the correction amount of the molding condition, the correction amount of the molding condition can be made more accurately suitable for the current situation.

[0382] (9-2. First example)

[0383] A first example of the processing of the molding condition determination support device 1003 is as follows. The primary processing step (S1101) performs processing by the detection data acquisition unit 1101. On the other hand, the secondary processing step (S1102) performs processing by the quality estimation unit 1102, the trend evaluation unit 1104, and the correction condition determination unit 1107.

[0384] (9-3. Second example)

[0385] A second example of the processing of the molding condition determination support device 1003 is as follows. The primary processing step (S1101) performs processing by the detection data acquisition unit 1101 and processing by the quality estimation unit 1102. On the other hand, the secondary processing step (S1102) performs processing by the trend evaluation unit 1104 and processing by the correction condition determination unit 1107.

[0386] (10. Configuration Example of Molding Machine System 1001)

[0387] (10-1. First example)

[0388] Reference Fig.38 The structure of the first example of the molding machine system 1001 is described. Fig.38 As shown, the molding machine system 1001 includes a plurality of molding machines 1002, 1002, edge computers 1004, 1004 integrally formed with the molding machines 1002, 1002, respectively, and a server 1005 forming the same network with the plurality of molding machines 1002, 1002. In addition, the edge computers 1004, 1004 may constitute a part of the molding machines 1002, 1002, or may be constituted separately from the molding machines 1002, 1002.

[0389] Furthermore, the edge computers 1004, 1004 and the server 1005 constitute a molding condition determination support device 1003. The edge computers 1004, 1004 include a detection data acquisition unit 1101. The server 1005 includes a quality estimation unit 1102, a quality transition storage unit 1103, a trend evaluation unit 1104, a molten state estimation unit 1105, a relationship storage unit 1106, and a correction condition determination unit 1107.

[0390] That is, the server 1005 receives the detection data acquired by the detection data acquisition unit 1101 from the edge computer 1004 that is separate from the molding machine 1002 or the edge computer 1004 built into the molding machine 1002. The server 1005 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 1002.

[0391] In this case, the server 1005 can accumulate information related to a plurality of molding machines 1002. Furthermore, by providing the server 1005 with a processor capable of high-speed processing, the processing of the quality estimation unit 1102, the processing of the trend evaluation unit 1104, the processing of the molten state estimation unit 1105, and the processing of the correction condition determination unit 1107 can be high-speed processed. On the other hand, since each edge computer 1004 does not need to be of high specification, cost reduction can be achieved.

[0392] (10-2. Second example)

[0393] The second example of the molding machine system 1001 is similar to the first example. The molding machine system 1001 includes a plurality of molding machines 1002, 1002, edge computers 1004, 1004 respectively connected to the molding machines 1002, 1002, and a server 1005 constituting the same network as the plurality of molding machines 1002, 1002.

[0394] The edge computers 1004, 1004 include a detection data acquisition unit 1101, a quality estimation unit 1102, and a molten state estimation unit 1105. The server 1005 includes a quality transition storage unit 1103, a trend evaluation unit 1104, a relationship storage unit 1106, and a correction condition determination unit 1107. That is, the server 1005 receives the quality of the molded product estimated by the quality estimation unit 1102 and the molten state estimated by the molten state estimation unit 1105 from the edge computer 1004 that is separate from the molding machine 1002 or the edge computer 1004 built into the molding machine 1002. The server 1005 determines the correction amount of the molding condition based on the received information, and sends the determined correction amount of the molding condition to the molding machine 2.

[0395] (10-3. The third example)

[0396] The third example of the molding machine system 1001 is that the server 1005 has all the functions. In this case, the edge computers 1004 and 1004 are not required. The detection data acquisition unit 1101 in the server 1005 receives the detection data detected by the sensors 1044 and 1045 from the molding machine 1002. In addition, the correction condition determination unit 1107 in the server 1005 sends the correction amount of the molding condition to the molding machine 1002.

Claims

1. A molding condition determination auxiliary device, which is used in a molding method for molding a molded product by supplying a molten material obtained by melting a molding material into a cavity of a mold of a molding machine, and determines molding conditions of the molded product, in, have: a detection data acquisition unit, the detection data acquisition unit acquiring detection data detected by a sensor installed in the molding machine during molding; A quality inference unit, which infers the quality of the molded product through machine learning based on the detection data; a quality transition storage unit that accumulates the estimated qualities of the molded products and stores quality transitions regarding the accumulated qualities of the plurality of molded products; a trend evaluation unit that evaluates a quality change trend relative to a predetermined quality standard based on the quality transition; a relationship storage unit storing a relationship between the quality change trend and a correction amount of a molding condition for returning the quality to the quality reference; as well as a correction condition determination unit that determines a correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit and the relationship stored in the relationship storage unit, The trend evaluation unit evaluates the degree of deviation of a plurality of quality categories from the quality standard as the quality change trend, The relationship storage unit stores, for each of the quality categories, a matrix expressing the relationship between the level of the deviation and the correction amount of the molding condition.

2. The molding condition determination auxiliary device according to claim 1, in, further comprising a molten state estimating unit for estimating a molten state of the molten material in the cavity based on the detection data, The relationship storage unit stores the relationship between the quality change trend and the correction amount of the molding condition for returning the quality to the quality standard in correspondence with the molten state. The correction condition determination unit determines a correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit, the molten state estimated by the molten state estimation unit, and the relationship stored in the relationship storage unit.

3. The molding condition determination auxiliary device according to claim 2, in, The molten state estimation unit estimates the degree of fluidity of the molten material as the molten state of the molten material through the machine learning.

4. The molding condition determination auxiliary device according to claim 3, in, The molten state estimation unit estimates the degree of fluidity of the molten material as the molten state of the molten material through the machine learning based on a filling time required from the start to the completion of filling of the molten material in the cavity as the detection data.

5. The molding condition determination auxiliary device according to claim 2, in, The melt state estimation unit estimates the melt state of the molten material based on the detection data and the molding conditions.

6. The molding condition determination auxiliary device according to any one of claims 2 to 5, in, The molten state estimation unit estimates the molten state of the molten material by referring to an inspection report of a raw material of the molding material in addition to the detection data.

7. The molding condition determination auxiliary device according to claim 1, in, The trend evaluation unit evaluates the quality change trend by excluding a quality change due to a sudden abnormality.

8. The molding condition determination auxiliary device according to claim 1, in, The correction condition determination unit determines a level closest to the degree of deviation evaluated by the trend evaluation unit in the relationship expressed by the matrix, The correction amount of the molding condition corresponding to the level is determined as the correction amount of the molding condition.

9. The molding condition determination auxiliary device according to claim 1, in, The relationship storage unit stores a relationship determined based on relationship information between the quality type as the quality change trend inferred by the machine learning and the type of molding conditions, or The relationship determined based on the relationship information between the feature amount of the detection data inferred by machine learning and the quality type as the quality change trend and the relationship information between the feature amount of the detection data inferred by the machine learning and the type of the molding condition is stored.

10. The molding condition determination auxiliary device according to claim 1, in, The type of the detection data is at least one of the temperature of the molten material in the cavity and the pressure exerted on the mold by the molten material. The quality type of the molded product is at least one of the mass of the molded product, the size of the molded product, and the void volume in the molded product. The type of the molding condition is at least one of injection speed, holding pressure, holding time, mold temperature during holding, and cooling time.

11. The molding condition determination auxiliary device according to claim 1, in, The molding condition determining auxiliary device is suitable for the molding method for continuously molding the molded product, When the second molded product is molded after the first molded product, The detection data acquisition unit performs processing based on the detection data related to the first molded product, The quality estimation unit, the trend evaluation unit, and the correction condition determination unit are processed in parallel with the preparation process of the molding machine until the molding of the second molded product starts. The correction condition determination unit determines a correction amount of the molding condition related to the second molded product.

12. The molding condition determination auxiliary device according to claim 1, in, The molding condition determination auxiliary device further includes a server that forms a same network with a plurality of molding machines having the molding machine. The server at least includes the relationship storage unit and the correction condition determination unit. The server receives, from the molding machine or a computer integrally formed with the molding machine, one of the detection data detected by the sensor, the detection data acquired by the detection data acquisition unit, and the quality of the molded product estimated by the quality estimation unit, determining a correction amount of the molding condition based on the received one information, The determined correction amount of the molding condition is sent to the molding machine.

13. A resin state estimation device for estimating the melting state of a resin in a mold cavity of an injection molding machine, in, have: a detection data acquisition unit that acquires detection data detected by a sensor installed in the injection molding machine during molding; a feature quantity generating unit for generating a feature quantity group consisting of a plurality of feature quantities related to the detection data based on the detection data; a control parameter acquisition unit for acquiring a control parameter value group consisting of a plurality of control parameter values ​​used for control in the injection molding machine; an identification parameter value calculation unit that defines the melting state of the resin as represented by the feature value group and the control parameter value group, and calculates a resin state identification parameter value representing the melting state of the resin corresponding to each of the feature values ​​based on the feature value group and the control parameter value group; as well as The group acquisition unit acquires the groups of the molten states of the resin by applying a multivariate analysis using the resin state identification parameter value as an explanatory variable based on the resin state identification parameter value when the molten states of the resin are defined as being classified into a plurality of groups.

14. The resin state estimation device according to claim 13, in, The resin state estimation device further includes a learned model storage unit. The learned model storage unit stores a learned model, which is generated by setting the resin state identification parameter value as an explanatory variable, setting the group of the molten state of the resin as a target variable, applying cluster analysis as the multivariate analysis, and performing machine learning of the cluster analysis using a training data set including the explanatory variable and the target variable, The group acquisition unit acquires a group of melting states of the resin as an output of the learned model when the resin state identification parameter value is input.

15. The resin state estimation device according to claim 13 or 14, in, The identification parameter value calculation unit calculates the resin state identification parameter value according to equations (1) and (2). in, A(Fj)=ΠAk…(2), Wherein, a plurality of resin state identification parameter values ​​are set as a resin state identification parameter value group P, The characteristic quantity group of the detection data during molding is F, The control parameter value group is A, The characteristic quantity is Fj, The control parameter value is Ak.

16. The resin state estimation device according to claim 13, in, The identification parameter value calculation unit calculates the resin state identification parameter value corresponding to each feature quantity using the feature quantity and one or more control parameter values ​​having a high degree of influence on the feature quantity.

17. The resin state estimation device according to claim 16, in, One or more control parameter values ​​having a high degree of influence on the feature quantity are extracted by using the feature quantity of the object and the control parameter value group and utilizing machine learning.

18. The resin state estimation device according to claim 14, in, The resin state estimation device further includes a learned model generation unit, the learned model generation unit generating the learned model by performing machine learning of the cluster analysis using a training data set including the explanatory variables and the target variable, The learned model generating unit acquires the group of the molten state of the resin with respect to the acquired detection data in a state where the number of groups of the molten state of the resin is set to an initial setting number, When the control parameter value is corrected based on the correction amount of the control parameter value corresponding to the acquired set of the molten state of the resin, determining whether the quality of the molded product satisfies a predetermined range, When the quality of the molded product does not satisfy a predetermined range, the number of groups is increased to repeatedly determine whether the groups of the molten state of the resin and the quality of the molded product satisfy a predetermined range. The number of groups when the quality of the molded product satisfies a predetermined range is set as the number of groups in the learned model.

19. A molding condition determination auxiliary device, which is suitable for a molding method for molding a molded product by supplying a molten material formed by melting a resin into a cavity of a mold of an injection molding machine, and determines the molding conditions of the molded product. in, have: The resin state estimation device according to any one of claims 13 to 18; A quality inference unit, which infers the quality of the molded product through machine learning based on the detection data; a quality transition storage unit that accumulates the estimated qualities of the molded products and stores quality transitions of the accumulated plurality of molded products; a trend evaluation unit that evaluates a quality change trend relative to a predetermined quality standard based on the quality transition; a relationship storage unit that stores the relationship between the quality change trend and the correction amount of the molding condition for returning the quality to the quality reference in correspondence with the set of the melting state of the resin; as well as A correction condition determination unit determines a correction amount of the molding condition based on the quality change trend evaluated by the trend evaluation unit, the group of molten states of the resin acquired by the group acquisition unit, and the relationship stored in the relationship storage unit.

Images