JP2021191621A - Resin condition estimator and molding condition determination support device - Google Patents

Abstract

To provide a resin state estimation device that can estimate the melting state of resin in a cavity.SOLUTION: The resin state estimating device 3 includes a feature value generating unit 102 that generates a group of feature values [F] related to the detection data based on the detection data detected during molding by sensors 44, 45 attached to the injection molding machine 2, and a discriminating parameter value calculating unit 104 that calculates, based on the group of feature values [F] and the group of control parameter values [A], a discriminating parameter value P1-Pm corresponding to each feature value. The discriminative parameter value calculating unit 104 calculates the discriminative parameter values P1-Pm representing the melting state of the resin, and based on the discriminative parameter values P1-Pm, applies multivariate analysis using the discriminative parameter values P1-Pm as explanatory variables, and the section is equipped with a group acquisition section 107 that acquires groups G1 and G2 of the resin melt state.SELECTED DRAWING: Figure 3

Description

The present invention relates to a resin state estimation device and a molding condition determination support device.

In a method of molding a molded product by supplying a molten material in which a resin is melted to a cavity of a mold of an injection molding machine, when a defective product occurs, the operator needs to correct the molding conditions. Skilled skills are required to modify the molding conditions. It is not easy for an unskilled person to determine which molding conditions should be changed and how much.

Therefore, in recent years, research on artificial intelligence has progressed, and for example, Patent Documents 1 and 2 disclose that the amount of modification of molding conditions is determined by machine learning. In the technique described in Patent Document 1, by acquiring the relationship between the quality seed of the molded product and the seed of the molding condition by machine learning, when a defect occurs in the seed of a certain quality, which molding condition is used. Is output as to whether or not to correct. In the technique described in Patent Document 2, the amount of correction of molding conditions is determined by machine learning based on the detection data detected at the time of molding by the sensor attached to the molding machine.

Japanese Unexamined Patent Publication No. 2020-49443 Japanese Unexamined Patent Publication No. 2020-49929

By the way, in injection molding, it was found that the quality of the molded product differs depending on the molten state of the resin in the cavity. For example, the quality of the finished molded product differs depending on whether the resin in the cavity has high fluidity or low fluidity.

The molten state of the resin in the cavity is naturally affected by the control parameters used for control in the injection molding machine, but other factors such as the composition and function of the parts constituting the injection molding machine, the environmental temperature, etc. It is considered to be affected by. That is, if the molten state of the resin in the cavity can be grasped, it becomes possible to determine an appropriate correction amount of the molding conditions. However, it is not easy to grasp the molten state of the resin in the cavity.

One of the objects of the present invention is to provide a resin state estimation device capable of estimating the melted state of the resin in the cavity. Another object of the present invention is to provide a molding condition determination support device capable of modifying the molding conditions so that the quality of the molded product approaches the quality standard by using the resin state estimation device.

(1. Resin state estimation device)
The resin state estimation device is a resin state estimation device that estimates the molten state of the resin in the cavity of the mold of the injection molding machine, and is a detection that acquires detection data detected at the time of molding by a sensor attached to the injection molding machine. It is composed of a data acquisition unit, a feature quantity generation unit that generates a feature quantity group composed of multiple types of feature quantities related to the detection data based on the detection data, and a plurality of types of control parameter values used for control in the injection molding machine. The control parameter acquisition unit that acquires the control parameter value group to be performed, and the control parameter value group, which defines that the molten state of the resin is represented by the feature quantity group and the control parameter value group, respectively, are based on the feature quantity group and the control parameter value group, respectively. The identification parameter value calculation unit that calculates the resin state identification parameter value that represents the molten state of the resin corresponding to the feature amount of, and the resin state identification parameter value when it is defined that the molten state of the resin is classified into multiple groups. Based on the above, a group acquisition unit for acquiring a group of the molten state of the resin is provided by applying a multivariate analysis using the resin state identification parameter value as an explanatory variable.

It is considered that the detection data detected at the time of molding by the sensor attached to the injection molding machine is influenced by the control parameters of the injection molding machine and the molten state of the resin in the cavity. In other words, it is defined that the molten state of the resin is represented by the feature quantity group and the control parameter value group generated from the detection data.

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

Then, when the group acquisition unit defines that the molten state of the resin is classified into a plurality of groups, a multivariate analysis using the resin state identification parameter value as an explanatory variable is applied based on the resin state identification parameter value. By doing so, the group of the molten state of the resin is acquired. Here, the group of the molten state of the resin does not need to be clearly defined, but for example, the degree of fluidity can be classified as one of the elements.

That is, according to the resin state estimation device, by performing arithmetic processing using the detection data and the control parameter, the group of the molten state of the resin in the cavity at the time of molding of the molded product, for example, the fluidity of the resin. Degrees and the like can be classified as one of the elements. By grouping the molten state of the resin in the cavity in this way, it is possible to determine the amount of modification of the molding conditions according to the group.

(2. Molding condition determination support device)
The molding condition determination support device is applied to a molding method for molding a molded product by supplying a molten material in which a resin is melted to a cavity of a mold of an injection molding machine, and is a molding condition determination support device for determining the molding conditions of the molded product. The above-mentioned resin state estimation device, a quality estimation unit that estimates the quality of the molded product by machine learning based on the detection data, and a plurality of molded products that accumulate the estimated quality of the molded product. A quality transition storage unit that stores the quality transition of the product, a tendency evaluation unit that evaluates the quality change tendency with respect to a predetermined quality standard based on the quality transition, and a molding condition for returning the quality change tendency and quality to the quality standard. The relationship storage unit that stores the relationship with the correction amount in association with the group of the molten state of the resin, the quality change tendency evaluated by the tendency evaluation unit, and the group of the molten state of the resin acquired by the group acquisition unit. , A modification condition determination unit for determining a modification amount of the molding condition based on the relationship stored in the relationship storage unit is provided.

That is, the correction amount of the molding condition is determined by using the group of the molten state of the resin acquired by the above-mentioned resin state estimation device. This makes it possible to easily determine the appropriate amount of modification of the molding conditions.

It is a figure which shows the whole structure of the molding machine system of 1st example. It is a figure which shows the relationship of a control parameter, a melting state of a resin, and detection data. It is a schematic representation of the relational expression of the resin state identification parameter values. It is a graph which shows the detection data. It is a schematic diagram which shows the group of the molten state of a resin. It is a figure which shows the functional block composition of the resin state estimation apparatus. It is a figure which shows the whole structure of the molding machine system of the 2nd example. It is a figure which shows the functional block composition of the molding condition determination support apparatus. It is a graph which shows the quality transition. It is a figure which shows the quality change tendency. It is a figure which shows the group of the molten state of a resin. It is a figure which shows the relationship between the degree of deviation of quality, and the correction amount. It is a figure which shows the structural example of a molding machine system.

(1. Applicable target)
The resin state estimation device and the molding condition determination support device are applied to a molding method for molding a molded product by supplying a molten material obtained by melting a molding material (resin) to a cavity of a mold of an injection molding machine. As the resin as a molding material, a thermoplastic resin such as a single polyamide or a reinforced resin in which a filler is added to a base material of the thermoplastic resin can be exemplified. Fillers can include micron-sized or nano-sized fillers. Examples of the filler include glass fiber and carbon fiber.

(2. Molding machine system 1A of the first example)
The molding machine system 1A of the first example including the resin state estimation device 3 will be described with reference to FIG. As shown in FIG. 1, the molding machine system 1A includes an injection molding machine 2 (hereinafter referred to as a “molding machine”) and a resin state estimation device 3.

The molding machine 2 molds a molded product of resin. The resin state estimation device 3 estimates the molten state of the resin in the cavity C of the mold 40 of the molding machine 2. The estimated molten state of the resin is used, for example, to determine the molding conditions in the molding machine 2.

The resin state estimation device 3 may be a device separate from the molding machine 2, or may be a built-in device of the molding machine 2. Further, the resin state estimation device 3 may be partially incorporated in the molding machine 2 and the rest may be separated from the molding machine 2. When all or part of the resin state estimation device 3 is separate from the molding machine 2, the separate portion may be connected to only one molding machine 2 or may be connected to a plurality of molding machines 2. It may be configured to be connected. In the latter case, the separate portion of the resin state estimation device 3 and the plurality of molding machines 2 form the same network and can communicate with each other.

(3. Molding machine 2)
(3-1. Configuration of molding machine 2)
The configuration of the molding machine 2 will be described with reference to FIG. The molding machine 2 mainly includes a bed 20, an injection device 30, a mold 40, a mold clamping device 50, and a control device 60.

The injection device 30 is arranged on the bed 20. The injection device 30 is a device that melts the molding material (resin) and applies pressure to the molten material to supply the molten material to the cavity 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 drive device 36, and a sensor 37 for the injection device.

The hopper 31 is an input port for pellets (granular molding material) which is a material for molding material. The heating cylinder 32 pressurizes the molten material formed by heating and melting the pellets charged into the hopper 31. Further, the heating cylinder 32 is provided so as to be movable in the axial direction of the heating cylinder 32 with respect to the bed 20. The screw 33 is arranged inside the heating cylinder 32 and is provided so as to be rotatable and axially movable. The nozzle 34 is an injection port provided at the tip 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.

The heater 35 is provided, for example, on the outside of the heating cylinder 32 and heats the pellets inside the heating cylinder 32. The drive device 36 moves the heating cylinder 32 in the axial direction, rotates the screw 33, moves in the axial direction, and the like. The injection device sensor 37 is a general term for sensors that acquire the stored amount of the molten material, the holding pressure, the holding time, the injection speed, the state of the driving device 36, and the like. However, the sensor 37 for the injection device is not limited to the above, and various information may be acquired.

The mold 40 is a mold including a first mold 41 on the fixed side and a second mold 42 on the movable side. The mold 40 forms a cavity C between the first mold 41 and the second mold 42 by molding the first mold 41 and the second mold 42. The first type 41 includes a supply path 43 (sprue, runner, gate) that guides the molten material supplied from the nozzle 34 to the cavity C. Further, the mold 40 includes a pressure sensor 44 and a temperature sensor 45. The pressure sensor 44 detects the pressure received from 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.

The mold clamping device 50 is arranged on the bed 20 so as to face the injection device 30. The mold clamping device 50 opens and closes the mounted mold 40, and prevents the mold 40 from opening due to the pressure of the molten material injected into the cavity C in the state where the mold 40 is tightened.

The mold clamping device 50 includes a fixed plate 51, a movable plate 52, a diver 53, a drive device 54, and a sensor 55 for the mold clamping device. The first mold 41 is fixed to the fixing plate 51. The fixing plate 51 can come into contact with the nozzle 34 of the injection device 30 and guide the molten material injected from the nozzle 34 to the mold 40. The second type 42 is fixed to the movable board 52. The movable platen 52 can approach and separate from the fixed platen 51. The diver 53 supports the movement of the movable platen 52. The drive device 54 is composed of, for example, a cylinder device, and moves the movable platen 52. The mold clamping device sensor 55 is a general term for sensors that acquire a mold clamping force, a mold temperature, a state of a driving device 54, and the like.

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 acquires various information from the injection device sensor 37 and the mold clamping device sensor 55, and performs an operation according to the operation command data. It controls the drive device 54 of 50.

(3-2. Molding method)
A method of molding a molded product by the molding machine 2 will be described. In the molding method by the molding machine 2, the weighing step, the mold clamping step, the injection filling step, the pressure holding step, the cooling step, and the mold taking out step are sequentially executed in one cycle. That is, in the molding of the next molded product, the above steps are sequentially executed again. Here, the weighing step and the mold clamping step constitute a start preparation step, the injection filling step, the pressure holding step, and the cooling step constitute a molding step, and the mold removal step constitutes a finishing process step. .. The initial stage of the mold release taking-out process (immediately after the mold is opened) may be included in the molding process, and the latter stage may be the end processing process.

In the weighing step, the molten material is stored between the tip of the heating cylinder 32 and the nozzle 34 while the pellets are melted by the heating of the heater 35 and the shear frictional heat accompanying the rotation of the screw 33. Since the screw 33 retracts as the stored amount of the molten material increases, the stored amount of the molten material is measured from the retracted position of the screw 33.

In the mold clamping step following the weighing step, the movable plate 52 is moved, the second mold 42 is aligned with the first mold 41, and the mold is clamped. Further, 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 fixing plate 51 of the mold clamping device 50. Subsequently, in the injection filling step, the molten material is injection-filled into the mold 40 at a high pressure by moving the screw 33 toward the nozzle 34 with a predetermined pushing force while the rotation of the screw 33 is stopped. When the cavity C is filled with the molten material, the process continues to the pressure holding step.

In the pressure holding step, the molten material is further pushed into the cavity C in a state where the cavity C is filled with the molten material, and a pressure holding process is performed in which a predetermined pressure (holding pressure) is applied to the molten material in the cavity C for a predetermined time. Specifically, by applying a constant pushing force to the screw 33, a predetermined holding pressure is applied to the molten material.

Then, after performing the pressure holding treatment for a predetermined time with a predetermined holding pressure, the process proceeds to the cooling step. In the cooling step, the pressing of the molten material is stopped to reduce the holding pressure, and the mold 40 is cooled. By cooling the mold 40, the molten material supplied to the mold 40 is solidified. Finally, in the mold release taking-out step, the molded product is taken out by separating the second mold 42 from the first mold 41.

(4. Basics for estimating the molten state of resin)
The basics of estimating the molten state of the resin in the cavity C will be described with reference to FIGS. 2 to 5. In this example, it is defined that the melted state of the resin is classified into a plurality of groups, and the group of the melted state of the resin is estimated as the estimation of the melted state of the resin.

FIG. 2 shows that the control parameters and the molten state of the resin affect the detection data at the time of molding. Here, the molten state of the resin depends on the components contained in the material of the molding material. For example, the variation of the contained components in the material of the molding material includes the water content, the length of the reinforcing fibers, the ratio of the reinforcing fibers, the molecular weight of the main component, and the like. Then, as described above, the molten state of the resin in the cavity C affects the detection data at the time of molding.

Specifically, the detection data detected at the time of molding by the sensors 44 and 45 attached to the molding machine 2 is affected by the control parameters used for control in the molding machine 2 and the molten state of the resin in the cavity C. think. In other words, the molten state of the resin is affected by the detection data and control parameters.

FIG. 3 shows the above relationship expressed as a relational expression having a value. Here, the information as a value related to the detection data at the time of molding is a feature amount group [F] composed of a plurality of types of feature amounts related to the detection data. The feature amount is a statistic (maximum value, minimum value, average value, dispersion, maximum value of differentiation, differentiation) in the detection data of the pressure sensor 44 in each molding process (injection filling process, pressure holding process, cooling process, etc.). (Minimum value, integrated value, etc.), the above statistics in the detection data of the temperature sensor 45 in each molding process. When a plurality of pressure sensors 44 and temperature sensors 45 are provided, the statistics of the respective sensors 44 and 45 are used as feature quantities. In this way, a large number of feature quantities can be obtained, and these plurality of feature quantities are collectively referred to as a feature quantity group [F].

For example, a part of the feature amount will be described using the detection data of the pressure sensor 44 shown in FIG. In the detection data of the pressure sensor 44, as shown in FIG. 4, the time T1 is the filling start time, the time T2 is the filling end time and the holding pressure start time, and the time T3 is the holding pressure end time and the cooling disclosure time. , Time T4 is the cooling end time and the mold opening time. That is, the section between T1 and T2 is the injection filling step, the section between T2-T3 is the pressure holding step, and the section between T3-T4 is the cooling step. In FIG. 4, the maximum pressure value in the pressure holding step is Pmax, the holding pressure area which is the pressure integral value in the pressure holding step is Sa, and the cooling area which is the pressure integral value in the cooling step is Sb. The maximum value Pmax, the holding pressure area Sa, and the cooling area Sb are a part of the feature amount.

Further, the information as the value regarding the control parameter is the control parameter value itself of a plurality of types used for control in the molding machine 2, and is a control parameter value group [A] composed of the plurality of types of control parameter values. The unit of each control parameter value may be adjusted so that the number of digits of the control parameter value becomes a set predetermined value. The types of control parameter values are, for example, injection speed, holding pressure, holding pressure time, mold temperature at the time of holding pressure, cooling time, nozzle temperature, injection pressure (nozzle pressure) and the like.

Further, the information as the value regarding the melted state of the resin is a resin state identification parameter value representing the melted state of the resin. Here, the resin state identification parameter value does not mean that the value itself has a meaning, but is an index used for grouping the molten state of the resin. The resin state identification parameter value is a value corresponding to each feature amount. That is, there are as many resin state identification parameter values as there are types of feature quantities. Therefore, a plurality of types of resin state identification parameter values are referred to as a resin state identification parameter value group [P].

Then, as shown in FIG. 3, the resin state identification parameter value group [P] is represented by the feature amount group [F] of the detection data at the time of molding and the control parameter value group [A]. Specifically, the resin state identification parameter value group [P] is defined as a value obtained by dividing the feature amount group [F] of the detection data at the time of molding by the control parameter value group [A].

Here, what is expressed as a mathematical formula is the formula (1). As shown in the proviso of the formula (1), the resin state identification parameter value group [P], the feature amount group [P], and the control parameter value group [A] are the resin state identification parameter value P1-Pm and the feature amount, respectively. It is a matrix represented by F1-Fm and control parameter values A (F1) -A (Fm).

The control parameter values A (F1) -A (Fm) are represented by the equation (2). The control parameter value A (Fj) is an infinite product value of a plurality of control parameter values Ak. As the control parameter value A (Fj), any one of the two types described below is applied.

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

The second control parameter value A (Fj) is an infinite product value of some control parameter values Ak. For example, A (F1) and A (F2) are shown in the equations (3) and (4).

In the second control parameter value A (Fj), some control parameter values Ak are one or a plurality of control parameter values having a high degree of influence on the feature amount Fj.

Subsequently, when a resin state identification parameter value group for each detection data is obtained using a large number of detection data, a multivariate analysis is performed using the resin state identification parameter value group to obtain a group of resin melt states. The division is done. That is, multivariate analysis is performed using each of a plurality of types of resin state identification parameter values as explanatory variables.

In order to facilitate the explanation, it is assumed that there are two types of resin state identification parameter values, P1 and P2. In this case, as shown in FIG. 5, it is represented by a two-dimensional coordinate system having P1 and P2 as explanatory variables. The points P1 and P2 for each detected data are plotted on this two-dimensional coordinate system.

FIG. 5 is a plot of the resin state identification parameter values obtained by using the detection data acquired in the learning phase and the control parameter values, and the group of the molten state of the resin is, for example, two groups G1. It is divided into G2. In this case, the group of the molten state of the resin obtained by using the newly acquired detection data and the control parameter value is classified into any of G1 and G2.

In particular, it is advisable to use the resin state identification parameter value as an explanatory variable, the group of the molten state of the resin as the objective variable, and apply the cluster analysis as a multivariate analysis. In this case, a trained model can be generated by performing machine learning of cluster analysis using a training data set including explanatory variables and objective variables. Then, the trained model generated can be used to determine the group of melted states of the resin.

Here, in the cluster analysis, the number of groups in the molten state of the resin is preset. That is, a trained model is generated so as to classify into a preset number of groups. Although the number of groups will be described later, the number of groups is such that the desired quality of the molded product can be obtained when the molding conditions are modified by the modification amount of the molding conditions set for each group of the molten state of the resin. It is good to set it to.

(5. Configuration of resin state estimation device 3)
The configuration of the resin state estimation device 3 will be described with reference to FIG. The resin state estimation device 3 is a device for acquiring a group of the melted states of the resin described above. The resin state estimation device 3 includes, for example, an arithmetic processing unit including a processor, a storage device, an interface, and the like, an input device that can be connected to the interface of the arithmetic processing unit, and an output device that can be connected to the interface of the arithmetic processing unit. .. The output device may include, for example, a display device. Further, the arithmetic processing unit, the input device, and the output device may form one unit without going through the interface. Further, a physical server or a cloud server can be applied to a part of the arithmetic processing unit and a part of the storage device.

As shown in FIG. 6, the resin state estimation device 3 includes a detection data acquisition unit 101, a feature amount generation unit 102, a control parameter acquisition unit 103, an identification parameter value calculation unit 104, a trained model generation unit 105, and a trained model storage unit. A unit 106 and a group acquisition unit 107 are provided.

The detection data acquisition unit 101 acquires the detection data detected at the time of molding by the sensors 44 and 45 attached to the molding machine 2. For example, the detection data detected by the pressure sensor 44 is time series data as shown in FIG. Although not shown, the detection data detected by the temperature sensor 45 is also acquired as time series data.

The feature amount generation unit 102 generates a feature amount group [A] composed of a plurality of types of feature amounts related to the detected data based on the detected data (see equation (1)). That is, a plurality of types of feature quantities are generated for each detected data. The feature amount is a statistic (maximum value, minimum value, average value, dispersion, maximum value of differentiation) in each detection data of the sensors 44 and 45 in each molding process (injection filling process, pressure holding process, cooling process, etc.). Minimum value of differentiation, integrated value, etc.).

The control parameter acquisition unit 103 acquires a control parameter value group [F] composed of a plurality of types of control parameter values used for control in the molding machine 2 (see equation (1)). The types of control parameter values are, for example, injection speed, holding pressure, holding pressure time, mold temperature at the time of holding pressure, cooling time, nozzle temperature, injection pressure (nozzle pressure) and the like.

The identification parameter value calculation unit 104 calculates the resin state identification parameter value P1-Pm representing the molten state of the resin corresponding to each feature amount based on the feature amount group [A] and the control parameter value group [F]. .. The identification parameter value calculation unit 104 calculates the resin state identification parameter value P1-Pm by the above-mentioned equations (1) and (2).

The identification parameter value calculation unit 104 calculates the resin state identification parameter value Pj using all the control parameter values Ak in each feature amount F1-Fm as in the above equations (3) and (4). May be. In this case, the resin state identification parameter value Pj is an infinite product value of all control parameter values Ak.

As in the above equations (5) and (6), the identification parameter value calculation unit 104 controls one or a plurality of features having a high degree of influence on the feature amount Fj and the feature amount Fj in each feature amount F1-Fm. The resin state identification parameter value Pj corresponding to the feature amount Fj may be calculated by using the parameter value Ak. In this case, the resin state identification parameter value Pj is an infinite product value of some control parameter values Ak.

For example, one or a plurality of control parameter values Ak having a high influence on the feature amount Fj may be extracted by machine learning using the target feature amount Fj and the control parameter value group [A]. For example, using the influence coefficient obtained by machine learning, a predetermined number of control parameter values Ak can be extracted from the one having the higher impact coefficient. The influence coefficient is, for example, a lasso coefficient obtained by lasso regression, a ridge coefficient obtained by ridge regression, or the like.

The trained model generation unit 105 generates a trained model of machine learning as a learning phase. When the trained model generation unit 105 defines that the molten state of the resin is classified into a plurality of groups G1, G2, ..., The molten state of the resin is set to the resin state identification parameter value P1-Pm as an explanatory variable. Group G1, G2, ... Are set as objective variables, and cluster analysis is applied as multivariate analysis.

The trained model generation unit 105 generates a trained model by performing machine learning of cluster analysis using a training data set including the above explanatory variables and objective variables. The generated trained model is stored in the trained model storage unit 106. That is, the trained model outputs a group when the resin state identification parameter value is input.

The group acquisition unit 107 acquires the groups G1, G2, ... In the molten state of the resin based on the resin state identification parameter values P1-Pm. In this example, the group acquisition unit 107 uses a trained model generated by applying a cluster analysis as a multivariate analysis using the resin state identification parameter value P1-Pm as an explanatory variable. That is, when the group acquisition unit 107 inputs the resin state identification parameter value P1-Pm as the application phase (also referred to as inference phase) of machine learning, the group G1 in the molten state of the resin as the output of the trained model. Acquire G2, ...

As described above, the resin state estimation device 3 uses the detection data at the time of molding the target molded product and the control parameter value used for molding the target molded product to mold the target molded product. The group of melted states of the resin in the cavity C at the time is determined.

(6. Effect of resin state estimation device 3)
As described above, the detection data detected by the sensors 44 and 45 attached to the molding machine 2 at the time of molding is considered to be influenced by the control parameters of the molding machine 2 and the molten state of the resin in the cavity C. In other words, it is defined that the molten state of the resin is represented by the feature amount group [F] and the control parameter value group [A] generated from the detection data.

Using this definition, the identification parameter value calculation unit 104 determines the molten state of the resin corresponding to each feature amount F1-Fm based on the feature amount group [F] and the control parameter value group [A] of the detection data. The resin state identification parameter value P1-Pm representing the above is calculated. That is, the resin state identification parameter values P1-Pm are generated in the same number as the number of types of the feature amount F1-Fm.

Then, when the group acquisition unit 107 defines that the molten state of the resin is classified into a plurality of groups G1, G2, ..., The resin state identification parameter value is based on the resin state identification parameter value P1-Pm. By applying multivariate analysis with P1-Pm as an explanatory variable, the groups G1, G2, ... Of the molten state of the resin are acquired. Here, the groups G1, G2, ... In the molten state of the resin do not need to be clearly defined, but for example, the degree of fluidity can be classified as one of the elements.

That is, according to the resin state estimation device 3, by performing arithmetic processing using the detection data and the control parameters, the groups G1, G2, ... -For example, the degree of fluidity of the resin can be classified as one of the elements. By grouping the molten state of the resin in the cavity C in this way, it is possible to determine the amount of modification of the molding conditions according to the groups G1, G2, ....

(7. Molding machine system 1B of the second example)
A second example molding machine system 1B including the molding condition determination support device 4 will be described with reference to FIG. 7. As shown in FIG. 7, the molding machine system 1B includes a molding machine 2 and a molding condition determination support device 4.

The molding condition determination support device 4 is a device for determining molding conditions in the molding machine 2. In particular, in this example, the molding condition determination support device 4 can improve the quality of the molded product when the molded product is molded according to the molding conditions already applied. Determine the amount of correction. Further, the molding condition determination support device 4 includes the resin state estimation device 3 described above, and processing using the group of the melted state of the resin obtained by the resin state estimation device 3 is performed.

(8. Configuration of molding condition determination support device 4)
The configuration of the molding condition determination support device 4 will be described with reference to FIGS. 8-12. The molding condition determination support device 4 includes, for example, an arithmetic processing unit including a processor, a storage device, an interface, etc., an input device that can be connected to the interface of the arithmetic processing unit, and an output device that can be connected to the interface of the arithmetic processing unit. Be prepared. The output device may include, for example, a display device. Further, the arithmetic processing unit, the input device, and the output device may form one unit without going through the interface. Further, a physical server or a cloud server can be applied to a part of the arithmetic processing unit and a part of the storage device.

As shown in FIG. 8, the molding condition determination support device 4 includes a resin state estimation device 3, a quality estimation unit 201, a quality transition storage unit 202, a tendency evaluation unit 203, a relationship storage unit 204, and a correction condition determination unit 205. The resin state estimation device 3 has the same configuration as the resin state estimation device 3 described in the molding machine system 1A of the first example.

The quality estimation unit 201 estimates the quality of the molded product by machine learning based on the detection data acquired by the detection data acquisition unit 101. For example, the quality estimation unit 201 estimates numerical values for one or more quality species in an article. The quality type of the article is at least one of the mass of the article, the dimensions of the article, and the void volume in the article.

The quality estimation unit 201 generates a trained model in which the relationship between the detection data and the quality of the molded product is learned in advance by machine learning. The trained model is generated for each quality species. Then, the quality estimation unit 201 stores the trained model and estimates the quality of the molded product using the newly acquired detection data and the trained model.

The quality transition storage unit 202 accumulates the quality of the molded product estimated by the quality estimation unit 201, and stores the quality transition of the accumulated plurality of molded products. The quality transition is information in which the qualities of a plurality of molded products are arranged in the order of molding. The quality transition is, for example, the data as shown in FIG. FIG. 9 is an example of the mass of the molded product. In FIG. 9, the predetermined quality standard is Std, the upper limit of the quality tolerance is Thmax, and the lower limit of the quality tolerance is Thmin. That is, the range between the upper limit value Thmax and the lower limit value Thmin means that the product is a non-defective product, and the deviation range is a defective product. Ideally, even if it is a good product, it meets the predetermined quality standard Std.

In FIG. 9, in the initial stage of molding, the quality of most molded products shows a value near the quality standard Std. However, as a sudden abnormality, there is a molded product showing a value exceeding the upper limit value Thmax (D2 in FIG. 9). That is, when the molded product having a sudden abnormality is excluded, the quality of the molded product at the initial stage of molding shows a value near the quality standard Std (D1 in FIG. 9). After that, as the molding is continued, the quality of the molded product gradually deviates from the quality standard Std. The quality of the molded product is in the vicinity of a value increased by + N% from the quality standard Std (D3 in FIG. 9).

The tendency evaluation unit 203 evaluates the quality change tendency with respect to the quality standard Std based on the quality transition stored in the quality transition storage unit 202. As shown in FIG. 10, the tendency evaluation unit 203 sets the tendency of quality change as the degree of quality deviation from the quality standard Std in each of a plurality of quality types and quality types, and the quality of a plurality of molded products in each quality type. Evaluate the degree of variability.

The tendency evaluation unit 203 presets the number of molded products of quality transition used for evaluating the quality change tendency. That is, the tendency evaluation unit 203 calculates the degree of quality deviation in the preset number of molded products. For example, the tendency evaluation unit 203 calculates the degree of deviation (absolute value or relative value) of the average value of the quality in the number of molded products from the quality standard Std as the degree of deviation of the quality. Further, the degree of variation indicates the degree of variation or stability in the preset number of molded products. As the degree of variation, for example, values such as standard deviation and variance may be used.

As shown in FIG. 10, for example, the tendency evaluation unit 203 evaluates the mass as the degree of deviation "+ 2.2%" and the degree of variation "stable", and the dimensions as the degree of deviation "+ 0.3%" and the degree of variation. It is evaluated as "stable", and the void volume is evaluated as the degree of deviation "-0.5%" and the degree of variation "stable".

Further, in D1 of the quality transition shown in FIG. 9, the tendency evaluation unit 203 evaluates that the quality change tendency is stably located near the quality standard Std. Here, the tendency evaluation unit 203 evaluates by excluding the molded product of D2 showing the quality change due to the sudden abnormality in D1 of the quality transition shown in FIG. In this case, the tendency evaluation unit 203 evaluates the target quality species as "0.1%" as the degree of deviation from the quality standard Std and as "stable" as the degree of variation. Since the quality of the molded product returns to normal after the sudden abnormality, the sudden abnormality is irrelevant to the molding conditions. Therefore, the evaluation excludes sudden abnormalities.

Further, in D3 of the quality transition shown in FIG. 9, it is evaluated that the tendency evaluation unit 203 stably shows a value of about + N% with respect to the quality standard Std as the quality change tendency. In this case, the tendency evaluation unit 203 evaluates the target quality species as "+ N%" as the degree of deviation from the quality standard Std and as "stable" as the degree of variation.

As described above, the group acquisition unit 107 in the resin state estimation device 3 acquires a group of the molten state of the resin in the cavity C. Here, in the group acquisition unit 107, the group in the molten state of the resin is classified into four types, that is, Type-A, Type-B, Type-C, and Type-D, for example, as shown in FIG. It shall be.

The relationship storage unit 204 stores the relationship between the quality change tendency and the correction amount of the molding conditions for returning the quality to the quality standard in association with the group of the molten state of the resin in the cavity C. As shown in FIG. 12, for example, the relationship storage unit 204 stores the relationship between the quality change tendency and the correction amount of the molding conditions for each group of the molten state of the resin. Specifically, the relationship storage unit 204 stores the relationship in which the stage of the degree of deviation and the correction amount of the molding conditions are represented by a matrix for each quality type in each of the groups of the molten state of the resin. For example, for each of the mass, the dimension, and the void volume, there are 6 stages of the degree of deviation. Further, as the seed of the molding condition to be modified, at least one of injection speed, holding pressure, holding pressure time, mold temperature at the time of holding pressure, cooling time and the like is set.

Here, the degree of the relationship between the quality seed and the seed of the molding condition and the relationship between the degree of deviation of the quality seed and the correction amount of the molding condition can be derived by using machine learning. That is, the relationship stored in the relationship storage unit 204 can be generated by machine learning. Of course, the relationship may be set based on experiments, past experiences, etc., instead of machine learning.

The correction condition determination unit 205 is based on the quality change tendency evaluated by the tendency evaluation unit 203, the group of the molten state of the resin acquired by the group acquisition unit 107, and the relationship stored in the relationship storage unit 204. , Determine the amount of modification of the molding conditions.

The correction condition determination unit 205 first selects from the relationships shown in FIG. 12 the one corresponding to the group in the molten state of the resin acquired by the group acquisition unit 107. Subsequently, for example, the correction condition determination unit 205 determines the stage closest to the degree of deviation evaluated by the tendency evaluation unit 203 in the relationship represented by the matrix as shown in FIG. 12, and the molding conditions corresponding to the stage are determined. The correction amount of is used as the correction amount of the molding conditions to be determined. For example, as shown in FIG. 10, when the degree of mass deviation is “+ 2.2%”, the step “+ 2%” of the degree of mass deviation is selected in the matrix shown in FIG.

Here, the modification condition determination unit 205 determines the modification amount of the molding conditions for each of the plurality of quality species. Therefore, the modification condition determination unit 205 determines the final modification amount for the molding condition based on the plurality of modification amounts for the same type of molding condition. In this case, the correction condition determination unit 205 may, for example, use the total value of a plurality of correction amounts as the final correction amount, or multiply the weighting coefficient according to the quality type and then add the total value. It may be the final correction amount.

Further, the correction condition determination unit 205 outputs the final correction amount to the control device 60 of the molding machine 2, and corrects the molding condition in the next molded product by the amount of the correction amount. Therefore, the control device 60 executes the molding of the next molded product according to the modified molding conditions. As a result, the quality of the molded product molded under the modified molding conditions can be brought closer to the quality standard Std.

(9. Effect of molding condition determination support device 4)
The effect of the molding condition determination support device 4 described above will be described. The quality transition storage unit 202 stores the quality of the molded product estimated by machine learning and stores the quality transition. The quality transition is information in which the qualities of a plurality of molded products are arranged in the order of molding. Therefore, the tendency evaluation unit 203 can evaluate the quality change tendency based on the quality transition of a plurality of consecutive molded products.

In particular, the trend evaluation unit 203 evaluates the quality change tendency with respect to the predetermined quality standard Std. For example, the tendency evaluation unit 203 keeps the state where the quality deviates from the predetermined quality standard Std as the quality change tendency, or the quality fluctuates within the quality permissible range including the predetermined quality standard. It is possible to evaluate things.

The relationship storage unit 204 stores in advance the relationship between the quality change tendency and the correction amount of the molding conditions in association with the group of the molten state of the resin in the cavity C. That is, the relationship storage unit 204 stores the relationship between the quality change tendency and the correction amount of the molding conditions for each group of the molten state of the resin. This relationship is set using the know-how of experts, output results by machine learning, experimental results, and the like.

The modification condition determination unit 205 determines the molding conditions based on the newly evaluated quality change tendency, the newly acquired group of the melted state of the resin in the cavity C, and the relationship stored in the relationship storage unit 204. Determine the amount of correction. Here, the relationship stored in the relationship storage unit 204 relates to the amount of modification of the molding conditions for returning the quality to the predetermined quality standard Std. In particular, the amount of modification of the molding conditions is determined according to the group of the molten state of the resin in the cavity C. Therefore, when the molding conditions of the molding machine 2 are modified according to the modification amount of the molding conditions determined by the modification condition determination unit 205, the quality of the molded product to be molded next can be brought closer to the predetermined quality standard Std. Will be.

That is, even if the quality of the molded product changes due to an external factor such as the environmental temperature or a slight difference in the components contained in the material of the molding material, by grasping the tendency of the quality change, the resin in the cavity C can be further improved. By grouping the molten states of, the molding conditions can be modified so that the quality of the molded product can be set to a predetermined quality standard Std. Therefore, not only a skilled person but also an unskilled person can modify the molding conditions so as to improve the quality of the molded product.

(10. How to set the number of groups)
In addition to generating the trained model described above, the trained model generation unit 105 performs a process of setting the number of groups in the molten state of the resin in cooperation with other configurations. Here, the number of groups can be set to an arbitrary number by the operator, but can be set to an appropriate number by using the molding condition determination support device 4 described above.

The trained model generation unit 105 has the number of groups of the molten state of the resin set to the initial set number (for example, 2), and the group of the melted state of the resin (either G1 or G2) for the acquired detection data. ) (Group acquisition process).

Then, when the number of groups is set to the initial set number, the relationship between the quality change tendency and the correction amount of the molding conditions is set in association with the group in the molten state of the resin (relationship setting step). Then, the relationship storage unit 204 stores the relationship. Subsequently, the correction condition determination unit 205 corrects 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 204. (Correction amount determination step).

Subsequently, when the trained model generation unit 105 corrects the control parameter value based on the correction amount of the control parameter value corresponding to the acquired group G1 and G2 of the molten state of the resin, the quality of the molded product is within a predetermined range. It is determined whether or not the condition is satisfied (determination step).

When the quality of the molded product does not satisfy the predetermined range, the number of groups is increased, and the above-mentioned group acquisition step, relationship setting step, correction amount determination step, and determination step are repeated. That is, if the quality of the molded product does not meet the predetermined range, the trained model generation unit 105 increases the number of groups to obtain the group in the molten state of the resin and whether the quality of the molded product satisfies the predetermined range. The determination of whether or not is repeated is repeated, and the number of groups when the quality of the molded product satisfies a predetermined range is set as the number of groups in the trained model.

By setting the number of groups in this way, the amount of modification of the molding conditions can be determined according to the group of the molten state of the resin. That is, it is possible to determine an appropriate amount of modification of molding conditions.

(11. Configuration example of molding machine system 1B)
(11-1. First example)
The configuration of the first example of the molding machine system 1B will be described with reference to FIG. As shown in FIG. 13, the molding machine system 1B includes a plurality of molding machines 2 and 2, an edge computer 5 and 5 integrally configured in each of the molding machines 2 and 2, and a plurality of molding machines 2 and 2. It includes a server 6 that constitutes the same network. The edge computers 5 and 5 may be configured as a part of the molding machines 2 and 2, or may be configured separately from the molding machines 2 and 2.

Then, the edge computers 5 and 5 and the server 6 constitute the molding condition determination support device 4. The edge computers 5 and 5 include a detection data acquisition unit 101. The server 6 includes a configuration excluding the detection data acquisition unit 101 in the resin state estimation device 3, a quality estimation unit 201, a quality transition storage unit 202, a tendency evaluation unit 203, a relationship storage unit 204, and a correction condition determination unit 205.

That is, the server 6 receives the detection data acquired by the detection data acquisition unit 101 from the edge computer 5 or the edge computer 5 built in the molding machine 2. The server 6 determines the modified amount of the molding condition based on the received information, and transmits the modified amount of the determined molding condition to the molding machine 2.

In this case, the server 6 can store information about the plurality of molding machines 2. Further, by providing the server 6 with a processor capable of executing high-speed processing, the processing of the feature amount generation unit 102, the processing of the identification parameter value calculation unit 104, the processing of the group acquisition unit 107, and the processing of the quality estimation unit 201. , The processing of the tendency evaluation unit 203 and the processing of the correction condition determination unit 205 can be processed at high speed. On the other hand, since it is not necessary to make each edge computer 5 have high specifications, it is possible to reduce the cost.

(11-2. Second example)
The second example of the molding machine system 1B is the same as the first example, in the molding machine system 1B, the plurality of molding machines 2 and 2 and the edge computers 5 and 5 connected to the molding machines 2 and 2 respectively. A server 6 constituting the same network as the molding machines 2 and 2 of the above is provided.

The edge computers 5 and 5 include all the configurations of the resin state estimation device 3 and the quality estimation unit 201. The server 6 includes a quality transition storage unit 202, a trend evaluation unit 203, a relationship storage unit 204, and a correction condition determination unit 205. That is, the server 6 determines the group of the molten state of the resin acquired by the group acquisition unit 107 of the resin state estimation device 3 and the quality of the molded product estimated by the quality estimation unit 201 by the edge computer 5 or the molding machine 2. Received from the edge computer 5 built in the. The server 6 determines the modified amount of the molding condition based on the received information, and transmits the modified amount of the determined molding condition to the molding machine 2.

(10-3. Third example)
In the third example of the molding machine system 1B, the server 6 has all the functions of the molding condition determination support device 4. In this case, the edge computers 5 and 5 are unnecessary. The detection data acquisition unit 101 in the server 6 receives the detection data detected by the sensors 44 and 45 from the molding machine 2. Then, the correction condition determination unit 205 in the server 6 transmits the correction amount of the molding conditions to the molding machine 2.

1A, 1B: Molding machine system, 2: Injection molding machine, 3: Resin state estimation device, 4: Molding condition determination support device, 5: Edge computer, 6: Server, 20: Bed, 30: Injection device, 40: Mold , 50: Molding device, 60: Control device, 101: Detection data acquisition unit, 102: Feature quantity generation unit, 103: Control parameter acquisition unit, 104: Identification parameter value calculation unit, 105: Trained model generation unit, 106 : Trained model storage unit, 107: Group acquisition unit, 201: Quality estimation unit, 202: Quality transition storage unit, 203: Trend evaluation unit, 204: Relationship storage unit, 205: Correction condition determination unit, C: Cavity

Claims (7)

It is a resin state estimation device that estimates the molten state of the resin in the cavity of the mold of the injection molding machine.
A detection data acquisition unit that acquires detection data detected during molding by a sensor attached to the injection molding machine, and a detection data acquisition unit.
Based on the detection data, a feature amount generation unit that generates a feature amount group composed of a plurality of types of feature amounts related to the detection data, and a feature amount generation unit.
A control parameter acquisition unit that acquires a control parameter value group composed of a plurality of types of control parameter values used for control in the injection molding machine, and a control parameter acquisition unit.
It is defined that the molten state of the resin is represented by the feature amount group and the control parameter value group, and the resin corresponding to the feature amount is based on the feature amount group and the control parameter value group. An identification parameter value calculation unit that calculates a resin state identification parameter value that represents a molten state,
When the molten state of the resin is defined to be classified into a plurality of groups, the multivariate analysis using the resin state identification parameter value as an explanatory variable based on the resin state identification parameter value is applied. The group acquisition unit that acquires the group of the molten state of the resin,
A resin state estimation device. The resin state estimation device further
The resin state identification parameter value is used as an explanatory variable, the group of melted states of the resin is used as an objective variable, cluster analysis is applied as the multivariate analysis, and the training data set including the explanatory variable and the objective variable is used. It has a trained model storage unit that stores the trained models generated by performing machine learning of cluster analysis.
The resin state estimation device according to claim 1, wherein the group acquisition unit acquires a group of the molten state of the resin as an output of the learned model when the resin state identification parameter value is input. The resin state estimation device according to claim 1 or 2, wherein the identification parameter value calculation unit calculates the resin state identification parameter value by the equations (1) and (2).

The identification parameter value calculation unit uses the feature amount and one or more control parameter values having a high influence on the feature amount in each of the feature amounts, and the resin corresponding to the feature amount. The resin state estimation device according to any one of claims 1-3, which calculates a state identification parameter value. The resin state according to claim 4, wherein the one or a plurality of the control parameter values having a high influence on the feature amount are extracted by machine learning using the target feature amount and the control parameter value group. Estimator. The resin state estimation device further
A trained model generator that generates the trained model by performing machine learning of the cluster analysis using the training data set including the explanatory variables and the objective variable is provided.
The trained model generation unit
With the number of groups in the molten state of the resin set to the initial set number, the group in the melted state of the resin for the acquired detection data is acquired.
When the control parameter value is corrected based on the corrected amount of the control parameter value corresponding to the acquired group of the molten state of the resin, it is determined whether or not the quality of the molded product satisfies a predetermined range.
When the quality of the molded product does not meet the predetermined range, the number of the groups is increased, and the acquisition of the group in the molten state of the resin and the determination of whether or not the quality of the molded product satisfies the predetermined range are repeated. The resin state estimation device according to claim 2, wherein the number of groups when the quality of the molded product satisfies a predetermined range is set to the number of groups in the trained model. It is a molding condition determination support device that is applied to a molding method that molds a molded product by supplying a molten material in which resin is melted to the cavity of the mold of an injection molding machine, and determines the molding conditions of the molded product.
The resin state estimation device according to any one of claims 1-6, and the resin state estimation device.
A quality estimation unit that estimates the quality of the molded product by machine learning based on the detection data,
A quality transition storage unit that accumulates the estimated quality of the molded product and stores the accumulated quality transition of the plurality of molded products.
A tendency evaluation unit that evaluates the tendency of quality change with respect to a predetermined quality standard based on the quality transition, and
A relational storage unit that stores the relationship between the quality change tendency and the amount of modification of the molding conditions for returning the quality to the quality standard in association with the group of the molten state of the resin.
The molding conditions are based on the quality change tendency evaluated by the tendency 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. The correction condition determination unit that determines the correction amount of
A molding condition determination support device.

Images