JP7761026B2 - Blending condition setting method, blending condition setting program, blending condition setting device, and polyester resin film manufacturing system - Google Patents

Description

The present invention relates to a trained model generation method, a blending condition setting method, a trained model generation program, a blending condition setting program, a learning device, a blending condition setting device, a resin film manufacturing method, a resin film manufacturing system, and a resin film.

Resin films are films that have excellent heat resistance, flame retardancy, chemical resistance, electrical insulation, moist heat resistance, mechanical strength, and dimensional stability. Such resin films are produced, for example, by biaxial stretching, in which resin chips are melted, stretched in one direction, and then stretched in a direction perpendicular to the melted resin chips (see, for example, Patent Documents 1 and 2). This biaxial stretching process results in a film with increased strength in both the longitudinal and transverse directions.

Japanese Patent Application Laid-Open No. 2001-287266 JP 2011-148186 A

However, in order to obtain the resin film described above with the desired physical properties, engineers have to go through extensive trial and error to design the compounding ratios and other compounding conditions, and there has been a demand for more efficient design.

The present invention has been made in consideration of the above, and aims to provide a trained model generation method, a blending condition setting method, a trained model generation program, a blending condition setting program, a learning device, a blending condition setting device, a resin film manufacturing method, a resin film manufacturing system, and a resin film that can efficiently set blending conditions for resin films.

To solve the above-mentioned problems and achieve the objectives, the trained model generation method of the present invention generates a trained model by having a computer train using training data in which the compounding conditions, including the physical properties and compounding ratios of the thermoplastic resin that makes up the resin film, are used as explanatory variables, and the physical properties of the resin film are used as objective variables.

Furthermore, in the trained model generation method according to the present invention, in the above invention, the physical properties of the thermoplastic resin include the intrinsic viscosity and solution haze of the thermoplastic resin.

Furthermore, in the trained model generation method according to the present invention, in the above invention, the physical properties of the resin film include the intrinsic viscosity and film haze of the resin film.

Furthermore, in the trained model generation method according to the present invention, in the above invention, the thermoplastic resin includes recycled resin.

Furthermore, in the trained model generation method according to the present invention, in the above invention, the thermoplastic resin is polyester resin.

Furthermore, in the trained model generation method according to the present invention, in the above invention, the resin film is a biaxially oriented polyester film.

In addition, the compounding condition setting method of the present invention involves inputting the physical properties of the target resin film into a search unit that includes a trained model generated by a computer learning using training data in which compounding conditions, including the physical properties and compounding ratios of the thermoplastic resin that makes up the resin film, are used as explanatory variables, and the physical properties of the resin film are used as target variables, and setting the compounding conditions based on the output results of the search unit.

Furthermore, in the blending condition setting method according to the present invention, in the above invention, when the output result contains multiple blending conditions, the step of setting the blending conditions selects the blending condition with the highest proportion of recycled resin.

Furthermore, in the blending condition setting method according to the present invention, in the above invention, the physical properties of the thermoplastic resin include the intrinsic viscosity and solution haze of the thermoplastic resin.

In addition, in the blending condition setting method according to the present invention, the physical properties of the resin film include the intrinsic viscosity and film haze of the resin film.

Furthermore, in the blending condition setting method according to the present invention, the thermoplastic resin includes recycled resin.

Furthermore, in the blending condition setting method according to the present invention, the thermoplastic resin is a polyester resin.

Furthermore, in the blending condition setting method according to the present invention, the resin film is a biaxially oriented polyester film.

The trained model generation program of the present invention also causes a computer to execute a trained model generation step of generating a trained model by learning using training data in which the blending conditions, including the physical properties and blending ratios of the thermoplastic resin that constitutes the resin film, are used as explanatory variables, and the physical properties of the resin film are used as objective variables.

The compounding condition setting program of the present invention also causes a computer to execute a compounding condition setting step of inputting the physical properties of the target resin film into a search unit that includes a trained model generated by learning using training data in which compounding conditions, including the physical properties and compounding ratios of the thermoplastic resins that make up the resin film, are used as explanatory variables and the physical properties of the resin film are used as target variables, and setting compounding conditions based on the output results of the search unit.

The learning device according to the present invention also includes a learning unit that generates a trained model by learning using training data in which the blending conditions, including the physical properties and blending ratios of the thermoplastic resin that constitutes the resin film, are explanatory variables, and the physical properties of the resin film are objective variables.

The blending condition setting device according to the present invention also includes a trained model generated by training using training data in which blending conditions, including the physical properties and blending ratios of the thermoplastic resins that make up the resin film, are used as explanatory variables, and the physical properties of the resin film are used as target variables. The device also includes a search unit that inputs the desired physical properties of the resin film into the trained model to search for blending conditions suitable for the properties, and a setting unit that sets the blending conditions based on the output results of the search unit.

The resin film manufacturing method according to the present invention also produces a resin film according to the thermoplastic resin blending conditions set by the blending condition setting method according to the above invention.

The resin film manufacturing system according to the present invention also includes the blending condition setting device according to the above invention, and a film manufacturing device that manufactures a resin film in accordance with the blending conditions for the thermoplastic resin set by the blending condition setting device.

The resin film according to the present invention is produced by mixing the thermoplastic resins according to the blending conditions set by the blending condition setting method according to the above invention, and molding the mixture into a sheet.

The present invention makes it possible to efficiently set the compounding conditions for resin films.

FIG. 1 is a diagram showing a schematic configuration of a compounding condition setting system according to an embodiment of the present invention. FIG. 2 is a diagram for explaining the flow of the compounding condition setting process performed by the compounding condition setting system according to one embodiment of the present invention. FIG. 3 is a diagram for explaining an example of a manufacturing method using biaxial stretching. FIG. 4 is a block diagram showing the configuration of a learning device provided in a compounding condition setting system according to an embodiment of the present invention. FIG. 5 is a block diagram showing the configuration of a blending condition setting device provided in a blending condition setting system according to one embodiment of the present invention. FIG. 6 is a flowchart showing an outline of the learning process performed by the learning device according to one embodiment of the present invention. FIG. 7 is a flowchart showing an outline of the blending condition setting process performed by the blending condition setting device according to one embodiment of the present invention. FIG. 8 is a flowchart showing an outline of the blending condition setting process performed by the blending condition setting device according to the modified embodiment of the present invention.

Below, an embodiment of the compounding condition setting system according to the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to this embodiment.

(Embodiment)
1 is a diagram showing a schematic configuration of a blending condition setting system according to one embodiment of the present invention. The blending condition setting system 1 includes a learning device 2 that creates training data and generates a trained model trained using the created training data, a blending condition setting device 3 that sets blending conditions that will give the resin film properties desired by the user using the trained model generated by the learning device 2, a display device 4 that displays information including the setting results of the blending condition setting device 3, and an input device 5.

The blending conditions set by the blending condition setting device 3 include the types of raw materials (resins) that make up the resin film and their blending ratios.

The resin that constitutes the resin film can be a thermoplastic resin. Examples of thermoplastic resins include polyester resin. Thermoplastic resins include new resins that have not been subjected to processing such as hot melt molding or heat treatment, and recycled resins that have been processed such as hot melt molding or heat treatment. Recycled resins can be used, for example, from process waste materials generated during product production or discarded or recovered products. There are no limitations on the form, but they can be used in the form of melt-extruded pellets or compressed flakes.

2 is a diagram illustrating the flow of a blending condition setting process performed by a blending condition setting system according to an embodiment of the present invention. The learning device 2 generates a trained model 100 by learning using training data LD.
The blending condition setting device 3 inputs physical property conditions IP indicating the desired physical properties into the trained model 100 to obtain blending conditions (predicted conditions), and sets and outputs blending conditions OP that satisfy the physical properties (predictions) of the resin film.
Alternatively, the formulation condition setting device 3 inputs physical property conditions IP indicating the desired physical properties into the trained model 100 to obtain formulation conditions (prediction conditions), and outputs the obtained one or more formulation conditions OP.

Here, an example of a method for manufacturing a resin film will be described with reference to Figure 3. Figure 3 is a diagram for explaining an example of a manufacturing method using biaxial stretching. In Figure 3, the longitudinal direction of the film along which the resin is fed in the processing step is the X-axis direction, the width direction of the resin film (the same as the rotation axis of the roll that feeds it to the next step) is the Y-axis direction, and the height direction perpendicular to the X-axis and Y-axis directions is the Z-axis direction.

The resin film manufacturing apparatus 200 includes a manufacturing unit 201 and a control unit 202 .
The blending condition setting device 3 and the manufacturing device 200 constitute a resin film manufacturing system. The manufacturing device 200 is not limited to a manufacturing unit that manufactures a resin film by biaxial stretching, such as the manufacturing unit 201 shown in FIG. 3 .

The manufacturing section 201 includes a vacuum drying section 211 that removes moisture from the pellets (resin), a storage section 212 that stores the vacuum-dried pellets, an extrusion section 213 that melts and extrudes the pellets sent from the storage section 212, a filtration filter 214 that filters the molten resin sent from the extrusion section 213, a die 215 that forms the filtered molten resin into a sheet, a casting drum 216 that wraps the sheet-like resin from the die 215 around and cools and solidifies it, a first stretching section 217 that stretches the sheet-like resin in the X-axis direction, a second stretching section 218 that stretches in the Y-axis direction perpendicular to the stretching direction by the first stretching section 217, a transfer conveying section 219 that conveys the biaxially stretched film that has passed through the second stretching section 218 while cooling it, and a winding section 220 that winds up the stretched resin sent from the transfer conveying section 219.

The extrusion section 213 mixes and melts the pellets and forces them through a die 215 .
The casting drum 216 rotates around an axis in the Y-axis direction, cools and solidifies the resin, and sends the sheet-like resin in the film longitudinal direction (X-axis direction).

The first stretching section 217 stretches the sheet-like resin delivered from the casting drum 216 in the longitudinal direction (X-axis direction) of the film using multiple rolls, while heating as necessary, to form a uniaxially stretched film.

The second stretching unit 218 stretches the uniaxially stretched film (221) sent from the first stretching unit 217 in a direction perpendicular to the longitudinal direction of the film using clips, while heating it as necessary. Here, it is stretched in the Y-axis direction (the film width direction/direction of the roll rotation axis) to form a biaxially stretched film.

The winding section 220 is a roll that winds up the biaxially stretched film after stretching in the second stretching section 218 via the transfer conveying section 219. Through this series of processes, multiple resin raw materials are mixed, melted, stretched, and formed into a sheet, producing a biaxially stretched resin film that is stretched in two perpendicular directions. Because biaxially stretched resin film is stretched in different directions, it exhibits higher isotropy than uniaxially stretched film, and exhibits, for example, excellent dimensional stability.

The control unit 202 controls the operation of the production unit 201. For example, when production conditions for biaxial stretching, such as roll temperature and rotation speed, are set in the blending condition setting device 3, the control unit 202 acquires the blending conditions and drives the production unit 201 in accordance with those production conditions. The control unit 202 is a computer configured using one or more hardware components, such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), and memory on which various programs are pre-installed.

Next, the configuration of the learning device 2 will be described with reference to Figure 4. The learning device 2 is electrically connected to the blending condition setting device 3. Figure 4 is a block diagram showing the configuration of the learning device provided in a blending condition setting system according to one embodiment of the present invention. The learning device 2 has a learning data generation unit 21, a learning unit 22, a control unit 23, and a memory unit 24.

The learning data generation unit 21 generates, from the data stored in the memory unit 24, data that associates the types and physical properties of multiple raw materials that make up the resin film, the blending conditions consisting of the blending ratios of the raw materials, and the physical properties of the resin film manufactured under those blending conditions, as learning data. In other words, the learning data is data on resin films whose physical properties are known. Note that the learning data generation unit 21 may extract data according to conditions input via the input device 5 and generate learning data.
The learning data generation unit 21 generates learning data in which the compounding conditions, including the physical properties and compounding ratio of the thermoplastic resin that constitutes the resin film, are explanatory variables, and the physical properties of the resin film corresponding to the compounding conditions are objective variables.
In this case, the blending conditions such as the type of raw material, physical properties, blending ratio, etc. may be quantified or visualized. In addition, a plurality of blending conditions may be combined and quantified or visualized to be expressed as a single blending condition.
For example, if there are 40 possible combinations of ingredient types and blending ratios, one way to express this is to assign a number corresponding to each combination, obtain a number from 1 to 40, and then quantify this number as a single blending condition related to ingredient types and blending ratios. Using the above-mentioned representation makes it possible to reduce the number of explanatory variables in the generated training data and reduce multicollinearity, leading to the construction of a highly accurate trained model.
The physical properties of the resin film may be expressed numerically or as an image. Furthermore, when a plurality of physical properties of the resin film are set, these physical properties may be combined and expressed as a single physical property of the resin film.
The experimental data stored in the memory unit 24, which contains a mixture of formulation conditions and physical properties, is separated into explanatory variables (x1, x2, ...) and a response variable (y) in the learning data generation unit 21. In this case, there may be multiple types of formulation conditions used as explanatory variables, and each formulation condition is associated with a physical property. Note that multiple formulation conditions may be collectively treated as one. Furthermore, the explanatory variables and the response variable may be stored separately in the memory unit in advance.

Here, the known blending ratio of a resin film refers to data accumulated in the past that expresses the composition ratio of raw materials in a resin film as a weight ratio, and includes resin ratios and additive ratios as necessary.

The physical properties used as training data include the physical properties of the raw materials and the physical properties of the resin film. Examples of the physical properties of the raw materials include intrinsic viscosity, solution haze (turbidity), and color tone. Examples of the physical properties of the resin film include intrinsic viscosity, film haze (turbidity), color tone, thickness, and mechanical strength. When the resin is manufactured using biaxial stretching, the mechanical strength includes the strength in each stretching direction. For recycled resins, the physical properties may also include the details of previous treatments (heating conditions, etc.).
In this case, even if a new resin and a recycled resin are the same product, they may differ from each other in at least some of their physical properties, such as intrinsic viscosity, solution haze (turbidity), and color tone. In particular, recycled resins that have been recycled repeatedly may have relatively large differences in physical properties, such as intrinsic viscosity, film haze (turbidity), and color tone, compared to new resins.

The raw materials for resin films are a mixture of multiple resins to achieve the desired physical properties. For example, even when creating a film using polyester resin, there are different types of polyester resin raw materials available, such as polyethylene terephthalate (PET), polyethylene naphthalate (PEN), polytrimethylene terephthalate (PTT), and polybutylene terephthalate (PBT). Furthermore, even the same polyester resin can have different physical properties depending on whether it is recycled or not.

An example of raw material parameters will now be described with reference to Table 1. Table 1 shows an example of the physical properties (intrinsic viscosity, solution haze (turbidity), color tone (b value)) of polyester resin raw materials, and whether or not they can be recycled. In Table 1, No. 1 and No. 2 indicate the physical properties of new resins, and No. 3 to No. 7 indicate the physical properties of recycled resins. As shown in Table 1, even if the polyester resin is the same, the physical properties will differ depending on the type.

The learning unit 22 performs learning using the learning data to generate a trained model. The learning performed by the learning unit 22 can employ known learning methods. Statistical models that can be used for learning include, for example, simple linear regression models, ridge regression, Lasso regression, elastic net regression, general additive models, random forest regression, rule-fit regression, gradient boosting trees, extra trees, support vector regression, Gaussian process regression, k-nearest neighbor regression, kernel ridge regression, and neural networks.

For example, when the learning unit 22 generates a trained model through learning using regularization, the learning unit 22 is given multiple candidate values for the hyperparameters of the trained model, performs learning for each of the given candidate values for the hyperparameters, generates a trained model for the target variable (physical properties of the resin film), and stores it in the storage unit 24. The learning unit 22 then uses the training data to calculate the prediction error through cross-validation or holdout validation for the models obtained by learning using each candidate value, and selects the trained model that provides the smallest prediction error. Note that the hyperparameters referred to here are parameters that are set in advance by the learning unit 22 for learning, and include, for example, regularization coefficients. In addition, in the case of a trained model using a neural network, the hyperparameters also include, for example, the number of layers in the neural network.

The control unit 23 controls the overall operation of the learning device 2.

The memory unit 24 stores various programs for operating the learning device 2, as well as data including various parameters necessary for the operation of the learning device 2. The various programs include a learning data generation program that generates learning data for generating a trained model, and a trained model generation program that uses the training data to learn and generate a trained model. The various parameters include hyperparameters and parameters acquired by the learning unit 22 through learning. The memory unit 24 also stores data for constituting the training data (for example, the above-mentioned formulation conditions and physical properties).

The storage unit 24 is composed of a ROM (Read Only Memory) in which various programs are pre-installed, a RAM (Random Access Memory) for storing calculation parameters and data for each process, a HDD (Hard Disk Drive), an SSD (Solid State Drive), etc.

The various programs can also be recorded on computer-readable recording media such as HDDs, flash memory, CD-ROMs, DVD-ROMs, and Blu-ray (registered trademark) for widespread distribution. The communications network referred to here can be configured using, for example, existing public line networks, LANs (Local Area Networks), WANs (Wide Area Networks), etc., and can be either wired or wireless.

The learning device 2 having the above functional configuration is a computer configured using one or more pieces of hardware, such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), etc.

Next, the blending condition setting device 3 will be described with reference to Figure 5. Figure 5 is a block diagram showing the configuration of the blending condition setting device provided in a blending condition setting system according to one embodiment of the present invention. The blending condition setting device 3 is electrically connected to the learning device 2 and the display device 4. The blending condition setting device 3 has a setting unit 31, a search unit 32, a control unit 33, and a memory unit 34.

The blending condition setting device 3 uses the physical properties to be predicted and the trained model acquired from the learning device 2 to set blending conditions, including the types of ingredients and their blending ratios that are suitable for the conditions.

The setting unit 31 inputs the set physical properties to the search unit 32, which includes a trained model, and sets the candidate blending conditions output from the search unit 32 as the blending conditions. The blending conditions set at this time may be the most suitable blending condition out of multiple blending conditions that satisfy the set physical property conditions, or may be all blending conditions that satisfy the physical property conditions.

The search unit 32 uses the trained model to search for candidate blending conditions based on the physical properties input from the setting unit 31, extracting the optimal blending conditions and outputting them to the setting unit 31. Here, searching for candidate blending conditions using the trained model means inputting candidate blending conditions generated based on the search algorithm described below into the trained model, and extracting the optimal blending conditions to achieve the physical properties input from the setting unit 31 from the physical properties of the resin film output from the trained model.

Here, optimal blending conditions refer to a state in which the target physical property is at its most desirable value. In the present embodiment, for example, in the case of strength, this is synonymous with maximizing or bringing the property closest to a set value. If there is a desirable range, the physical property must fall within that range. Furthermore, if the set conditions include multiple types of physical properties, candidate blending conditions are extracted based on the difference between the respective physical property values, the priority assigned to each physical property, etc.

Search algorithms for obtaining optimal formulation conditions include a grid search method, which selects formulation conditions from a trained model that satisfy the target range for the physical properties of the desired resin film, and a random search method, which determines a predetermined upper limit on the number of searches and randomly extracts formulation conditions to obtain formulation conditions that maximize or minimize the physical properties. Another search method that can be used is the Bayesian optimization method, which selects and evaluates extreme values in the accumulated data within a confidence interval represented by the average physical property value ± prediction error, taking into account the prediction error of the physical properties, and then repeats the search. Another search method that can be used is the genetic algorithm selection method, which selects formulation conditions close to the desired physical properties and repeatedly searches for formulation conditions while changing some of those formulation conditions. Here, when using search algorithms that require a starting point before searching, such as the Bayesian optimization method or the genetic algorithm selection method, the starting point can be the data generated by grid search or random plotting, or the training data used to generate the trained model.

The control unit 33 controls the overall operation of the blending condition setting device 3. The control unit 33 has a display control unit 331 that displays the setting results of the setting unit 31 on the display device 4. In addition to the setting results, the display control unit 331 may also display information such as raw materials, the raw materials, and their predicted physical properties on the display device 4.

The memory unit 34 stores various programs for operating the blending condition setting device 3, as well as data including various parameters necessary for the operation of the blending condition setting device 3. The various programs include a blending condition setting program that is executed using a trained model. The memory unit 34 may also store trained models. In this case, the trained models may be updated in synchronization with the learning device 2. The memory unit 34 is composed of a ROM in which various programs are pre-installed, and a RAM, HDD, SSD, etc. that stores calculation parameters and data for each process.

The various programs can be recorded on computer-readable recording media such as HDDs, flash memory, CD-ROMs, DVD-ROMs, and Blu-ray (registered trademark) for widespread distribution. The blending condition setting device 3 can also obtain the various programs via a communications network. The communications network referred to here may be configured using, for example, an existing public line network, LAN, WAN, etc., and may be wired or wireless.

The blending condition setting device 3, which has the above functional configuration, is a computer configured using one or more hardware components such as a CPU, GPU, ASIC, FPGA, etc.

The display device 4 is a display such as a liquid crystal or organic electroluminescence (EL) display, and is electrically connected to the blending condition setting device 3. The display device 4 acquires and displays display data output from the blending condition setting device 3 under the control of the display control unit 331. The display device 4 may also have an audio output function such as a speaker.

The input device 5 accepts input of various information, including information such as set mechanical property values related to the process of setting blending conditions, and outputs the accepted information to the learning device 2 and blending condition setting device 3. The input device 5 is configured using a user interface such as a keyboard, mouse, microphone, and touch panel.

Next, the flow of the learning process performed by the learning device 2 will be described with reference to FIG. 6. FIG. 6 is a flowchart showing an overview of the learning process performed by a learning device according to one embodiment of the present invention. First, the learning device 2 generates a dataset to be used for learning by referencing the storage unit 24 (step S11). Here, the learning data generation unit 21 generates a dataset that pairs the blending conditions, including the types, physical properties, and blending ratios of the raw materials in the resin film, with the physical properties of the resin film manufactured under those blending conditions. This dataset is used as learning data for generating a trained model. The learning data generation unit 21 generates learning data for each resin film.

The learning unit 22 generates a trained model by performing machine learning using the training data generated in step S11 (step S12). The learning unit 22 performs learning using the compounding conditions, including the physical properties and compounding ratio of the thermoplastic resin that constitutes the resin film, as explanatory variables and the physical properties of the resin film as objective variables, to generate a trained model.

Then, the blending condition setting device 3 sets blending conditions consisting of the optimal raw material types and blending ratios for the resin film that are suited to the physical properties being set. Figure 7 is a flowchart showing an overview of the blending condition setting process performed by the blending condition setting device according to one embodiment of the present invention.

The setting unit 31 acquires the trained model from the learning device 2 and inputs the physical properties of the target resin film to the search unit 32 (step S21). The search unit 32 receives the set physical properties treated as the target variables during learning and outputs the blending conditions. When the physical properties of the target resin film are input to the search unit in this way, blending conditions (prediction conditions) suitable for those properties are output. The target physical properties input at this time are based on, for example, information input by the user via the input unit of the blending condition setting device 3.
For example, a blending condition using raw materials A and B, where the blending ratio is A:B = XX (%): △△% (%), is set and output. At this time, the proportion of recycled resin to the total amount of raw materials (recycle rate) may also be displayed. The trained model may output the blending conditions in the form of digits or images, and the setting unit 31 or the like may convert them to represent the blending conditions.

The setting unit 31 sets the blending conditions based on the prediction results of the search unit 32 (step S22). In this embodiment, when there are multiple blending conditions for the target physical property, the setting unit 31 may set a single optimal blending condition that is closest to the target physical property, may set a blending condition that satisfies the target physical property or its allowable setting range, or may select and set the top few blending conditions that satisfy the target physical property. For example, when priorities are set for each type of physical property, the blending conditions are preferentially selected based on the physical property with the highest priority and the condition that is closest to the target physical property.

When the display control unit 331 acquires the setting results from the setting unit 31, it outputs the setting results to the display device 4 and performs display control to display the setting results on the display device 4. The display device 4 displays the blending conditions, including the types of raw materials for the optimal resin film that are predicted to exhibit the desired physical properties and their blending ratios.

Then, for example, raw materials are prepared according to the blending conditions set by the setting method described above, and a resin film is produced using the manufacturing apparatus 200 described above.

In the embodiment described above, the compounding conditions for a resin film with the desired physical properties are set by inputting the desired physical properties into a trained model generated by training using the type and physical properties of raw materials and their blending ratios as explanatory variables and the desired physical properties as the objective variable. This embodiment allows the compounding conditions for a resin film to be set efficiently.

Furthermore, in the embodiment, blending conditions can be searched for that include recycled resins in addition to new resins as raw materials for the resin film, and blending conditions that include the recycled resins can be set, thereby improving the recycling rate of raw materials in manufacturing. In this case, when selecting raw materials, the priority of recycled resins can be set higher than the priority of new resins.

In the embodiments, the objective variables are described as the physical properties of the resin film, and the explanatory variables are the physical properties and blending conditions of the raw materials. However, in addition to these, the objective variables may also include spectral intensity, process condition values during manufacturing (e.g., stretching ratio, etc.), and measurement condition values.

In this case, the spectral intensity refers to the intensity of the spectrum at each wavelength obtained by spectral measurement of the resin film, and examples include spectral measurements taken using light sources in each wavelength range, such as infrared light, near-infrared light, visible light, ultraviolet light, and X-rays.

Process condition values refer to control parameters for producing resin films, and include, for example, manufacturing conditions such as the temperature when using biaxial stretching, the resin feed speed (roll rotation speed), and the distance between clips in the film width direction. Furthermore, if isotropy or other properties are specified as the desired physical properties, the need for biaxial stretching may also be included in the formulation conditions.

These may also include values that can be calculated from the physical properties and characteristics specific to each raw material of the resin film and its blending conditions, as well as physical quantities such as molecular weight, glass transition point, and melting point. Here, the physical quantities specific to each raw material can be calculated from the amount of specific functional groups or the number of partial structures contained in each raw material, or from the chemical structure.

In addition, when contributions from the chemical structure of the resin components contained in the resin film are expected, further improvements in the accuracy of the trained model can be expected by using as explanatory variables not only the component blend ratios but also the amount of specific functional groups calculated based on the chemical structural formula of the components, the number of partial structures and related values, or molecular weights that can be calculated from the chemical structure of each component, or physical quantities including physical property values such as glass transition point and melting point.

(Modification)
Next, a modified example of the embodiment will be described with reference to Fig. 8. In the modified example, the compounding condition setting process is different from the above-described embodiment. The processes and system configuration other than the compounding condition setting process are the same as in the embodiment, so the description will be omitted.

Figure 8 is a flowchart showing an overview of the blending condition setting process performed by a blending condition setting device according to a modified embodiment of the present invention. Similar to step S21, the setting unit 31 acquires a trained model from the learning device 2 and inputs the target physical properties to the search unit 32 (step S31). When the physical properties of the target resin film are input to the search unit 32, blending conditions (prediction conditions) appropriate for those physical properties are output.

The setting unit 31 determines whether the combination conditions output by the trained model are multiple (step S32). If the setting unit 31 determines that there are not multiple combination conditions (step S32: No), it proceeds to step S34. On the other hand, if the setting unit 31 determines that there are multiple combination conditions (step S32: Yes), it proceeds to step S33.

In step S33, the setting unit 31 calculates the recycle rate for each blending condition and selects the blending condition with the highest recycle rate. The setting unit 31, for example, calculates the proportion of recycled resin in the raw materials to be blended and sets this as the recycle rate. The proportion is calculated, for example, by adding up the blending rate or mass percentage of recycled resin. After calculating the recycle rate, the setting unit 31 proceeds to step S34.

In step S34, the setting unit 31 sets the compounding conditions. In this modified example, if there is only one compounding condition in step S32, the setting unit 31 sets that compounding condition, and if a compounding condition is selected in step S33, the setting unit 31 sets the selected compounding condition.

Then, when the display control unit 331 acquires the setting results from the setting unit 31, it outputs the setting results to the display device 4 and performs display control to display the setting results on the display device 4. The display device 4 displays the optimal resin film raw material types that are predicted to exhibit the physical properties of the target resin film, as well as the blending conditions, including their blending ratios.

Then, for example, raw materials are prepared according to the blending conditions set by the setting method described above, and a resin film is produced using the manufacturing apparatus 200 described above.

In the modified example described above, as in the embodiment, the compounding conditions for a resin film with the desired physical properties are set using a trained model generated by training using the type and physical properties of the raw materials and their compounding ratio as explanatory variables and the physical properties of the resin film as the target variable. This modified example makes it possible to efficiently set the compounding conditions for a resin film.

In addition, in this modified example, the compounding conditions are set so that a higher proportion or amount of recycled resin is used as a raw material for the resin film, thereby improving the recycling rate of raw materials during manufacturing.

(Other embodiments)
Although the embodiments of the present invention have been described above, the present invention should not be limited to the above-described embodiments. For example, the compounding condition setting device may have the function of a learning unit. In this case, the compounding condition setting device not only sets the compounding conditions to be set, but also sequentially updates the trained model.

REFERENCE SIGNS LIST 1 Blending condition setting system 2 Learning device 3 Blending condition setting device 4 Display device 5 Input device 21 Learning data generation unit 22 Learning unit 23, 33, 202 Control unit 24, 34 Storage unit 31 Setting unit 32 Search unit 200 Manufacturing device 201 Manufacturing unit 331 Display control unit

Claims (4)

The computer
Inputting the physical properties of a target polyester resin film into a search unit including a trained model generated by training using training data in which the physical properties and blending conditions, including the blending ratio, of a polyester resin including a recycled resin constituting a polyester resin film are used as explanatory variables, and the physical properties of the polyester resin film are used as objective variables;
When a plurality of blending conditions are present in the output result of the search unit, a blending condition having a high proportion of recycled resin is selected from among the blending conditions, and the selected blending condition is set.
How to set compounding conditions. On the computer,
a blending condition setting step of inputting the physical properties of the target polyester resin film into a search unit including a trained model generated by learning using learning data in which blending conditions including the physical properties and blending ratio of polyester resins including recycled resins constituting the polyester resin film are used as explanatory variables and the physical properties of the polyester resin film are used as objective variables, and selecting, from among the blending conditions and in which the proportion of recycled resin is high, a blending condition in which the selected blending condition is set;
A compounding condition setting program that executes the above. a search unit that has a trained model generated by training using training data in which the physical properties and blending conditions, including the blending ratio, of the polyester resin, including the recycled resin, that constitutes the polyester resin film are used as explanatory variables, and the physical properties of the polyester resin film are used as objective variables, and that inputs the physical properties of the target polyester resin film into the trained model to search for blending conditions that are suitable for the physical properties;
a setting unit that selects a blending condition having a high proportion of recycled resin from among the blending conditions when a plurality of blending conditions are present in the output result of the search unit, and sets the selected blending condition;
A compounding condition setting device comprising: The blending condition setting device according to claim 3 ;
a film manufacturing device that manufactures a polyester resin film in accordance with the blending conditions of the polyester resin set by the blending condition setting device;
A polyester resin film manufacturing system comprising: