Apparatus and method for designing a multilayer film

Abstract

An apparatus and method for designing a multilayer film are disclosed. The apparatus for designing a multilayer film may perform: modeling a multilayer film to be designed as a single-layer structure having a preset thickness; collecting physical property data about the film corresponding to the single layer structure; reading a value pre-stored in a storage space accessible to a device for designing the multilayer film, and obtaining a feature setting mode for specifying different feature setting modes according to the read value; performing feature setting based on a plurality of physical indicators selected from the physical characteristic data according to a feature setting mode; selecting at least one of a plurality of supervised learning models capable of performing regression analysis as a machine learning model; predicting a dart impact strength of the multilayer film by using a machine learning model learned by using the feature as an independent variable and a dart impact strength of the multilayer film as a target variable; and generating design data on the multilayer film by combining the predicted values on the other characteristics and the predicted value on the falling dart impact strength so as to meet the design requirements of the multilayer film.

Description

Equipment and methods for designing multilayer films Technical Field

[0001] Cross-references to related applications

[0002] This application claims priority and benefit to Korean Patent Application No. 10-2023-0129783, filed with the Korean Intellectual Property Office on September 26, 2023, and Korean Patent Application No. 10-2024-0130232, filed with the Korean Intellectual Property Office on September 25, 2024, the entire contents of which are incorporated herein by reference.

[0003] This disclosure relates to apparatus and methods for designing multilayer films. Background Technology

[0004] Polymer films are thin, flat structures made of polymer materials. Here, the polymer can have a long-chain molecular structure with repeating units called monomers linked within these chains. The thickness of polymer films can be set in various units, down to millimeters (mm). Polymer films offer a variety of advantages, including flexibility, lightweight, cost-effectiveness, and processability. Specifically, polymer films, due to their excellent flexibility, can be manufactured in various shapes and sizes, making them suitable for a wide range of product designs. Furthermore, the lightweight nature of polymer films makes them suitable for products requiring portability, thereby reducing transportation and handling costs. Additionally, polymer films are cost-effective because they can be manufactured using relatively inexpensive raw materials and can be mass-produced. Moreover, because polymer films can be applied to various manufacturing and processing techniques, such as heat treatment, extrusion, and bonding, the shape, size, and thickness of products can be easily adjusted. Additionally, polymer films possess high durability and strength, and some exhibit excellent optical properties, are waterproof, and can be breathable if desired.

[0005] Multilayer membranes can be composed of two or more layers of polymers or other materials. This multilayer structure can improve the overall performance of the membrane by combining the unique properties of each layer. For example, one layer can be used to improve durability and strength, another layer can provide water or moisture resistance, and yet another layer can provide resistance to certain chemicals. In this way, by combining layers with different properties, multilayer membranes can be adapted to a wider range of applications and specific requirements than single-layer membranes. Summary of the Invention

[0006] Technical issues

[0007] The objective is to provide an apparatus and method for designing multilayer membranes that can predict the dart impact strength of a multilayer membrane without having information about the single-layer membranes that constitute the multilayer membrane, and to design the multilayer membrane according to given design requirements.

[0008] Technical solution

[0009] An apparatus for designing multilayer films according to an embodiment is used to predict the physical properties of the multilayer film and design the multilayer film based on the predicted physical properties by executing program code loaded on one or more memory devices by one or more processors, wherein the program code is configured to perform the following operations when executed: modeling the multilayer film to be designed as a single-layer structure with a preset thickness; collecting physical property data about the film corresponding to the single-layer structure; reading values ​​pre-stored in a storage space accessible to the apparatus for designing the multilayer film, and obtaining a feature setting pattern for specifying different feature setting methods based on the read values; performing feature setting based on a plurality of physical indices selected from the physical property data according to the feature setting pattern; selecting at least one as a machine learning model from a plurality of supervised learning models capable of performing regression analysis; predicting the dart impact strength of the multilayer film by using a machine learning model that learns by using features as independent variables and the dart impact strength of the multilayer film as a target variable; and generating design data about the multilayer film by combining the predicted values ​​of other properties and the predicted value of the dart impact strength to meet the design requirements of the multilayer film.

[0010] In some implementations, when the feature setting mode may include a first feature setting mode, performing feature setting may include setting the thickness and at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on multilayer membranes with monolayer structures as features.

[0011] In some implementations, selection as a machine learning model may include selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model as the machine learning model, and the machine learning model selected from the PLS model, lasso regression model, and random forest model may be configured to receive a vector generated from at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain, as well as the thickness, and to learn to predict the dart impact strength of the multilayer film.

[0012] In some implementations, when the feature setting mode may include a second feature setting mode, performing feature setting may include setting the thickness and strain energy calculated based on at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on multilayer membranes with monolayer structures as features.

[0013] In some implementations, selection as a machine learning model may include selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model as the machine learning model, and the machine learning model selected from the PLS model, lasso regression model, and random forest model may be configured to receive a vector generated from the strain energy and thickness, and to learn to predict the dart impact strength of the multilayer film.

[0014] In some implementations, the strain energy may include a first strain energy (W). yield ), and performing feature settings may include calculating the first strain energy (W) according to the following formula. yield ), and the calculated first strain energy (W) yield Set as a feature,

[0015]

[0016] Where, σ y ε represents the yield stress. y σ represents the yield strain. br Let represent the fracture stress, d represent the contact diameter of the falling dart, t represent the thickness of the measured sample, and a represent an experimentally determined constant.

[0017] In some implementations, the strain energy may include a second strain energy (W). neck ), and performing feature settings may include calculating the second strain energy (W) according to the following formula. neck ), and the calculated second strain energy (W) neck Set as a feature,

[0018]

[0019] Where, σ n Represents necking stress, ε n ε represents necking strain. y σ represents the yield strain. br σ represents the fracture stress. y Let represent the yield stress, d represent the contact diameter of the dart, t represent the thickness of the measured sample, and b represent an experimentally determined constant.

[0020] In some implementations, the strain energy may include a third strain energy (W). str-hard ), and performing feature settings may include calculating the third strain energy (W) according to the following formula. str-hard ), and the calculated third strain energy (W) str-hard Set as a feature,

[0021]

[0022] Where, ε br ε represents fracture strain. n σ represents the necking strain. br σ represents the fracture stress. n Let L represent the necking stress, L represent the final tensile propagation length, t represent the thickness of the measured sample, and c represent an experimentally determined constant.

[0023] In some implementations, the strain energy may include a fourth strain energy (W). elast ), and performing feature settings may include calculating the fourth strain energy (W) according to the following formula. elast ), and the calculated fourth strain energy (W) elast Set as a feature,

[0024]

[0025] Where, σ y ε represents the yield stress. y Let represent the yield strain, d represent the dart contact diameter, D represent the diameter of the fixed disk, L represent the final tensile propagation length, t represent the thickness of the measured sample, and e represent an experimentally determined constant.

[0026] In some implementations, generating design data for the multilayer membrane may include: combining predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature (SIT), haze, moisture permeability, and air permeability, with a predicted value of dart impact strength; generating design data based on the combination when the multilayer membrane designed based on the combination of predicted values ​​meets the design requirements; and redesigning the multilayer membrane by changing the combination to another combination when the multilayer membrane designed based on the combination of predicted values ​​does not meet the design requirements.

[0027] A method for designing a multilayer membrane according to an embodiment is executed by a computing device including one or more processors and one or more memory devices to predict the physical properties of the multilayer membrane and design the multilayer membrane based on the predicted physical properties. The method may include: modeling the multilayer membrane to be designed as a single-layer structure with a predetermined thickness; collecting physical property data of the membrane corresponding to the single-layer structure; reading values ​​pre-stored in a device-accessible storage space for designing the multilayer membrane, and obtaining a feature setting pattern for specifying different feature setting methods based on the read values; performing feature setting based on a plurality of physical indices selected from the physical property data according to the feature setting pattern; selecting at least one as a machine learning model from a plurality of supervised learning models capable of performing regression analysis; predicting the dart impact strength of the multilayer membrane using a machine learning model that learns by using features as independent variables and the dart impact strength of the multilayer membrane as a target variable; and generating design data for the multilayer membrane by combining the predicted values ​​of other properties and the predicted value of the dart impact strength to meet the design requirements of the multilayer membrane.

[0028] In some implementations, when the feature setting mode may include a first feature setting mode, performing feature setting may include setting the thickness and at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on multilayer membranes with monolayer structures as features.

[0029] In some implementations, selection as a machine learning model may include selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model as the machine learning model, wherein the machine learning model selected from the PLS model, lasso regression model, and random forest model may be configured to receive a vector generated from at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain, as well as thickness, and to learn to predict the dart impact strength of the multilayer film.

[0030] In some implementations, when the feature setting mode may include a second feature setting mode, performing feature setting may include setting the thickness and strain energy calculated based on at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on multilayer membranes with monolayer structures as features.

[0031] In some implementations, selection as a machine learning model may include selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model as the machine learning model, and the machine learning model selected from the PLS model, lasso regression model, and random forest model may be configured to receive a vector generated from the strain energy and thickness, and to learn to predict the dart impact strength of the multilayer film.

[0032] In some implementations, the strain energy may include a first strain energy (W). yield ), and performing feature settings may include calculating the first strain energy (W) according to the following formula. yield ), and the calculated first strain energy (W) yield Set as a feature,

[0033]

[0034] Where, σ y ε represents the yield stress. y σ represents the yield strain. br Let represent the fracture stress, d represent the contact diameter of the falling dart, t represent the thickness of the measured sample, and a represent an experimentally determined constant.

[0035] In some implementations, the strain energy may include a second strain energy (W). neck ), and performing feature settings may include calculating the second strain energy (W) according to the following formula. neck ), and the calculated second strain energy (W) neck Set as a feature,

[0036]

[0037] Where, σ n Represents necking stress, ε n ε represents necking strain. y σ represents the yield strain. br σ represents the fracture stress. y Let represent the yield stress, d represent the contact diameter of the dart, t represent the thickness of the measured sample, and b represent an experimentally determined constant.

[0038] In some implementations, the strain energy may include a third strain energy (W). str-hard ), and performing feature settings may include calculating the third strain energy (W) according to the following formula. str-hard ), and the calculated third strain energy (W) str-hard Set as a feature,

[0039]

[0040] Where, ε br ε represents fracture strain. n σ represents the necking strain. br σ represents the fracture stress. n Let L represent the necking stress, L represent the final tensile propagation length, t represent the thickness of the measured sample, and c represent an experimentally determined constant.

[0041] In some implementations, the strain energy may include a fourth strain energy (W). elast ), and performing feature settings may include calculating the fourth strain energy (W) according to the following formula. elast ), and the calculated fourth strain energy (W) elast Set as a feature,

[0042]

[0043] Where, σ y ε represents the yield stress. y Let represent the yield strain, d represent the dart contact diameter, D represent the diameter of the fixed disk, L represent the final tensile propagation length, t represent the thickness of the measured sample, and e represent an experimentally determined constant.

[0044] In some implementations, generating design data for the multilayer membrane may include: combining predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature (SIT), haze, moisture permeability, and air permeability, with a predicted value of dart impact strength; generating design data based on the combination when the multilayer membrane designed based on the combination of predicted values ​​meets the design requirements; and redesigning the multilayer membrane by changing the combination to another combination when the multilayer membrane designed based on the combination of predicted values ​​does not meet the design requirements.

[0045] Beneficial effects

[0046] According to the implementation method, a multilayer membrane that can meet a given design requirement can be designed by predicting the dart impact strength of the multilayer membrane without having information about the single-layer membranes that constitute the multilayer membrane, and by taking into account the prediction results for other properties together. Attached Figure Description

[0047] Figure 1 is a block diagram illustrating an apparatus for designing multilayer films according to an embodiment.

[0048] Figure 2 is a graph illustrating the first stress-strain data for the multilayer film according to an embodiment.

[0049] Figure 3 is a graph illustrating the second stress-strain data for the multilayer film according to an embodiment.

[0050] Figures 4 and 5 are diagrams illustrating environmental data for measuring dart impact intensity according to the embodiments.

[0051] Figure 6 is a flowchart illustrating a method for designing a multilayer film according to an embodiment.

[0052] Figure 7 is a diagram illustrating a computing device according to an embodiment. Detailed Implementation

[0053] In the following description, exemplary embodiments of the present disclosure will be described more fully with reference to the accompanying drawings to facilitate practice by those skilled in the art to which this disclosure pertains. As will be appreciated by those skilled in the art, modifications can be made to the described embodiments in various different ways without departing from the spirit or scope of this disclosure. Furthermore, the drawings and description are to be considered illustrative rather than restrictive in nature, and similar reference numerals denote similar elements throughout the specification.

[0054] Throughout the specification and claims, unless explicitly stated otherwise, the words “comprising” and variations such as “including” or “having” shall be understood to mean including the stated element, but not excluding any other element. Terms including serial numbers such as first, second, etc., may be used to describe various components, but the components are not limited to those terms. The above terms are used only for the purpose of distinguishing one component from others.

[0055] Terms such as “…unit,” “…device,” and “module” used in this specification may refer to a unit capable of performing at least one function or operation described in the specification, which may be implemented as hardware or circuitry, software, or a combination of hardware or circuitry and software. Furthermore, at least some components or functions of the design apparatus and method for multilayer films according to the exemplary embodiments described below may be implemented as programs or software, and such programs or software may be stored on a computer-readable medium.

[0056] Figure 1 is a block diagram illustrating an apparatus for designing a multilayer film according to an embodiment. Figure 2 is a graph illustrating first stress-strain data for the multilayer film according to an embodiment. Figure 3 is a graph illustrating second stress-strain data for the multilayer film according to an embodiment. Figures 4 and 5 are graphs illustrating environmental data for measuring dart impact strength according to an embodiment.

[0057] Referring to FIG1, the apparatus 10 for designing multilayer films according to an embodiment can execute program code or instructions loaded on one or more memory devices by one or more processors. For example, the apparatus 10 for designing multilayer films can be implemented as a computing device 50, which will be described later with reference to FIG7. In this case, one or more processors may correspond to processor 510 of computing device 50, and one or more memory devices may correspond to memory 520 of computing device 50. The program code or instructions can be executed by one or more processors to predict the dart impact strength of the multilayer film without having information about the single-layer films constituting the multilayer film, and to design the multilayer film according to given design requirements. In this specification, the term "module" is used to logically separate these functions executed by the program code or instructions.

[0058] Dart impact strength can refer to the strength of a physical impact that a material (in this case, a membrane) can withstand. Dart impact strength can be used as an indicator to measure, for example, the degree of damage to a membrane when a heavy object is dropped from a predetermined height. The device 10 for designing multilayer membranes may include all or at least some of the following: a data collection module 110, a feature setting module 120, a model selection module 130, a learning module 140, a prediction module 150, and a design data generation module 160. For example, the learning module 150 may be implemented within the device 10 for designing multilayer membranes. Alternatively, for example, the learning module 140 may be implemented externally to the device 10 for designing multilayer membranes, and the device 10 for designing multilayer membranes may be provided with a trained machine learning model from the learning module 140.

[0059] A multilayer membrane can include multiple monolayer membranes. Here, the multiple monolayer membranes can include specific types and different types of monolayer membranes, and the unique properties of the monolayer membranes can vary depending on the type. Additionally, the lamination order of the monolayer membranes within the multilayer membrane can also be changed. The apparatus 10 for designing the multilayer membrane can predict the dart impact strength of the multilayer membrane using a feature-based machine learning model, which is set based on physical property data collected by the data collection module 110 described below. In this case, the apparatus 10 for designing the multilayer membrane can model the multilayer membrane to be designed as a monolayer structure with a preset thickness, rather than modeling it as a form with multiple monolayer membranes laminated. Here, the preset thickness can be selected, for example, the thickness of the final multilayer membrane to be manufactured (i.e., the final product). By modeling the multilayer membrane as a monolayer structure, the dart impact strength can be predicted based solely on data about the multilayer membrane in the form of the final product, without needing to experiment on each monolayer membrane or obtain information about each monolayer membrane.

[0060] When predicting dart impact strength, there are cases where the physical properties of the multilayer film itself, such as those of the final product, are particularly important, and on the other hand, there are cases where it is necessary to fix the physical properties of specific single-layer films among the single-layer films forming the multilayer film. In the latter case, in order for some single-layer films within the multilayer film to meet the physical properties required by the product, the composition of these films must be fixed, and for example, when the multilayer film is used for food, it is important to consider the properties of the single-layer films responsible for functions such as moisture permeability or oxygen permeability individually. In such cases, it can be advantageous to set features applied to a machine learning model for predicting dart impact strength based on data about the physical properties of the single-layer films forming the multilayer film. However, in the former case, such as for heavy products or agricultural products where the mechanical properties of the final film are important, mechanical properties such as stiffness, strength, and tear strength can also be important factors in addition to impact strength. Therefore, in this case, it is effective to set features applied to a machine learning model for predicting dart impact strength based on data about the physical properties of the multilayer film itself, which is more suitable for analytical purposes and uses fewer features compared to the case based on data about the physical properties of the single-layer films forming the multilayer film.

[0061] The data collection module 110 can collect physical property data about the membrane corresponding to the single-layer structure. That is, the data collection module 110 can collect physical property data by assuming that the multilayer membrane itself is a single layer, without considering the type of single-layer membrane forming the multilayer membrane or the lamination order of the single-layer membrane.

[0062] In some embodiments, the data collection module 110 can collect stress-strain data (hereinafter, first stress-strain data) about the multilayer film as physical property data. The first stress-strain data can represent the mechanical properties of a single-layer structure (i.e., a multilayer film considered as a single layer), and together with FIG2, the first stress-strain data can be represented in the form of a stress-strain curve. The stress-strain curve can represent the stress generated when a force is applied to the multilayer film and how the multilayer film deforms as a result.

[0063] In some implementations, the first stress-strain data may include at least one of yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus. Here, the yield point represents the point where the material moves from an elastic deformation region to an inelastic deformation region, and beyond this yield point, even if the stress applied to the material is removed, the material may not fully recover its original shape. The yield stress is the stress value at the yield point, and a higher value may indicate that the material can withstand greater stress before deformation. Resilience is the area below the stress-strain curve up to the yield point and can be a measure of how much the material can recover its original shape after the stress is removed following deformation by receiving stress. Strain hardening refers to the phenomenon where a material strengthens with increasing strain to withstand additional stress beyond the yield point, while failure indicates the point where the material can no longer withstand stress and fractures. The strain hardening modulus is expressed as the slope of the stress-strain curve in the strain hardening region and can be used as an indicator of the rate of material strengthening. Flow energy can be represented by the area under the stress-strain curve from the yield point to strain hardening, and strain hardening energy can be represented by the area under the stress-strain curve from strain hardening to failure. Toughness is the total area under the stress-strain curve, indicating how much energy a material can absorb, and the 1% modulus and 2% modulus indicate the stress at 1% and 2% strain, respectively. The values ​​of the 1% modulus and 2% modulus can be used as measurements indicating strength within the elastic range of the material.

[0064] In some embodiments, the data collection module 110 can collect data on the dart impact strength of the multilayer film by means of a test according to Method A of ASTM D1709. Here, the test can be used to evaluate the dart impact strength or toughness of the film by using a free-falling dart. More detailed information about ASTM D1709 can be found in the ASTM standard document, and its description is not included in this specification.

[0065] In some embodiments, the data collection module 110 can acquire stress-strain curves as data based on the results obtained by performing uniaxial tensile tests on the multilayer film in the machine direction (MD) and transverse direction (TD). Here, MD can be the direction in which the film moves between the rollers of the manufacturing machine during the film manufacturing process, and TD can be a direction perpendicular to the rollers of the manufacturing machine, perpendicular to MD. Alternatively, the uniaxial tensile test can be performed by measuring the resistance of the multilayer film when stretched in a uniaxial direction. In this case, the data collection module 120 can collect data on the stress-strain curves by means of a test according to ASTM D882. More detailed information about ASTM D882 can be found in the ASTM standard document, and its description is not included in this specification.

[0066] In some other embodiments, the data collection module 110 can collect stress-strain data (hereinafter, second stress-strain data) obtained by universal testing machine (UTM) experiments on the multilayer film as physical property data. The UTM experiment can be designed to perform various mechanical tests on the material, and the load value can be measured when a load due to a force load is generated by applying work to the specimen at specific times, forces, and directions. That is, experiments such as compression, tension, and bending can be performed by applying force to the specimen and measuring the load value until the specimen deforms. The data collection module 110 can collect the results obtained by UTM experiments on the multilayer film specimen as stress-strain data.

[0067] In some embodiments, the second stress-strain data may include at least one of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain. Here, yield stress may represent the stress at the yield point, which is the point where the material exceeds its elastic limit and begins permanent deformation, i.e., the point where the material moves from an elastic deformation region to an inelastic deformation region, and yield strain may represent the strain corresponding to the yield stress. Necking stress may represent the stress at the necking point, which is the point where the cross-section of the material begins to locally narrow, and typically after the necking point, the load-bearing capacity of the material may decrease sharply. Necking strain may represent the strain at the necking point. Fracture stress may represent the stress at the point where the material finally fractures, and at the fracture point, the material may no longer bear a load. Fracture strain may represent the strain corresponding to the fracture stress.

[0068] Referring to Figure 3, on the stress-strain curve, the example point P corresponding to the yield stress and yield strain can be determined. y Example point P corresponding to necking stress and necking strain n And example point P corresponding to fracture stress and fracture strain. br .

[0069] The feature setting module 120 can perform feature setting based on multiple physical indicators selected from physical characteristic data collected by the data collection module 110, according to multiple feature setting modes. The multiple feature setting modes are stored as predetermined values ​​in processor-accessible storage space of the device 10 for designing multilayer films and managed therein. The device 10 for designing multilayer films can read values ​​pre-stored in this storage space and determine the feature setting mode used to specify different feature setting methods based on the read values.

[0070] In machine learning, features can be input variables used to predict a target (i.e., the physical properties of a multilayer membrane) and can refer to individual independent variables of the data. Features can provide information about the target variable or dependent variable to be predicted within the dataset and have a significant impact on the performance of the machine learning model 20. The learning module 150 can perform learning of the machine learning model 20 by using features. Here, features can constitute part of the data input to the machine learning model 20, and the machine learning model 20 can learn patterns and perform predictions based on this data. In particular, feature engineering, such as selecting or transforming features, can be an important process for improving the performance of the machine learning model 20. That is, selecting appropriate features or appropriately transforming features can have a significant impact on the high performance of the machine learning model 20.

[0071] In some implementations, when the feature setting mode includes a first feature setting mode, the feature setting module 120 can perform feature setting based on a plurality of physical parameters selected from the first stress-strain data.

[0072] In some embodiments, the feature setting module 120 may set the following as features: the thickness of the multilayer film; and at least some of the following: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus. That is, the feature setting module 120 may set the following as features: only a portion of the first stress-strain data collected by the data collection module 110; and the thickness of the multilayer film. Then, the learning module 140 may perform learning of the machine learning model 20 based on the features including only a portion of the first stress-strain data and the thickness of the multilayer film. This embodiment can be employed in environments with limited computational resources for performing the learning and prediction of the machine learning model 20.

[0073] In some embodiments, the feature setting module 120 may set the following as features: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus; and the thickness of the multilayer film. That is, the feature setting module 120 may set the following as features: all of the first stress-strain data collected by the data collection module 110; and the thickness of the multilayer film. Then, the learning module 140 may perform learning of the machine learning model 20 based on the features including all of the first stress-strain data and the thickness of the multilayer film. This embodiment can be employed in an environment with sufficient computational resources for performing the learning and prediction of the machine learning model 20.

[0074] In some embodiments, the first stress-strain data may include stress-strain data in a first direction and stress-strain data in a second direction. The stress-strain data in the first direction may represent the mechanical properties of the multilayer film in the first direction, and the stress-strain data in the second direction may represent the mechanical properties of the multilayer film in a second direction perpendicular to the first direction. Specifically, the first direction may be the machine direction (MD), and the second direction may be the transverse direction (TD). The MD may be the direction in which the film moves between the rollers of the manufacturing machine during the film manufacturing process, and the TD may be a direction perpendicular to the rollers of the manufacturing machine, which is perpendicular to the MD. The feature setting module 120 may set a plurality of physical parameters selected from at least one of the stress-strain data in the first direction and the stress-strain data in the second direction as features.

[0075] In some embodiments, the feature setting module 120 may set the following as features: at least some of the following in a first direction: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus; at least some of the following in a second direction: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus; and the thickness of the multilayer film. That is, the feature setting module 120 may set the following as features: only a portion of the first stress-strain data in the first direction collected by the data collection module 120, only a portion of the first stress-strain data in the second direction, and the thickness of the multilayer film.

[0076] In some embodiments, the feature setting module 120 may set the following as features: all of the following in a first direction: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus; all of the following in a second direction: yield point, yield stress, resilience, strain hardening, failure, strain hardening modulus, flow energy, strain hardening energy, toughness, 1% modulus, and 2% modulus; and the thickness of the multilayer film. That is, the feature setting module 120 may set the following as features: all of the first stress-strain data in the first direction collected by the data collection module 120, all of the first stress-strain data in the second direction, and the thickness of the multilayer film.

[0077] In some other implementations, when the feature setting mode includes a second feature setting mode, the feature setting module 120 can perform feature setting based on a plurality of physical parameters selected from the second stress-strain data.

[0078] In some embodiments, the feature setting module 120 may set the following as features: stress-strain data obtained experimentally from a multilayer membrane with a single-layer structure (i.e., at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from second stress-strain data), and the thickness of the multilayer membrane. The learning module 140 may then perform machine learning model 20 learning based on features including only a portion of the second stress-strain data and the thickness of the multilayer membrane. This embodiment can be employed in environments with limited computational resources for performing the learning and prediction of the machine learning model 20.

[0079] In some embodiments, the feature setting module 120 may set the following as features: the thickness of the multilayer film, and stress-strain data obtained through experiments on a single-layer multilayer film (i.e., all of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from the second stress-strain data). The learning module 140 may then perform machine learning model 20 learning based on the features including all of the second stress-strain data and the thickness of the multilayer film. This embodiment can be employed in an environment with sufficient computational resources for performing the learning and prediction of the machine learning model 20.

[0080] In some embodiments, the feature setting module 120 may set the following as features: at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain in a first direction; at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain in a second direction; and the thickness of the multilayer film. Then, the learning module 140 may perform machine learning model 20 learning based on features including only a portion of the second stress-strain data in the first direction, only a portion of the second stress-strain data in the second direction, and the thickness of the multilayer film. This embodiment can be employed in environments with limited computational resources for performing the learning and prediction of the machine learning model 20.

[0081] In some embodiments, the feature setting module 120 may set the following as features: all of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain in a first direction; all of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain in a second direction; and the thickness of the multilayer film. Then, the learning module 140 may perform machine learning model 20 learning based on the features including all of the second stress-strain data in the first direction, all of the second stress-strain data in the second direction, and the thickness of the multilayer film. This embodiment can be employed in an environment with sufficient computational resources for performing the learning and prediction of the machine learning model 20.

[0082] In some other embodiments, when the feature setting mode includes a third feature setting mode, the feature setting module 120 can calculate the strain energy based on a plurality of physical parameters selected from the second stress-strain data and set the strain energy as a feature.

[0083] In some embodiments, the feature setting module 120 may set the following as features: strain energy calculated from at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from second stress-strain data (i.e., stress-strain data obtained through experiments on multilayer membranes with monolayer structures); and the thickness of the multilayer membrane. The learning module 140 may then perform learning of the machine learning model 20 based on features including the strain energy calculated from only a portion of the second stress-strain data and the thickness of the multilayer membrane. This embodiment can be employed in environments with limited computational resources for performing the learning and prediction of the machine learning model 20.

[0084] In some embodiments, the feature setting module 120 may set the following as features: the thickness of the multilayer film, and all strain energies selected from the second stress-strain data (i.e., stress-strain data obtained through experiments on a multilayer film with a single-layer structure), including yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain. The learning module 140 may then perform learning of the machine learning model 20 based on the strain energy calculated according to all the features in the second stress-strain data and the thickness of the multilayer film. This embodiment can be employed in an environment with sufficient computational resources for performing the learning and prediction of the machine learning model 20.

[0085] In some implementations, the feature setting module 120 can calculate the first strain energy (W) based on the second stress-strain data. yield ) and set it as a feature. Specifically, the feature setting module 120 can calculate the first strain energy (W) according to the following formula 1. yield ).

[0086] Formula 1

[0087]

[0088] Here, σ y It can represent the yield stress, ε, in the second stress-strain data. y It can represent the yield strain in the second stress-strain data, and σ in the stress-strain data br This can represent the fracture stress in the second stress-strain data. These variables can correspond to experimental data measured by methods other than dart impact testing (e.g., UTM testing).

[0089] On the other hand, in order to calculate the strain energy based on the second stress-strain data, variables related to the environmental settings of the dart impact test according to a specific predetermined standard can be additionally applied to the multilayer membrane whose dart impact strength is to be predicted. Here, the standard can refer to the American Society for Testing and Materials (ASTM) standard, and can refer to any other standard customized considering the learning performance of the machine learning model 20. Specifically, as variables related to the environmental settings of the dart impact test, d can represent the dart contact diameter, t can represent the thickness of the measured sample, and a can represent an experimentally determined constant considering the experimental environment or product test environment. Referring together to Figure 4, the engagement diameter d can refer to the diameter of the circular area engaged with the additional membrane when the weight is dropped from a predetermined height. The weight includes a fixed disc with a diameter D, the area represented by l is the area on which the tension propagates, and the area represented by L represents the area (or length) on which the tension ultimately propagates. The variables (d, D, l, L, t, etc.) related to the environmental settings of the individual dart impact test can be appropriately adjusted to improve the accuracy of the final predicted dart impact strength.

[0090] In some embodiments, the feature setting module 120 can calculate the second strain energy (W) based on the second stress-strain data. neck ) and set it as a feature. Specifically, the feature setting module 120 can calculate the second strain energy (W) according to the following formula 2. neck ).

[0091] Formula 2

[0092]

[0093] Here, σ n This can represent the necking stress, ε, in the second stress-strain data. n This can represent the necking strain, ε, in the second stress-strain data. y This can represent the yield strain, σ, in the second stress-strain data. br This can represent the fracture stress in the second stress-strain data, σ. y The yield stress can be represented in the second stress-strain data, and these variables can correspond to experimental data measured by methods other than the dart impact test as described above.

[0094] On the other hand, as a variable related to the environmental settings of the dart impact test, d can be the dart contact diameter. t can represent the thickness of the measured sample, and b can represent an experimentally determined constant considering the experimental environment or product testing environment.

[0095] In some implementations, the feature setting module 120 can calculate the third strain energy (W) based on the second stress-strain data. str-hard ) and set it as a feature. Specifically, the feature setting module 120 can calculate the third strain energy (W) according to the following formula 3. str-hard ).

[0096] Formula 3

[0097]

[0098] Here, ε br It can represent the fracture strain in the second stress-strain data, ε n This can represent the necking strain in the second stress-strain data, σ. br It can represent the fracture stress in the second stress-strain data, and σ n This can represent the necking stress in the second stress-strain data. These variables can correspond to experimental data measured by methods other than the dart impact test described above.

[0099] On the other hand, as variables related to the environmental settings of the dart impact test, L can represent the final tensile propagation length, t can represent the thickness of the measured sample, and c can represent a constant determined by the experiment considering the experimental environment or product test environment.

[0100] In some implementations, the feature setting module 120 can calculate the fourth strain energy (W) based on the second stress-strain data. elast ) and set it as a feature. Specifically, the feature setting module 120 can calculate the fourth strain energy (W) according to the following formula 4. elast ).

[0101] Formula 4

[0102]

[0103] Here, σ y It can represent the yield stress in the second stress-strain data, and ε y This can represent the yield strain in the second stress-strain data. These variables can correspond to experimental data measured by methods other than the dart impact test described above.

[0104] On the other hand, as variables related to the environmental settings of the dart impact test, d can represent the dart contact diameter, D can represent the diameter of the fixed plate, L can represent the final tensile propagation length, t can represent the thickness of the measured sample, and e can represent a predetermined constant.

[0105] In some implementations, the final tensile propagation length L can be calculated based on experimental data measured by methods other than dart impact testing. Specifically, the final tensile propagation length L can be calculated according to the following formula 5.

[0106] Formula 5

[0107]

[0108] Here, σ br This can represent the fracture stress in the second stress-strain data, σ. y It can represent the yield stress in the second stress-strain data, and d can represent the contact diameter of the falling dart.

[0109] In some implementations, the feature setting module 120 may set the following together as a feature: first strain energy (W) yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (W elast The learning module 140 can perform learning of the machine learning model 20 based on the features configured in this way. This implementation can be employed in an environment with sufficient computing resources for performing the learning and prediction of the machine learning model 20.

[0110] Alternatively, in some other embodiments, the feature setting module 120 may set the following as a feature: first strain energy (W) yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (W elast At least some of the features, and the thickness of the multilayer film. The learning module 140 can perform learning of the machine learning model 20 based on the features configured in this way. This implementation can be employed in environments with limited computational resources for performing the learning and prediction of the machine learning model 20.

[0111] In some implementations, the feature setting module 120 may set the following together as a feature: a first strain energy (W) in a first direction. yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (W elast All of the above; the first strain energy (W) in the second direction. yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (Welast The learning module 140 can perform learning of the machine learning model 20 based on the features configured in this way. This implementation can be employed in an environment with sufficient computing resources for performing the learning and prediction of the machine learning model 20.

[0112] Alternatively, in some other embodiments, the feature setting module 120 may set the following as a feature: a first strain energy (W) in a first direction. yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (W elast At least some of them; the first strain energy (W) in the second direction yield ), second strain energy (W) neck ), third strain energy (W) str-hard ) and fourth strain energy (W elast At least some of the features in the model; and the thickness of the multilayer film. The learning module 140 can perform learning of the machine learning model 20 based on the features configured in this way. This implementation can be employed in environments with limited computing resources for performing learning and prediction of the machine learning model 20.

[0113] The model selection module 130 can select a machine learning model for predicting the physical properties (i.e., dart impact strength) of multilayer membranes. Specifically, the model selection module 130 can select at least one machine learning model from a plurality of supervised learning models capable of performing regression analysis as the machine learning model for predicting the dart impact strength of multilayer membranes. Supervised learning is a method of training a model using labeled training data, and in supervised learning, the model can learn the relationship between input variables and target variables, thereby enabling the model to predict the value of the target variable with respect to new input data. In some implementations, the model selection module 130 can select a supervised learning model capable of performing regression analysis from partial least squares (PLS), linear regression, multinomial regression, ridge regression, lasso regression, logistic regression, decision trees, random forests, gradient boosting, and neural networks. PLS models can be used to model the relationship between various independent variables and various dependent variables (or target variables), and independent variables determined to be important for predicting the dependent variable can be identified. In linear regression, the model can learn a linear relationship between input variables and target variables, and in multinomial regression, the model can learn complex relationships by extending linear regression to include higher-order terms of the input variables. Ridge regression and lasso regression can prevent overfitting and adjust model complexity by adding regularization to linear regression. Logistic regression, a classification algorithm used to solve binary classification problems, can also be used for regression analysis because it predicts results as probabilities. Decision trees, random forests, and gradient boosting learn complex nonlinear relationships and can therefore also be used for regression and classification problems. Neural networks can solve various types of regression problems, and in particular, deep learning can be very effective at learning complex nonlinear relationships. Of course, the supervised learning models capable of performing regression analysis that can be selected by the model selection module 130 are not limited to the models listed above, and other types of models not listed can be selected considering specific implementation objectives, implementation environments, required performance, etc.

[0114] The learning module 140 can train the machine learning model 20 by using the features set by the feature setting module 120 as independent variables and the dart impact intensity of the multilayer film as the target variable. Therefore, the machine learning model 20 can learn the relationship between the features and the target variable to be trained as a model capable of predicting the dart impact intensity of the multilayer film with respect to new data (i.e., data on the physical properties of the multilayer film). Specifically, the learning module 140 can use different learning methods depending on the model selected by the model selection module 130.

[0115] The prediction module 150 can predict the dart impact intensity of the multilayer membrane using the machine learning model 20 trained by the learning module 140. In other words, the prediction module 150 can feed new data (i.e., data on the physical properties of the multilayer membrane) to the machine learning model 20 and obtain a prediction result on the dart impact intensity of the multilayer membrane.

[0116] More specifically, the model selection module 130 can perform model selection based on multiple model selection modes. These multiple model selection modes are stored as predetermined values ​​in processor-accessible storage space of the device 10 for designing multilayer films and managed therein. The device 10 for designing multilayer films can read the values ​​pre-stored in this storage space and determine the model selection mode for specifying different model selection methods based on the read values.

[0117] In some implementations, when the model selection mode is the first model selection mode, the model selection module 130 can select the PLS model as the machine learning model for predicting the dart impact intensity of the multilayer membrane. As described above, the features set by the feature setting module 120 according to the feature setting mode can, for example, be combined in the form of a single vector and used as input to the PLS model. The dart impact intensity of the multilayer membrane is set as the target variable, and the learning module 150 can train the PLS model in the direction that maximizes the correlation between the input features and the target variable. At this time, the learning module 150 can optimize the "n_components" parameter so that the PLS model correctly interprets the data, and the optimal "n_components" value can be determined through cross-validation. With the PLS model learned in this way, the dart impact intensity of the multilayer membrane can be predicted by inputting features about a new monolayer membrane.

[0118] In some other implementations, when the model selection mode is the second model selection mode, the model selection module 130 can select a lasso regression model as the machine learning model for predicting the dart impact intensity of the multilayer membrane. As described above, the features set by the feature setting module 120 according to the feature setting mode can, for example, be combined in the form of a single vector and used as input to the lasso regression model. The dart impact intensity of the multilayer membrane is set as the target variable, and the learning module 150 can train the lasso regression model through L1 regularization (a regularization that adds the sum of the absolute values ​​of the regression coefficients as a penalty to the target function) to guide the model to select only important features. At this time, the learning module 150 can optimize the "alpha" parameter used to adjust the intensity during model learning, and the optimal "alpha" value can be determined through cross-validation. With the lasso regression model learned in this way, the dart impact intensity of the multilayer membrane can be predicted by inputting features about a new monolayer membrane.

[0119] In some other implementations, when the model selection mode is the third model selection mode, the model selection module 130 can select a random forest model as the machine learning model for predicting the dart impact intensity of the multilayer membrane. As described above, the features set by the feature setting module 120 according to the feature setting mode can, for example, be combined in the form of a single vector and used as input to the random forest model. The dart impact intensity of the multilayer membrane is set as the target variable, and the learning module 150 can be trained by setting "n_estimators", "max_depth", "min_samples_split", etc., as hyperparameters of the random forest model. With the random forest model learned in this way, the dart impact intensity of the multilayer membrane can be predicted by inputting features about a new monolayer membrane.

[0120] In some implementations, a machine learning model selected from PLS models, lasso regression models, and random forest models can be configured to receive vectors generated from the thickness of the multilayer film and at least some or all of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain, and to learn to predict the dart impact strength of the multilayer film.

[0121] In some other implementations, a machine learning model selected from PLS models, lasso regression models, and random forest models can be configured to receive a vector generated from strain energy calculated based on at least some or all of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain, as well as the thickness of the multilayer film, and to learn to predict the dart impact strength of the multilayer film.

[0122] The design data generation module 160 can generate design data for the multilayer membrane by combining predicted values ​​of other properties with predicted values ​​of dart impact strength obtained by the prediction module 150, in order to meet the design requirements of the multilayer membrane. In some embodiments, other properties may include at least one of the following: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature (SIT), haze, moisture permeability, and air permeability of the multilayer membrane.

[0123] The design data generation module 160 can primarily combine predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature, haze, moisture permeability, and air permeability, along with a predicted value of the dart impact strength. Design requirements for the multilayer membrane to be designed may include requirements to ensure that mechanical, thermal, optical, barrier, and chemical properties achieve the desired performance. When the multilayer membrane designed based on the primary combination of predicted values ​​meets the design requirements, the design data generation module 160 can generate design data based on the primary combination. In this case, the design data may include specific predicted values.

[0124] Alternatively, when the multilayer membrane designed based on the primary combination of predicted values ​​does not meet the design requirements, the design data generation module 160 can redesign the multilayer membrane by changing the primary combination to another combination. That is, the design data generation module 160 can secondarily combine predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature, haze, moisture permeability, and air permeability, with predicted values ​​of dart impact strength, to have a combination different from the primary combination. In this case, the predicted values ​​of dart impact strength in the primary combination and the predicted values ​​of dart impact strength in the secondary combination can be different values.

[0125] When a multilayer membrane designed based on a secondary combination of predicted values ​​meets the design requirements, the design data generation module 160 can generate design data based on the secondary combination. Alternatively, when a multilayer membrane designed based on a secondary combination of predicted values ​​also fails to meet the design requirements, the design data generation module 160 can repeat the process of redesigning the multilayer membrane by changing the secondary combination to yet another combination.

[0126] According to this embodiment, when the physical properties of the multilayer film itself, such as those of the final product, are particularly important, compared to the case based on physical property data of the single-layer film forming the multilayer film, by learning and predicting using elements selected from physical property data collected by treating the multilayer film as a single layer as features, it is possible to predict the dart impact strength of the multilayer film that is more suitable for the analysis purpose, using fewer features, and the dart impact strength has high accuracy. Furthermore, by distinguishing between the number of indicators selected as features prioritizing performance and the number of indicators selected as features saving computational resources, flexible design can be performed according to the environment of the equipment used to design the multilayer film and the purpose of implementation. Moreover, by considering the prediction results related to other properties of the multilayer film and the prediction results of the dart impact strength, a multilayer film that can meet given design requirements can be designed.

[0127] Figure 6 is a flowchart illustrating a method for designing a multilayer film according to an embodiment.

[0128] Referring to Figure 6, the method for designing a multilayer membrane according to an embodiment may include: step S601, modeling the multilayer membrane to be designed as a single-layer structure with a preset thickness; step S602, collecting physical property data of the membrane corresponding to the single-layer structure; step S603, reading values ​​pre-stored in a storage space accessible to the device for designing the multilayer membrane, and obtaining a feature setting mode for specifying different feature setting methods based on the read values; step S604, setting features according to the feature setting mode and according to a plurality of physical indices selected from the physical property data; step S605, selecting at least one as a machine learning model from a plurality of supervised learning models capable of performing regression analysis; step S606, training the machine learning model with features as independent variables and the dart impact strength of the multilayer membrane as a target variable; step S607, predicting the dart impact strength of the multilayer membrane using the learned machine learning model; and step S608, generating design data for the multilayer membrane by combining the predicted values ​​of other properties and the predicted value of the dart impact strength, so as to meet the design requirements of the multilayer membrane.

[0129] Further details of the method for designing multilayer films can be found in the description of the features disclosed in the embodiments described herein, and redundant descriptions are not included herein.

[0130] Figure 7 is a diagram illustrating a computing device according to an embodiment.

[0131] Referring to FIG7, the apparatus and method for designing multilayer films according to the embodiments can be implemented using a computing device 50. The computing device 50 can be implemented in various forms of electronic devices, servers or similar devices, and its functions can be implemented through a combination of software and hardware.

[0132] The computing device 50 may include at least one of a processor 510, a memory 530, a user interface input device 540, a user interface output device 550, and a storage device 560, which communicate via a bus 520. The computing device 50 may also include a network interface 570 electrically connected to the network 40. The network interface 570 can send or receive signals with other entities via the network 40.

[0133] The processor 510 can be implemented as various types of computing devices, such as a microcontroller unit (MCU), application processor (AP), central processing unit (CPU), graphics processing unit (GPU), neural processing unit (NPU), quantum processing unit (QPU), etc. The processor 510 is a semiconductor device that executes instructions stored in memory 530 or storage device 560 and can play a core role in the system. The program code and data stored in memory 530 or storage device 560 can command the processor 510 to perform specific tasks, thereby realizing the overall operation of the system. Therefore, the processor 510 can be configured to implement the various functions and methods described above with reference to Figures 1 to 6.

[0134] Memory 530 and storage device 560 may include various types of volatile or non-volatile storage media for storing and accessing system data. For example, memory 530 may include read-only memory (ROM) 531 and random access memory (RAM) 532. In some embodiments, memory 530 may be embedded inside processor 510, and in this case, data transfer speeds between memory 530 and processor 510 can be very fast. In some other embodiments, memory 530 may be located outside processor 510, and in this case, memory 530 may be connected to processor 510 via various data buses or interfaces. This connection can be made through various known means, such as a peripheral component rapid interconnect (PCIe) interface for high-speed data transfer or a memory controller.

[0135] In some embodiments, at least some components and features of the apparatus and method for designing multilayer films according to the embodiments may be implemented as programs or software executed by computing device 50, and such programs or software may be stored in a computer-readable medium. Specifically, the computer-readable medium according to the embodiments may record programs for causing a computer, including a processor 510 that executes programs or instructions stored in memory 530 or storage device 560, to execute the steps included in the apparatus and method for designing multilayer films according to the embodiments.

[0136] In some embodiments, at least some components and features of the apparatus and method for designing multilayer films according to the embodiments may be implemented using the hardware or circuitry of the computing device 50, or may be implemented as a separate hardware or circuitry that can be electrically connected to the computing device 50.

[0137] According to the implementation method, when the physical properties of the multilayer film itself, such as those of the final product, are particularly important, the dart impact strength of the multilayer film can be predicted with high accuracy without information about the single-layer films constituting the multilayer film. That is, compared to the case based on physical property data about the single-layer films forming the multilayer film, by learning and predicting using elements selected from physical property data collected by treating the multilayer film as a single layer as features, the dart impact strength of the multilayer film more suitable for the analysis purpose can be predicted, and fewer features are used. Furthermore, by distinguishing between the number of indicators selected to prioritize performance and the number of indicators selected to save computational resources, flexible design can be performed according to the environment and objectives of the equipment and method used to design the multilayer film. Moreover, by considering the prediction results related to other properties of the multilayer film and the prediction results of the dart impact strength, a multilayer film that meets given design requirements can be designed.

[0138] Although the embodiments of this disclosure have been described in detail above, the scope of this disclosure is not limited thereto, but may include various modifications and alterations made by those skilled in the art using the basic concept of this disclosure as defined in the claims.

Claims

1. An apparatus for designing multilayer films, the apparatus predicting and designing the multilayer films based on the predicted physical properties by executing program code loaded on one or more memory devices by one or more processors. The program code is configured to perform the following operations when executed: model the multilayer membrane to be designed as a single-layer structure with a preset thickness; Collect data on the physical properties of the membrane corresponding to the monolayer structure; Values ​​pre-stored in device-accessible storage space for designing multilayer films are read, and a feature setting pattern for specifying different feature setting methods is obtained based on the read values; feature setting is performed based on multiple physical indicators selected from the physical property data according to the feature setting pattern; at least one of multiple supervised learning models capable of performing regression analysis is selected as a machine learning model; the dart impact intensity of the multilayer film is predicted by using the machine learning model learned by using the features as independent variables and the dart impact intensity of the multilayer film as the target variable; Design data for the multilayer membrane is generated by combining predicted values ​​of other properties with the predicted value of the dart impact strength, in order to meet the design requirements of the multilayer membrane.

2. The device according to claim 1, wherein, When the feature setting mode includes a first feature setting mode, performing the feature setting includes setting the thickness and at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained by experiments on the multilayer membrane of the monolayer structure as the feature.

3. The device according to claim 2, wherein: Selecting the machine learning model includes selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model; and the machine learning model selected from the PLS model, the lasso regression model, and the random forest model is configured to receive a vector generated from at least some of the yield stress, the yield strain, the necking stress, the necking strain, the fracture stress, and the fracture strain, as well as the thickness, and to learn to predict the dart impact strength of the multilayer film.

4. The device according to claim 1, wherein, When the feature setting mode includes a second feature setting mode, performing the feature setting includes setting the thickness and strain energy calculated based on at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on the multilayer membrane of the monolayer structure as the feature.

5. The device according to claim 4, wherein: Selecting the machine learning model includes selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model; and the machine learning model selected from the PLS model, the lasso regression model, and the random forest model is configured to receive a vector generated from the strain energy and the thickness, and to learn to predict the dart impact strength of the multilayer film.

6. The device according to claim 4, wherein, The strain energy includes a first strain energy (W) yield ), and wherein performing the feature setting includes: calculating the first strain energy (W) according to the following formula. yield ), and the calculated first strain energy (W) yield ) is set as the feature, Where, σ y ε represents the yield stress. y σ represents the yield strain. br Let represent the fracture stress, d represent the contact diameter of the falling dart, t represent the thickness of the measured sample, and a represent an experimentally determined constant.

7. The device according to claim 4, wherein, The strain energy includes a second strain energy (W) neck ), and wherein performing the feature setting includes: calculating the second strain energy (W) according to the following formula. neck ), and the calculated second strain energy (W) neck ) is set as the feature, Where, σ n The necking stress, ε n ε represents the necking strain. y σ represents the yield strain. br The fracture stress, σ, represents the stress. y Let d represent the yield stress, d represent the contact diameter of the dart, t represent the thickness of the measured sample, and b represent an experimentally determined constant.

8. The device according to claim 4, wherein, The strain energy includes a third strain energy (W) str-hard ), and wherein performing the feature setting includes: calculating the third strain energy (W) according to the following formula. str-hard ), and the calculated third strain energy (W) str-hard ) is set as the feature, Where, ε br ε represents the fracture strain. n The necking strain, σ, represents the necking strain. br The fracture stress, σ, represents the stress. n Let L represent the necking stress, L represent the final tensile propagation length, t represent the thickness of the measured sample, and c represent an experimentally determined constant.

9. The device according to claim 4, wherein, The strain energy includes a fourth strain energy (W) elast ), and wherein performing the feature setting includes: calculating the fourth strain energy (W) according to the following formula. elast ), and the calculated fourth strain energy (W) elast ) is set as the feature, Where, σ y ε represents the yield stress. y Let d represent the yield strain, d represent the dart contact diameter, D represent the diameter of the fixed disk, L represent the final tensile propagation length, t represent the thickness of the measured sample, and e represent an experimentally determined constant.

10. The device according to claim 1, wherein, Generating the design data for the multilayer membrane includes: combining predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature (SIT), haze, moisture permeability, and air permeability, with a predicted value of the dart impact strength; generating the design data based on the combination of the predicted values ​​when the multilayer membrane designed based on the combination of the predicted values ​​meets the design requirements; and redesigning the multilayer membrane by changing the combination to another combination when the multilayer membrane designed based on the combination of the predicted values ​​does not meet the design requirements.

11. A method for designing a multilayer film, the method being performed by a computing device including one or more processors and one or more memory devices to predict the physical properties of the multilayer film and design the multilayer film based on the predicted physical properties, the method comprising: The multilayer membrane to be designed is modeled as a single-layer structure with a preset thickness; Collect data on the physical properties of the membrane corresponding to the monolayer structure; Values ​​pre-stored in device-accessible storage space for designing multilayer films are read, and a feature setting pattern for specifying different feature setting methods is obtained based on the read values; feature setting is performed based on multiple physical indicators selected from the physical property data according to the feature setting pattern; at least one of multiple supervised learning models capable of performing regression analysis is selected as a machine learning model; the dart impact intensity of the multilayer film is predicted by using the machine learning model learned by using the features as independent variables and the dart impact intensity of the multilayer film as the target variable; Design data for the multilayer membrane is generated by combining predicted values ​​of other properties with the predicted value of the dart impact strength, in order to meet the design requirements of the multilayer membrane.

12. The method according to claim 11, wherein, When the feature setting mode includes a first feature setting mode, performing the feature setting includes setting the thickness and at least some of the yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained by experiments on the multilayer membrane of the monolayer structure as the feature.

13. The method according to claim 12, wherein, Selecting the machine learning model includes selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model as the machine learning model, wherein the machine learning model selected from the PLS model, the lasso regression model, and the random forest model is configured to receive a vector generated from at least some of the yield stress, the yield strain, the necking stress, the necking strain, the fracture stress, and the fracture strain, as well as the thickness, and to learn to predict the dart impact strength of the multilayer film.

14. The method according to claim 11, wherein, When the feature setting mode includes a second feature setting mode, performing the feature setting includes setting the thickness and strain energy calculated based on at least some of yield stress, yield strain, necking stress, necking strain, fracture stress, and fracture strain selected from stress-strain data obtained from experiments on the multilayer membrane of the monolayer structure as the feature.

15. The method according to claim 14, wherein, Selecting the machine learning model includes selecting one of a partial least squares (PLS) model, a lasso regression model, and a random forest model, wherein the machine learning model selected from the PLS model, the lasso regression model, and the random forest model is configured to receive a vector generated from the strain energy and the thickness, and to learn to predict the dart impact strength of the multilayer film.

16. The method of claim 14, wherein, The strain energy includes a first strain energy (W) yield ), and wherein performing the feature setting includes: calculating the first strain energy (W) according to the following formula. yield ), and the calculated first strain energy (W) yield ) is set as the feature, Where, σ y ε represents the yield stress. y σ represents the yield strain. br Let represent the fracture stress, d represent the contact diameter of the falling dart, t represent the thickness of the measured sample, and a represent an experimentally determined constant.

17. The method according to claim 14, wherein, The strain energy includes a second strain energy (W) neck ), and wherein performing the feature setting includes: calculating the second strain energy (W) according to the following formula. neck ), and the calculated second strain energy (W) neck ) is set as the feature, Where, σ n The necking stress, ε n ε represents the necking strain. y σ represents the yield strain. br The fracture stress, σ, represents the stress. y Let d represent the yield stress, d represent the contact diameter of the dart, t represent the thickness of the measured sample, and b represent an experimentally determined constant.

18. The method according to claim 14, wherein, The strain energy includes a third strain energy (W) str-hard ), and wherein performing the feature setting includes: calculating the third strain energy (W) according to the following formula. str-hard ), and the calculated third strain energy (W) str-hard ) is set as the feature, Where, ε br ε represents the fracture strain. n The necking strain, σ, represents the necking strain. br The fracture stress, σ, represents the stress. n Let L represent the necking stress, L represent the final tensile propagation length, t represent the thickness of the measured sample, and c represent an experimentally determined constant.

19. The method of claim 14, wherein, The strain energy includes a fourth strain energy (W) elast ), and wherein performing the feature setting includes: calculating the fourth strain energy (W) according to the following formula. elast ), and the calculated fourth strain energy (W) elast ) is set as the feature, Where, σ y ε represents the yield stress. y Let d represent the yield strain, d represent the dart contact diameter, D represent the diameter of the fixed disk, L represent the final tensile propagation length, t represent the thickness of the measured sample, and e represent an experimentally determined constant.

20. The method according to claim 11, wherein, Generating the design data for the multilayer membrane includes: combining predicted values ​​of at least one of the following characteristics of the multilayer membrane: stiffness, strength, strain, tear strength, high-speed dart impact strength, seal initiation temperature (SIT), haze, moisture permeability, and air permeability, with a predicted value of the dart impact strength; generating the design data based on the combination of the predicted values ​​when the multilayer membrane designed based on the combination of the predicted values ​​meets the design requirements; and redesigning the multilayer membrane by changing the combination to another combination when the multilayer membrane designed based on the combination of the predicted values ​​does not meet the design requirements.

Images