Estimation device, model learning device, estimation method, model learning method, and program

Abstract

To enable suitable conditions for substrate processing to be estimated.SOLUTION: An estimation device 1 includes: a first input unit 104 that receives substrate information representing the properties of a pre-processing substrate; and an estimation unit 106 that estimates estimated performance information representing the performance of the processed substrate, into which the pre-processing substrate has been processed, based on the substrate information representing the properties of the pre-processing substrate, and one of a plurality of processing information related to the processing conditions of the substrate.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an estimation device, a model learning device, an estimation method, a model learning method, and a program. [Background technology]

[0002] It is known that latex such as Saran latex is coated on a substrate such as paper, plastic, cloth, etc. For example, Patent Document 1 describes a laminate having a vinyl halide polymer layer on a paper substrate, the vinyl halide polymer layer containing the vinyl halide polymer and having a specific range of coating efficiency. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2022 / 059472 Summary of the Invention [Problem to be solved by the invention]

[0004] However, with the technique described in Patent Document 1, it was difficult to know what kind of processing should be performed on what kind of substrate in order to obtain a coated substrate that satisfies the desired performance.

[0005] In view of the above circumstances, an object of the present disclosure is to provide an estimation device, a model learning device, an estimation method, a model learning method, and a program capable of estimating appropriate conditions for processing a substrate. [Means for solving the problem]

[0006] That is, the present invention is as follows. [1] A first input unit that accepts input of substrate information indicating characteristics of the substrate before processing; an estimation unit that estimates estimated performance information indicating the performance of a processed substrate obtained by processing the substrate before processing, based on substrate information indicating the characteristics of the substrate before processing and any one of a plurality of processing information related to processing conditions of the substrate; An estimation device comprising: [2] The method further includes a model that learns a relationship between substrate information indicating a characteristic of a substrate, the processing information, and performance information indicating a performance of a processed substrate obtained by processing the substrate, and outputs the performance information when the substrate information and the processing information are input, The estimation unit inputs substrate information indicating characteristics of the substrate before processing and any one of the multiple pieces of processing information to the model, and outputs the estimated performance information. [3] The first input unit further receives input of target performance information indicating a target performance of a processed substrate obtained by processing the substrate before processing, The estimation device according to [1] or [2], further comprising a recommendation unit that outputs recommended information including at least one of recommended substrate information indicating recommended characteristics of the substrate and recommended processing information indicating recommended processing conditions for the substrate, based on the target performance information and the estimated performance information. [4] The recommendation section A generation unit that generates candidate information including at least one of the base material information and the processing information; an acquisition unit that acquires the estimated performance information from the estimation unit based on the candidate information; a determination unit that determines whether or not new candidate information is to be generated based on a search end condition; The estimation device according to [3], [5] the determination unit determines whether or not a condition, which is the search end condition, is satisfied, the condition being that the performance indicated by the estimated performance information satisfies the target performance, or the number of times that candidate information has been generated is a predetermined upper limit number of times; a display unit that displays the recommendation information when it is determined that the search end condition is satisfied; The estimation device according to [4], comprising: [6] The first input unit further receives an input of a range of conditions related to the processing, The estimation device according to [3], wherein the recommendation unit outputs the recommendation information based on search range information according to a range of conditions related to the processing. [7] The first input unit further receives an input of a country or field in which the processed base material is used, The estimation device described in [6], further comprising a search range generation unit that generates the search range information based on substrate information indicating characteristics of the substrate before processing and processing information corresponding to the country or the field among the multiple pieces of processing information. [8] The estimation device according to any one of [1] to [7], wherein the processing information includes a type of polymer to be formulated for the substrate, a type of additive to be added to the polymer, and coating conditions for coating the polymer on the substrate. [9] The estimation device according to any one of [1] to [8], wherein the estimation unit has a feature calculation unit that calculates a feature representing one of the features of a base material indicated by the base material information and a processing indicated by the processing information.

[10] The estimation device according to any one of [1] to [9], wherein the first input unit further receives barrier target substance information indicating a substance that is a barrier to the processed base material.

[11] The estimation unit includes a feature amount calculation unit that calculates a feature amount representing a feature of either a base material indicated by the base material information or a processing indicated by the processing information, The estimation device described in

[10] , wherein the feature calculation unit uses at least one of the substrate information and the processing information, and the barrier target substance information to calculate a first feature indicating a characteristic of at least one of the substrate information and the processing information, and the barrier target substance information.

[12] The estimation unit estimates barrier performance information indicating the barrier performance that the processed base material has against the barrier target substance based on the first feature amount.

[13] The first input unit further receives input of target performance information indicating a target performance of a processed substrate obtained by processing the substrate before processing, The estimation device described in

[12] , wherein the target performance information includes target barrier performance information indicating a target of barrier performance, which is the performance of the processed base material to barrier substances.

[14] The performance includes at least one of the barrier property, heat sealability, mechanical stability, bending resistance, aroma retention, oil resistance, chemical resistance, flame retardancy, high frequency sealability, and color of the processed substrate.

[15] The model includes a model for each of the performance types and the base material, The estimation unit inputs the substrate information and processing information related to the processing of the substrate before processing into a model corresponding to the substrate indicated by the performance type and substrate information indicating the characteristics of the substrate before processing, and outputs the estimated performance information.

[16] A second input unit that accepts input of conditions related to the processing is further provided, The estimation device described in any of [1] to

[15] , wherein the estimation unit estimates the estimated performance information based on substrate information indicating characteristics of the substrate before processing and processing information indicating processing that satisfies the condition accepted as input by the second input unit.

[17] A third input unit that accepts input of substrate information indicating characteristics of a substrate, processing information related to processing conditions of the substrate, and performance information indicating performance of a processed substrate obtained by subjecting the substrate to the processing; a model learning unit that learns a relationship between the substrate information, the processing information, and the performance information, and learns a model that outputs the performance information based on the substrate information and the processing information; A model learning device comprising:

[18] Further comprising a model storage unit that stores the model, The third input unit additionally accepts input of the base material information, the processing information, and the performance information, The model learning device described in

[17] , wherein the model learning unit re-learns the model using the substrate information, the processing information, and the performance information used to learn the model stored in the model memory unit, and the additionally accepted substrate information, the processing information, and the performance information.

[19] An estimation method executed by an estimation device, Receiving input of substrate information indicating characteristics of the substrate before processing; A step of estimating estimated performance information indicating performance of a processed substrate obtained by processing the substrate before processing, based on substrate information indicating characteristics of the substrate before processing and any one of a plurality of processing information related to processing conditions of the substrate; An estimation method including:

[20] A step of receiving input of substrate information indicating characteristics of a substrate, processing information related to processing conditions of the substrate, and performance information indicating performance of a processed substrate obtained by subjecting the substrate to the processing; A step of learning a relationship between the substrate information, the processing information, and the performance information, and learning a model that outputs the performance information based on the substrate information and the processing information; A model training method including: [twenty one] A program for causing a computer to function as an estimation device according to any one of [1] to

[16] . [twenty two] A program for causing a computer to function as the model learning device described in

[17] or

[18] . Effect of the Invention

[0007] According to the estimation device, model learning device, estimation method, model learning method, and program of the present disclosure, it is possible to estimate appropriate conditions for processing a substrate. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a functional block diagram illustrating an example of an estimation device according to an embodiment of the present invention. [Diagram 2] 1 is an example of an operation screen for inputting information to a first input unit. [Diagram 3] 13 is an example of a screen for displaying recommended information and inputting information into a second input unit. [Figure 4] FIG. 2 is a functional block diagram illustrating an example of a model learning device according to the present embodiment. [Diagram 5] 2 is a flowchart showing an example of an operation of the estimation device shown in FIG. 1. [Figure 6] 5 is a flowchart showing an example of the operation of the model learning device shown in FIG. [Figure 7] FIG. 2 is a hardware block diagram of the estimation device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] <Configuration of the estimation device> The configuration of an estimation device 1 of this embodiment will be described with reference to FIG.

[0010] The estimation device 1 according to this embodiment includes a processing information storage unit 101, a regulation information storage unit 102, a model storage unit 103, a first input unit 104, a search range generation unit 105, an estimation unit 106, a recommendation unit 107, a display unit 108, and a second input unit 109.

[0011] The processing information storage unit 101, the regulation information storage unit 102, and the model storage unit 103 are configured by a memory. The memory may be a hard disk drive (HDD), a solid state drive (SSD), an electrically erasable programmable read-only memory (EEPROM), a read only memory (ROM), and a random access memory (RAM), etc. The first input unit 104 and the second input unit 109 are configured by an input interface. The input interface may be a pointing device, a keyboard, a mouse, etc. The search range generation unit 105, the estimation unit 106, and the recommendation unit 107 are configured by a controller. The controller may be configured by dedicated hardware such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA), may be configured by a processor, or may be configured to include both. The display unit 108 is configured by a display interface. The display interface may be an organic electroluminescence (EL), a liquid crystal panel, etc.

[0012] The processing information storage unit 101 stores processing information related to processing of a substrate. The substrate may be, but is not limited to, a plastic sheet, a woven fabric, a nonwoven fabric, wood, paper, a metal foil, or the like. The processing of the substrate is, for example, coating the substrate with latex, and the latex is composed of a polymer and an additive. The processing information is information related to processing conditions of the substrate. The processing information may include polymer information related to a polymer to be formulated for the substrate, additive information related to an additive to be added to the polymer, and coating conditions for coating the substrate with the polymer. The polymer information may include information on the type of polymer, the monomers constituting the polymer, the polymerization ratio of the monomers constituting the polymer, the molecular weight of the polymer, the particle size of the polymer, and the like. The additive information may include information on the type of additive, and the like. The processing information may also include composition information related to the composition of the latex. The composition information may further include information on the additive ratio. The additive ratio is the ratio of the amount of additive to the amount of polymer.

[0013] An example of the processing information may be information indicating that the polymer is "latex A", the additives are "silica" in an amount of "80" and "wax" in an amount of "40", the amount applied is "80", the number of applications is "2", the drying temperature is "100", and the drying time is "2". Another example of the processing information may be information indicating that the polymer is "latex A", the additives are "silica" in an amount of "70" and "wax" in an amount of "50", the amount applied is "60", the number of applications is "3", the drying temperature is "120", and the drying time is "1".

[0014] The regulatory information storage unit 102 stores polymers and additives that are permitted for use in a country where the processed base material obtained by processing the base material is used, in association with the country. The regulatory information storage unit 102 also stores polymers and additives that are permitted for use in a field where the processed base material obtained by processing the base material is used, in association with the field.

[0015] The model storage unit 103 stores a model that learns the relationship between substrate information indicating the characteristics of the substrate, processing information, and performance information indicating the performance of the processed substrate obtained by processing the substrate, and outputs performance information when the substrate information and processing information are input. The performance indicated by the performance information can include processed substrate performance, which is the performance of the processed substrate, and processing stability performance, which is related to the stability of processing. For example, in a configuration in which the substrate is paper and latex is applied, the processed substrate performance can be the coated paper performance, and the processed substrate stability performance can be the coating stability performance. In addition, the performance can include at least one of the barrier property, heat seal property, mechanical stability, bending resistance, aroma retention, oil resistance, chemical resistance, flame retardancy, high frequency seal property, and color of the processed substrate. The barrier property includes gas barrier property, moisture resistance, etc.

[0016] The model storage unit 103 may store a model received via a communication network from the model learning device 2, which will be described in detail later, or may store a model acquired from the model learning device 2 via any medium. In addition, when the estimation device 1 is configured integrally with the model learning device 2, the model storage unit 103 of the estimation device 1 may be the model storage unit 203 of the model learning device 2, which will be described in detail later.

[0017] The models stored in the model storage unit 103 may include models for each type of performance and each substrate. For example, the models stored in the model storage unit 103 may include a model corresponding to barrier properties and paper, and a model corresponding to barrier properties and a plastic sheet. Also, for example, the models stored in the model storage unit 103 may include a model corresponding to heat seal properties and paper, and a model corresponding to heat seal properties and a plastic sheet.

[0018] The first input unit 104 accepts input of substrate information indicating the characteristics of the substrate before processing, for example, via an operation screen as shown in Fig. 2. The substrate information includes the type of substrate. The substrate type is, for example, a plastic sheet, a woven fabric, a nonwoven fabric, wood, paper, metal foil, etc. as described above. The substrate information may also include the properties of the substrate for each type of substrate. For example, when the substrate type is paper, the properties of the substrate can be represented by coating base paper parameters such as porosity, degree of beating, degree of sizing, and amount of crimp.

[0019] The first input unit 104 may further receive an input of a country or field in which the processed base material obtained by processing the base material will be used.

[0020] The first input unit 104 may further receive input of target performance information indicating a target performance of a processed substrate obtained by processing the substrate before processing.

[0021] As an example, the first input unit 104 may accept substrate information, country or field, and target performance information input by a user using an operation screen as shown in Fig. 2. In the example shown in Fig. 2, the user can select a substrate, a country, and a field in a substrate selection section, a region selection section, and a field selection section, respectively, displayed on the operation screen.

[0022] The operation screen may also display a substrate property input section that allows the user to input the properties of the substrate according to the substrate selected in the substrate selection section. In the example shown in Fig. 2, since "paper" is selected in the substrate selection section, input fields for the properties of "paper", such as porosity, degree of beating, degree of sizing, and amount of voids, are displayed, and the first input section 104 can receive the properties of "paper" input by the user in the input fields.

[0023] The first input unit 104 may further receive input of barrier target substance information indicating a barrier target substance that the processed substrate acts as a barrier. Examples of the barrier target substance include barrier target gases such as water vapor, oxygen, fragrances, and odorous substances, and barrier target liquids such as oils and chemicals. The barrier target substance information may include information on the type, molecular structure, molecular weight, and the like of the barrier target substance. Note that, as the barrier target substance information, a feature amount (described later) calculated from information on the structure of the barrier target substance and the like may be used. In such a configuration, the above-mentioned target performance information may include target barrier performance information indicating a target of the barrier performance, which is the performance of the processed substrate to barrier substances.

[0024] Also, the operation screen may display a target performance input section for inputting target performance. In the example shown in FIG. 2, input fields for the minimum values ​​of barrier property and heat seal property are displayed, and the first input unit 104 can accept performance equal to or greater than the minimum value input by the user in the input fields as target performance. It should be noted that the example is not limited to that shown in FIG. 2, and an input field for an arbitrary performance range may be displayed on the operation screen, and the first input unit 104 can accept performance within the range as target performance. It should be noted that the first input unit 104 may accept the input of the information when the user inputs the above-mentioned information and presses the "start search" button.

[0025] The first input unit 104 may further receive an input of a range of processing conditions. Specifically, the first input unit 104 may receive an input of a range of one or more of the coating amount, the number of coatings, the drying temperature, and the drying time included in the coating conditions as the processing conditions.

[0026] The search range generating unit 105 generates search range information based on the substrate information indicating the characteristics of the substrate before processing and the processing information corresponding to the country or field among the multiple processing information. The search range information is information indicating the range of the substrate information and the processing information when the estimation unit 106 estimates the estimated performance information.

[0027] Specifically, the search range generating unit 105 may generate, as the search range, a range including polymers and additives approved for use in the country and field whose input is accepted by the first input unit 104. More specifically, the search range generating unit 105 may generate, as the search range, a range including polymers and additives stored in the regulatory information storage unit 102 in association with the country and field whose input is accepted by the first input unit 104.

[0028] The estimation unit 106 estimates estimated performance information indicating the performance of a processed substrate obtained by processing the substrate before processing, based on substrate information indicating the characteristics of the substrate before processing and any one of a plurality of processing information related to the processing conditions of the substrate. Specifically, the estimation unit 106 may input the substrate information indicating the characteristics of the substrate before processing and any one of a plurality of processing information to the above-mentioned model, and output the estimated performance information. The estimation unit 106 inputs each of various processing information, which is a combination of a polymer, an additive, the additive ratio, coating conditions, etc., as described above, and substrate information, and outputs each performance information based on each of the processing information and the substrate information.

[0029] As an example, the estimation unit 106 may input substrate information indicating the characteristics of the substrate before processing and processing information stored in the processing information storage unit 101 that indicates processing whose processing conditions are within the range accepted as input by the first input unit 104, and output estimated performance information.

[0030] Furthermore, as described above, in a configuration in which the model includes a model for each performance type and each substrate, the estimation unit 106 can input substrate information and processing information related to the processing of the substrate before processing to a model corresponding to the substrate indicated by the performance type and the substrate information indicating the characteristics of the substrate before processing, and output estimated performance information.

[0031] The estimation unit 106 may also have a feature calculation unit 106a that calculates a feature representing the characteristics of the substrate indicated by the substrate information and the processing indicated by the processing information. In such a configuration, the estimation unit 106 can input a feature representing the characteristics of the substrate indicated by the substrate information and the processing indicated by the processing information into the above-mentioned model, and output estimated performance information. Here, the feature may include physical property value information related to physical property values ​​and partial structure information related to partial structures. For example, the physical property value information can include surface charge, surface area, and solubility index. Furthermore, the partial structure information can include a molecular fingerprint indicating a partial structure contained in a molecule. For example, such a feature can be calculated or generated using information on the monomers that constitute the polymer, for example, in the case of polymer information in the processing information. Furthermore, the solubility index can include the Hansen solubility parameter. The solubility index may be calculated using the Hansen solubility parameter of each substance obtained experimentally, or may be calculated using an existing estimation tool such as HSPiP that can be calculated based on the molecular structure of each substance, or an estimation tool created based on public data and data obtained by the user. Furthermore, the relationship between the solubility indices of a plurality of substances can be used as a feature. Such a relationship can be, for example, the difference in solubility between two substances, for example, the difference between the solubility index of a polymer and the solubility index of a barrier target substance. By using such a relationship as a feature, it is possible to improve the prediction performance of the estimation unit. For example, by using the difference between the solubility index of a polymer and the solubility index of a barrier target substance as a feature, it is possible to reflect the affinity between the polymer and the barrier target substance, and it is possible to improve the prediction performance of an index related to barrier properties.

[0032] For this reason, the feature amount calculation unit 106a may use at least one of the substrate information and the processing information, and the barrier target substance information to calculate a first feature amount indicating a feature of at least one of the substrate information and the processing information, and the barrier target substance information. In such a configuration, the estimation unit 106 can estimate barrier performance information indicating the barrier performance that the processed substrate has against the barrier target substance, based on the first feature amount.

[0033] The recommendation unit 107 outputs recommendation information including at least one of recommended substrate information indicating recommended characteristics of the substrate and recommended processing information indicating recommended processing conditions of the substrate based on the target performance information and the estimated performance information. The recommendation unit 107 can output recommendation information based on search range information according to the range of processing-related conditions. The recommended processing conditions can include recommended polymer formulation conditions that are recommended conditions related to a polymer to be formulated for the substrate, recommended additive formulation conditions that are recommended conditions related to an additive to be added to a polymer formulated for the substrate, and recommended coating conditions that are conditions for coating a polymer on the substrate. At this time, the recommendation unit 107 may output recommendation information such that the multiple performances indicated by the estimated performance information become a Pareto optimal solution.

[0034] Specifically, the recommendation unit 107 includes a generation unit 107a, an acquisition unit 107b, and a determination unit 107c.

[0035] The generating unit 107a generates candidate information including at least one of the substrate information and the processing information. In the first search, the generating unit 107a may generate candidates by random numbers within the range indicated by the search range information based on the substrate information and the processing information generated by the search range generating unit 105. In the second search and thereafter, the generating unit 107a may generate new candidates for the candidates generated in the previous search using an evaluation function based on the estimated performance information obtained by the acquiring unit 107b. The generating unit 107a may generate candidate information that changes the output value of the evaluation function in a desired direction using an existing condition search algorithm based on the gradient of the evaluation function, such as gradient descent, stochastic gradient descent, AdaGrad, RMSProp, and Adam. As the evaluation function based on the estimated performance information, a function that outputs the value of the estimated performance information itself may be used when the target of the performance information is specified as exceeding or falling below a specific value. In this case, the desired direction of the output value of the evaluation function is the direction that maximizes or minimizes the output value. Also, when the target is specified to be within a specific range, the square of the difference between any point within the specific range (for example, the average value of the upper and lower limits of the range) and the estimated performance information may be used as the evaluation function. In this case, the desired direction of the output value of the evaluation function is the direction that minimizes the output value. The generating unit 107a may calculate the gradient of the evaluation function based on the value of the evaluation function corresponding to the candidate information generated previously. In order to generate the candidate information more appropriately, the generating unit 107a may generate the candidate information by utilizing an optimization method for solving various combinatorial optimization problems, and may generate the candidate information by using, but is not limited to, an existing optimization method such as simulated annealing or a multi-objective genetic algorithm.

[0036] The acquiring unit 107b acquires the estimated performance information based on the candidate information from the estimating unit 106. Specifically, the acquiring unit 107b acquires the estimated performance information estimated by the estimating unit 106 based on the substrate information and processing information included in the candidate information generated by the generating unit 107a.

[0037] The determining unit 107c determines whether or not to generate new candidate information based on a search end condition, which is a condition that the target performance is satisfied or the number of times that the candidate information has been generated reaches a predetermined upper limit number of times.

[0038] Specifically, the determining unit 107c determines whether or not a search end condition is satisfied. That is, the determining unit 107c determines whether or not a condition, which is a search end condition, that the performance indicated by the estimated performance information satisfies the target performance or the number of times the candidate information has been generated is a predetermined upper limit number is satisfied. Then, when the determining unit 107c determines that the search end condition is not satisfied, the generating unit 107a generates new candidate information. Also, when the determining unit 107c determines that the search end condition is satisfied, the generating unit 107a does not generate new candidate information.

[0039] The display unit 108 displays the recommended information output by the recommendation unit 107. Specifically, the display unit 108 may display the recommended information when the determination unit 107c determines that the search end condition is satisfied. At this time, the display unit 108 may display the estimated performance information corresponding to the processing information selected by the user from the recommended information output by the recommendation unit 107. For example, when the condition set "A" is selected in the condition set selection field of the screen as shown in FIG. 3, the display unit 108 can display the recommended information output corresponding to the processing information corresponding to the condition set "A" (for example, the processing information that the polymer is "latex A", the additives are "silica" in an amount of "80" and "wax" in an amount of "40", the application amount is "80", the number of applications is "2", the drying temperature is "100", and the drying time is "2").

[0040] Furthermore, as described above, in a configuration in which the recommendation unit 107 outputs recommendation information such that the performance indicated by the estimated performance information becomes a Pareto optimal solution, the conditions that can be selected from the condition set displayed by the display unit 108 can be a condition set such that the performance becomes a Pareto optimal solution.

[0041] Furthermore, when it is determined that the performance indicated by the estimated performance information satisfies the target performance, the display unit 108 may display the estimated performance information and the substrate information and processing information corresponding to the estimated performance information. In such a configuration, the display unit 108 does not display the estimated performance information indicating the performance that does not satisfy the target performance and the substrate information and processing information corresponding to the estimated performance information.

[0042] When the display unit 108 outputs the recommended information, the second input unit 109 accepts the input of the processing conditions. Specifically, when the estimation unit 106 outputs the performance information and the display unit 108 displays the performance information and the processing information corresponding to the performance information, the second input unit 109 can accept the input of the processing conditions by the user's operation. At this time, the user can perform an operation to input the processing conditions using a numerical input field, a slider, etc. displayed together with the recommended information as shown in FIG. 3. As a result, the above-mentioned estimation unit 106 can estimate the estimated performance information based on the substrate information indicating the characteristics of the substrate before processing and the processing information indicating the processing that satisfies the conditions inputted and accepted by the second input unit 109. As a result, the user can know how the estimated performance information is changed by changing the processing conditions while referring to the performance information and processing information displayed by the display unit 108.

[0043] <Configuration of model learning device> The configuration of the model learning device 2 of this embodiment will be described with reference to Fig. 4. The model learning device 2 may be configured integrally with the estimation device 1, or may be configured separately.

[0044] The model learning device 2 according to this embodiment includes a third input unit 201, a model learning unit 202, a model storage unit 203, and an output unit 204. The third input unit 201 is configured by an input interface. The model learning unit 202 is configured by a controller. The model storage unit 203 is configured by a memory. The output unit 204 is configured by an output interface. The output interface may include a communication interface. For the communication interface, standards such as Ethernet (registered trademark), FDDI (Fiber Distributed Data Interface), and Wi-Fi (registered trademark) may be used.

[0045] The third input unit 201 accepts input of substrate information indicating characteristics of the substrate, processing information related to processing of the substrate, and performance information indicating performance of the processed substrate obtained by processing the substrate. For example, the third input unit 201 can accept substrate information, processing information, and performance information that have already been obtained by experiments or the like. The third input unit 201 can accept input of substrate information, processing information, and performance information that have been newly clarified by experiments.

[0046] Furthermore, the third input unit 201 can additionally accept input of substrate information, processing information, and performance information. For example, the third input unit 201 can accept input of substrate information indicating the substrate when the substrate is processed by the user, processing information indicating the processing, and performance information indicating the performance obtained by processing the substrate, based on the recommended information recommended by the estimation device 1.

[0047] The model learning unit 202 learns the relationship between the substrate information, the processing information, and the performance information, and learns a model that outputs the performance information based on the substrate information and the processing information. Here, the data used for the model learning may be substrate information, processing information, and performance information that have already been obtained by an experiment or the like and stored in an arbitrary storage unit, or may be substrate information, processing information, and performance information that have been newly clarified by an experiment, or may be a combination of both of these.

[0048] Specifically, when the performance information is a quantitative variable, the model learning unit 202 can learn various other models, such as multiple regression analysis, ridge regression, partial least squares regression, multilayer perceptron, neural networks such as CNN (Convolutional Neural Network) and RNN (Recurrent Neural Network), support vector machines using any kernel function such as Gaussian kernel, random forest modeled as a regression tree, models using hidden Markov models, statistical models, and probability models. The model learning unit 202 can also learn a model that combines various models to make a comprehensive judgment. Furthermore, when the performance information is a qualitative variable, various discriminant analysis methods such as logistic regression, random forest, and neural network may be used.

[0049] Furthermore, the model learning unit 202 learns the model again using the substrate information, processing information, and performance information used in learning the model stored in the model storage unit 203, and the additionally received substrate information, processing information, and performance information. Here, the additionally received substrate information, processing information, and performance information may include, for example, substrate information indicating the substrate when the user processed the substrate based on the recommendation information recommended by the estimation device 1 as described above, processing information indicating the processing, and performance information indicating the performance obtained by processing the substrate.

[0050] Here, the model learning unit 202 may calculate a feature amount representing the characteristics of the substrate indicated by the substrate information and the processing indicated by the processing information. In such a configuration, the model learning unit 202 may learn a model by inputting a feature amount representing the characteristics of the substrate indicated by the substrate information and the processing indicated by the processing information.

[0051] The model storage unit 203 stores the model learned by the model learning unit 202 .

[0052] The output unit 204 outputs the model learned by the model learning unit 202. Specifically, the output unit 204 can output the model to the estimation device 1.

[0053] <Operation of the estimation device> Here, the operation of the estimation device 1 according to this embodiment will be described with reference to FIG. 5. FIG. 5 is a flowchart showing an example of the operation of the estimation device 1 according to this embodiment. The operation of the estimation device 1 described with reference to FIG. 5 corresponds to an example of an estimation method of the estimation device 1 according to this embodiment. As described above, the estimation device 1 may be provided with a model that learns the relationship between substrate information indicating the characteristics of the substrate, processing information, and performance information indicating the performance of the processed substrate obtained by processing the substrate, and outputs the performance information when the substrate information and processing information are input. In this embodiment, the estimation device 1 is provided with a model by storing the model in the model storage unit 103 provided in the estimation device 1.

[0054] In step S11, the first input unit 104 receives input of substrate information indicating the characteristics of the substrate before processing.

[0055] In step S12, the first input unit 104 further receives input of target performance information indicating target performance of a processed substrate obtained by processing the substrate before processing.

[0056] In step S13, the first input unit 104 further receives input of processing information related to processing conditions of the substrate. For example, the first input unit 104 further receives input of a range of conditions related to processing. Alternatively, the first input unit 104 may receive input of a country or field in which the processed substrate is used.

[0057] In step S14, the generating unit 107a generates candidate information including at least one of the base material information and the processing information.

[0058] In step S15, the acquisition unit 107b acquires estimated performance information from the estimation unit 106 based on the candidate information. Specifically, the estimation unit 106 estimates estimated performance information indicating the performance of the processed substrate obtained by processing the substrate before processing, based on substrate information indicating the characteristics of the substrate before processing and any one of a plurality of processing information related to the processing conditions of the substrate. Then, the acquisition unit 107b acquires the estimated performance information. Note that, in step S13, if the input of the country or field is accepted, the search range generation unit 105 generates search range information based on the substrate information indicating the characteristics of the substrate before processing and the processing information corresponding to the country or field among the plurality of processing information, and the estimation unit 106 inputs the estimated performance information based on the substrate information and processing information of the search range indicated in the search range information.

[0059] In step S16, the determining unit 107c determines whether a search end condition is satisfied. For example, the determining unit 107c may determine whether a search end condition is satisfied, which is that the performance indicated by the estimated performance information satisfies the target performance, or the number of times that the candidate information has been generated is a predetermined upper limit number of times.

[0060] If it is determined in step S16 that the search end condition is satisfied, the recommendation unit 107 outputs recommendation information including at least one of recommended substrate information indicating recommended properties of the substrate and recommended processing information indicating recommended processing conditions for the substrate, based on the target performance information and the estimated performance information. If it is determined in step S16 that the search end condition is not satisfied, the estimation device 1 returns to step S14 and repeats the subsequent operations.

[0061] When the recommendation unit 107 outputs the recommended information, the display unit 108 may display the recommended information, and the process may return to step S13, where the second input unit 109 may receive input of conditions related to processing input by a user operation. When the second input unit 109 receives input of conditions related to processing, the estimation device 1 may repeat the operations from step S13 onward.

[0062] <Operation of the model learning device> Here, the operation of the model learning device 2 according to this embodiment will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of the operation of the model learning device 2 according to this embodiment. The operation of the model learning device 2 described with reference to Fig. 6 corresponds to an example of a model learning method of the model learning device 2 according to this embodiment.

[0063] In step S21, the third input unit 201 accepts input of substrate information indicating the characteristics of the substrate, processing information related to the processing conditions of the substrate, and performance information indicating the performance of the processed substrate obtained by processing the substrate.

[0064] In step S22, the model learning unit 202 learns the relationship between the substrate information, the processing information, and the performance information, and learns a model that outputs the performance information based on the substrate information and the processing information.

[0065] In step S23, the model storage unit 203 stores the model.

[0066] In step S24, the third input unit 201 additionally accepts input of substrate information, processing information, and performance information.

[0067] In step S25, the model learning unit 202 learns the model again using the substrate information, processing information, and performance information used to learn the model stored in the model memory unit 203, and the additional substrate information, processing information, and performance information accepted.

[0068] As described above, according to this embodiment, the estimation device 1 includes a first input unit 104 that receives input of substrate information indicating the characteristics of the substrate before processing, and an estimation unit 106 that estimates estimated performance information indicating the performance of a processed substrate obtained by processing the substrate before processing, based on the substrate information indicating the characteristics of the substrate before processing and any one of a plurality of pieces of processing information related to the processing conditions of the substrate. This allows the estimation device 1 to estimate appropriate conditions for processing the substrate.

[0069] According to this embodiment, the model learning device 2 includes a third input unit 201 that receives input of substrate information indicating substrate characteristics, processing information related to processing conditions of the substrate, and performance information indicating performance of the processed substrate obtained by processing the substrate, and a model learning unit 202 that learns the relationship between the substrate information, the processing information, and the performance information, and learns a model that outputs the performance information based on the substrate information and the processing information. This allows the model learning device 2 to learn a model for the estimation device 1 to estimate appropriate conditions for processing the substrate.

[0070] <Program> The above-mentioned estimation device 1 and model learning device 2 can each be realized by a computer 401. A program for causing each of the estimation device 1 and the model learning device 2 to function may be provided. The program may be stored in a storage medium or provided through a network. FIG. 7 is a block diagram showing a schematic configuration of a computer 401 functioning as the estimation device 1. A computer functioning as the model learning device 2 may also be configured similarly to the computer 401. Here, the computer 401 may be a general-purpose computer, a dedicated computer, a workstation, a PC (Personal Computer), an electronic notepad, or the like. The program instructions may be program code, code segments, or the like for performing required tasks.

[0071] 7, the computer 401 includes a processor 410, a ROM 420, a RAM 430, a storage 440, an input unit 450, an output unit 460, and a communication interface (I / F) 470. Each component is connected to each other so as to be able to communicate with each other via a bus 480. The processor 410 is specifically a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), a SoC (System on a Chip), or the like, and may be configured by a plurality of processors of the same type or different types.

[0072] The processor 410 controls each component and executes various arithmetic processing. That is, the processor 410 reads a program from the ROM 420 or the storage 440, and executes the program using the RAM 430 as a working area. The processor 410 controls each component and executes various arithmetic processing according to the program stored in the ROM 420 or the storage 440. In the above-described embodiment, the program according to the present disclosure is stored in the ROM 420 or the storage 440.

[0073] The program may be stored in a storage medium readable by the computer 401. By using such a storage medium, the program can be installed in the computer 401. Here, the storage medium in which the program is stored may be a non-transitory storage medium. The non-transitory storage medium is not particularly limited, and may be, for example, a CD-ROM, a DVD-ROM, a USB (Universal Serial Bus) memory, or the like. In addition, the program may be in a form that is downloaded from an external device via a network.

[0074] The ROM 420 stores various programs and various data. The RAM 430 temporarily stores programs or data as a working area. The storage 440 is configured with an HDD or SSD, and stores various programs including an operating system and various data.

[0075] The input unit 450 includes one or more input interfaces that receive an input operation from a user and acquire information based on the user's operation. For example, the input unit 450 is a pointing device, a keyboard, a mouse, etc., but is not limited to these.

[0076] The output unit 460 is a display that outputs information in the form of a video, but is not limited to this. If the output unit 460 is a touch panel type display, it also functions as the input unit 450.

[0077] The communication interface 470 is an interface for communicating with external devices.

[0078] All publications, patent applications, and standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, and standard was specifically and individually indicated to be incorporated by reference.

[0079] Although the above-mentioned embodiment has been described as a representative example, it is obvious to those skilled in the art that many modifications and substitutions can be made within the spirit and scope of the present disclosure. Therefore, the present invention should not be interpreted as being limited by the above-mentioned embodiment, and various modifications or changes can be made without departing from the scope of the claims. For example, it is possible to combine multiple building blocks shown in the configuration diagram of the embodiment into one, or to divide one building block. [Explanation of symbols]

[0080] 1 Estimation device 2. Model learning device 101 Processing information storage section 102 Regulatory information storage unit 103 Model Memory Unit 104 First input section 105 Search range generation unit 106 Estimation part 107 Recommendation section 107a Generation part 107b Acquisition Department 107c Judgment section 108 Display section 109 Second input section 201 Third Input Section 202 Model Learning Department 203 Model Memory Unit 204 Output section 401 Computer 410 Processor 420 ROM 430 RAM 440 Storage 450 Input section 460 Output section 470 Communication Interface (Communication I / F) 480 Bus

Claims

1. a first input unit that receives input of substrate information indicating characteristics of the substrate before processing; an estimation unit that estimates estimated performance information that indicates the performance of a processed substrate obtained by processing the substrate before processing, based on substrate information that indicates the characteristics of the substrate before processing and any one of a plurality of processing information related to processing conditions of the substrate; An estimation device comprising:

2. The method further includes a model that learns the relationship between substrate information indicating the characteristics of the substrate, the processing information, and performance information indicating the performance of a processed substrate obtained by processing the substrate, and outputs the performance information when the substrate information and the processing information are input, The estimation device according to claim 1 , wherein the estimation unit inputs substrate information indicating characteristics of the substrate before processing and any one of the plurality of pieces of processing information to the model, and outputs the estimated performance information.

3. The first input unit further receives input of target performance information indicating target performance of a processed substrate obtained by processing the unprocessed substrate, 3. The estimation device according to claim 1, further comprising a recommendation unit that outputs recommended information including at least one of recommended substrate material information indicating recommended characteristics of the substrate and recommended processing information indicating recommended processing conditions for the substrate, based on the target performance information and the estimated performance information.

4. The recommendation section a generation unit that generates candidate information including at least one of the base material information and the processing information; an acquisition unit that acquires the estimated performance information from the estimation unit based on the candidate information; a determination unit that determines whether or not to generate new candidate information based on a search termination condition; The estimation device of claim 3 , comprising:

5. the determination unit determines whether or not the search termination condition is satisfied, the condition being that the performance indicated by the estimated performance information satisfies the target performance or that the number of times candidate information has been generated is a predetermined upper limit number of times; a display unit that displays the recommended information when it is determined that the search termination condition is satisfied; The estimation device according to claim 4 , comprising:

6. the first input unit further accepts input of a range of conditions related to the processing; The estimation device according to claim 3 , wherein the recommendation unit outputs the recommendation information based on search range information corresponding to a range of conditions related to the processing.

7. The first input unit further accepts input of a country or field in which the processed base material is used, The estimation device according to claim 6, further comprising a search range generation unit that generates the search range information based on substrate information indicating characteristics of the substrate before processing and processing information corresponding to the country or the field among the plurality of pieces of processing information.

8. The estimation device according to claim 1 or 2, wherein the processing information includes a type of polymer to be applied to the substrate, a type of additive to be added to the polymer, and coating conditions for coating the polymer on the substrate.

9. The estimation device according to claim 1 , wherein the estimation unit includes a feature amount calculation unit that calculates a feature amount that represents a feature of either the base material indicated by the base material information or the processing indicated by the processing information.

10. The estimation device according to claim 1 , wherein the first input unit further receives barrier target substance information indicating a substance that acts as a barrier to the processed base material.

11. the estimation unit includes a feature amount calculation unit that calculates a feature amount representing a feature of either the base material indicated by the base material information or the processing indicated by the processing information, The estimation device according to claim 10, wherein the feature calculation unit calculates a first feature indicating a characteristic of at least one of the substrate information and the processing information and the barrier target substance information, using at least one of the substrate information and the processing information and the barrier target substance information.

12. The estimation device according to claim 11 , wherein the estimation unit estimates barrier performance information indicating a barrier performance of the processed base material against the barrier target substance based on the first feature amount.

13. The first input unit further receives input of target performance information indicating target performance of a processed substrate obtained by processing the unprocessed substrate, The estimation device according to claim 12 , wherein the target performance information includes target barrier performance information indicating a target of barrier performance, which is the performance of the processed substrate to barrier substances.

14. The estimation device according to claim 1 or 2, wherein the performance includes at least one of a barrier property, a heat seal property, a mechanical stability, a bending resistance, an aroma retention property, an oil resistance, a chemical resistance, a flame retardancy, a high frequency seal property, and a color of the processed substrate.

15. The model includes a model for each of the performance types and the substrate, 3. The estimation device according to claim 2, wherein the estimation unit inputs the substrate information and processing information related to processing of the substrate before processing into a model corresponding to the substrate indicated by the performance type and substrate information indicating characteristics of the substrate before processing, and outputs the estimated performance information.

16. a second input unit that accepts input of conditions related to the processing; 3. The estimation device according to claim 1, wherein the estimation unit estimates the estimated performance information based on substrate information indicating characteristics of the substrate before processing and processing information indicating processing that satisfies the conditions accepted as input by the second input unit.

17. a third input unit that receives input of substrate information indicating characteristics of a substrate, processing information related to processing conditions of the substrate, and performance information indicating performance of a processed substrate obtained by subjecting the substrate to the processing; a model learning unit that learns the relationship between the substrate information, the processing information, and the performance information, and learns a model that outputs the performance information based on the substrate information and the processing information; A model learning device comprising:

18. further comprising a model storage unit that stores the model; the third input unit additionally accepts input of the substrate information, the processing information, and the performance information, 18. The model learning device according to claim 17, wherein the model learning unit learns the model again using the substrate information, the processing information, and the performance information used in learning the model stored in the model storage unit, and the additionally accepted substrate information, the processing information, and the performance information.

19. An estimation method executed by an estimation device, receiving input of substrate information indicating characteristics of the substrate before processing; a step of estimating estimated performance information indicating the performance of a processed substrate obtained by processing the substrate before processing, based on substrate information indicating the characteristics of the substrate before processing and any one of a plurality of processing information related to processing conditions of the substrate; Estimation methods including:

20. receiving input of substrate information indicating characteristics of a substrate, processing information relating to processing conditions of the substrate, and performance information indicating performance of a processed substrate obtained by subjecting the substrate to the processing; a step of learning a relationship between the substrate information, the processing information, and the performance information, and learning a model that outputs the performance information based on the substrate information and the processing information; Model learning methods including:

21. A program for causing a computer to function as the estimation device according to claim 1 or 2.

22. A program for causing a computer to function as the model learning device according to claim 17 or 18.

Images