Estimation method, estimation model generation method, and model determination model generation method

Abstract

The performance of the coating film can be estimated without actually applying the paint. [Solution] An estimation method that uses a coating performance estimation model learned from variables including information on the formulation and / or properties of the paint and the measurement results of the coating performance, and outputs an estimated value of the measurement results of the coating performance by inputting information on the formulation and / or properties of the paint into the coating performance estimation model.

Description

Technical Field

[0001] The present invention relates to an estimation method, a method for generating an estimation model, and a method for generating a usage model determination model. This application claims priority based on Japanese Patent Application No. 2024-232633 filed in Japan on December 27, 2024, and incorporates its content herein.

Background Art

[0002] A coating film is formed by applying paint to an object to be coated and drying it. The coating film has roles such as imparting aesthetics to the object to be coated and protecting it from the external environment, and it is very important to grasp various coating film performances according to the coating purpose. However, it is difficult to predict various performances of the coating film from the formulation information and the state of the paint. Therefore, it is common to grasp various performances of the coating film by measuring the coating film formed by applying the actually prepared paint with a predetermined film thickness and drying it. However, since a great deal of man-hours are required for the preparation of the paint and the coating work, a technique for estimating the performance of the coating film without preparing the paint and forming the coating film is demanded.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present invention is to provide an estimation method, a method for generating an estimation model, and a method for generating a usage model determination model that can estimate coating film performance without preparing paint and forming a coating film.

Means for Solving the Problems

[0005] One aspect of the present invention is an estimation method that uses a coating performance estimation model learned from variables including information on the formulation and / or properties of the paint and measurement results of the coating performance, and outputs an estimated value of the measurement results of the coating performance by inputting information on the formulation and / or properties of the paint into the coating performance estimation model. [Effects of the Invention]

[0006] According to the present invention, the performance of a coating film can be estimated without applying paint. [Brief explanation of the drawing]

[0007] [Figure 1] This figure shows an example of the configuration of the estimation model generation device 10 according to the first embodiment. [Figure 2] This is a flowchart showing the operation of the estimation model generation device 10 according to the first embodiment. [Figure 3] This figure shows an example of the configuration of the estimation device 20 according to the first embodiment. [Figure 4] This is a flowchart showing the operation of the estimation device 20 according to the first embodiment. [Figure 5] This figure shows the configuration of the estimation model generation device 10 according to the second embodiment. [Figure 6] This figure shows the configuration of the estimation device 20 according to the second embodiment. [Figure 7] This is a flowchart showing the operation of the estimation model generation device 10 according to the second embodiment. [Figure 8] This is a flowchart showing the operation of the estimation device 20 according to the second embodiment. [Figure 9] This is a scatter plot with the amount of white pigment in the paint on the horizontal axis and the opacity film thickness of the paint on the vertical axis. [Figure 10] This figure shows the experimental results. [Modes for carrying out the invention]

[0008] Embodiments of the present invention will be described in detail below with reference to the drawings. (First Embodiment) Figure 1 shows an example of the configuration of the estimation model generation device 10 according to the first embodiment. The estimation model generation device 10 includes a dataset acquisition unit 110, an estimation model generation unit 120, and an estimation model output unit 130.

[0009] The dataset acquisition unit 110 acquires a dataset. The dataset includes information as variables regarding the paint formulation and / or properties of the paint and the measurement results of the coating performance.

[0010] The performance of the coating film is not limited to the performance of a single coating film obtained by applying and curing one type of paint. Examples include the shielding performance of the coating film against electromagnetic waves (light, heat, etc.), coating film hardness, color, surface properties (hydrophilicity, stain resistance, etc.), adhesion, glass transition temperature of the coating film, coating film properties (viscoelasticity, smoothness, flexibility, etc.), processability, corrosion resistance of processed parts, abrasion resistance, biodegradability, rust prevention, tactile feel, moisture resistance, salt water resistance, low-temperature impact resistance, low-temperature flexibility, low-temperature brittleness, cold resistance, heat resistance, weathering resistance, scratch resistance, non-flammability, flame retardancy, thermal shock resistance, suitability for recoating, crack resistance, and water whitening resistance. The estimated model generated by the estimated model generation device 10 according to this embodiment is particularly effective in estimating the shielding performance of the coating film against electromagnetic waves. The electromagnetic wave shielding performance of a coating film includes at least one selected from the group consisting of, for example, opacity, light transmittance, light reflectance, and heat shielding. Hereinafter, the electromagnetic wave shielding performance of a coating film will simply be referred to as "shielding performance." Opacity refers to the ability of a paint film to conceal the color of the underlying surface. Paints with low opacity are more susceptible to the influence of the underlying color. Opacity can be evaluated using methods such as those specified in Japanese Industrial Standard K5600-4-1.

[0011] Light transmittance is the rate at which light is transmitted through a coating film; a higher value indicates greater transparency of the coating film. Types of light transmittance include total light transmittance and directional transmittance. The value of light transmittance varies depending on the measurement wavelength, and it can be measured using light in any wavelength range, including ultraviolet, visible, and infrared light. Light reflectance is the reflectivity of light by a coating film; a higher value indicates higher reflectance. Types of light reflectance include total light transmittance, diffuse reflectance, and specular reflectance. The value of light reflectance varies depending on the measurement wavelength, and it can be measured using light in any wavelength range, such as ultraviolet light, visible light, and infrared light.

[0012] Heat shielding refers to the heat-blocking performance of a coating film; the higher the heat shielding, the less heat the coating film conducts.

[0013] The paint formulation is information regarding the amount and proportion of each component that makes up the paint. In this embodiment, the paint formulation used as a variable or input information for the coating performance estimation model includes at least one selected from the group consisting of: the type of glossing agent contained in the paint, the amount of glossing agent contained in the paint, the color of the glossing agent contained in the paint, the particle size of the glossing agent contained in the paint, the thickness of the glossing agent contained in the paint, the type of surface treatment of the glossing agent contained in the paint, the type of coloring pigment contained in the paint, the amount of coloring pigment contained in the paint, the color of the coloring pigment contained in the paint, the chemical structure of the coloring pigment contained in the paint, the particle size of the coloring pigment contained in the paint, the amount of resin contained in the paint, the chemical structure of the resin contained in the paint, the amount of dispersant contained in the paint, the chemical structure of the dispersant contained in the paint, the amount of thickener contained in the paint, and the chemical structure of the thickener contained in the paint.

[0014] The brightening material is a flaky pigment or plate-shaped pigment that imparts a glittering or optical interference property to the coating film. Examples of the types of brightening materials include flaky aluminum (aluminum flakes), vapor-deposited aluminum, aluminum oxide, oxybisthmus chloride, mica, mica coated with titanium oxide, mica coated with iron oxide, iron oxide mica, silica coated with titanium oxide, alumina coated with titanium oxide, silica coated with iron oxide, alumina coated with iron oxide, glass flakes, colored glass flakes, vapor-deposited glass flakes, hologram films, and the like.

[0015] Examples of coloring pigments include white pigments. Examples of white pigments include titanium dioxide. Examples of coloring pigments include black pigments. Examples of black pigments include carbon black, acetylene black, lamp black, bone black, graphite, iron black, and aniline black. Examples of coloring pigments include yellow pigments. Examples of yellow pigments include yellow iron oxide, titanium yellow, monoazo yellow, condensed azo yellow, azomethine yellow, bismuth vanadate, benzimidazolone, isoindolinone, isoindoline, quinophthalone, benzidine yellow, and permanent yellow. Examples of coloring pigments include orange pigments. Examples of orange pigments include permanent orange. Examples of coloring pigments include red pigments. Examples of red pigments include red iron oxide, naphthol AS-based azo red, anthenslon, anthraquinonyl red, perylene maroon, quinacridone-based red pigment, diketopyrrolopyrrole, watching red, and permanent red. Examples of coloring pigments include brown pigments. Examples of brown pigments include brown iron oxide. Examples of coloring pigments include purple pigments. Examples of purple pigments include cobalt purple, quinacridone violet, and dioxazine violet. Examples of coloring pigments include blue pigments. Examples of blue pigments include cobalt blue, phthalocyanine blue, and suren blue. Examples of coloring pigments include green pigments. Examples of green pigments include phthalocyanine green. Coloring pigments may be combinations of these pigments.

[0016] The resin is a binder that forms a coating film. The resin is not particularly limited as long as it is generally used in paint applications, and it may be a solvent-based or water-based liquid, or a powder for powder coatings. Examples of resin types include acrylic resin, phenolic resin, alkyd resin, polyester resin, urethane resin, polyether resin, polyolefin resin, epoxy resin, silicone resin, fluororesin, ethylene vinyl acetate copolymer, amino resin, polycarbonate resin, polyvinyl chloride resin, isocyanate compound, melamine resin, and the like.

[0017] The dispersant is an additive for uniformly dispersing a brightening material, a coloring pigment, or the like. The dispersant is not particularly limited as long as it is generally used in paint applications, and commercially available products generally sold as dispersants can be used. Examples of dispersant types include nonionic, cationic, or anionic functional group-containing compounds.

[0018] The thickener is an additive for adjusting the viscosity of the paint. The thickener is not particularly limited as long as it is generally used in paint applications, and commercially available products generally sold as thickeners can be used. Examples of thickener types include alkali thickening type, nonionic associative type, cellulose type, water-soluble polymer type, polyamide type, and clay type thickeners.

[0019] The properties of the paint are information about the paint obtained by performing respective measurements on the paint. In the present embodiment, the properties of the paint used as variables and input information of the coating film performance estimation model include at least one selected from the group consisting of the viscosity of the paint, the solid content of the paint, the pH of the paint, the color of the paint, the transmittance of the paint, the surface tension of the paint, and the drying conditions of the paint.

[0020] Formulation information for each paint can be obtained from, for example, product data for each paint or experimental level data during paint development. The properties of each paint are obtained by performing individual measurements. Then, paint films are created by applying and curing each paint, and the performance of the created paint films is measured. This allows for the creation of combinations of the formulation and / or properties of each paint and the measurement results of the paint film performance. This allows for the creation of a dataset containing information on the formulation and / or properties of the paint and the measurement results of the paint film performance as variables. Alternatively, the properties of each paint may be obtained by performing individual measurements on each paint after measuring the performance of the created paint films.

[0021] The following describes the case where the estimation model generated by the estimation model generation device 10 estimates the interruption performance.

[0022] The estimation model generation unit 120 generates an estimation model based on the dataset. The estimation model is a model that outputs an estimated value of the measurement result of the barrier performance when information on the paint formulation and / or properties of the paint is input. The estimation model generation unit 120 generates an estimation model using machine learning methods, with information on the paint formulation and / or properties of the paint as explanatory variables and the measurement result of the barrier performance as the dependent variable. Examples of machine learning techniques include decision trees, decision trees with gradient boosting, linear regression, logistic regression, simple perceptron, MLP, neural networks, support vector machines, random forests, Gaussian processes, Bayesian networks, k-nearest neighbors, Ridge regression, Lasso regression, graph neural networks, K-means algorithm, naive Bayes, CNN, and other algorithms used in machine learning. These algorithms may be used individually or in combination. In this embodiment, it is preferable to use a decision tree with gradient boosting. An example of a decision tree with gradient boosting is LightGBM.

[0023] The estimated model output unit 130 outputs the generated estimated model.

[0024] Figure 2 is a flowchart illustrating the operation of the estimation model generation device 10 according to the first embodiment. The dataset acquisition unit 110 acquires the dataset (step S101). The estimation model generation unit 120 generates an estimation model based on the dataset (step S102). The estimation model output unit 130 outputs the generated estimation model (step S103).

[0025] Figure 3 shows an example of the configuration of the estimation device 20 according to the first embodiment. The estimation device 20 is a device that estimates the coating performance of a paint based on paint data. The following describes the case in which the estimation device 20 estimates the barrier performance of a paint.

[0026] The estimation device 20 comprises a paint data acquisition unit 210, a barrier performance measurement result estimation unit 220, a barrier performance measurement result output unit 230, and a storage unit 290. The storage unit 290 stores the estimation model output by the estimation model generation device 10.

[0027] The paint data acquisition unit 210 acquires paint data. The paint data consists of the paint composition and / or paint properties, which are explanatory variables of the estimation model.

[0028] The barrier performance measurement result estimation unit 220 estimates the barrier performance measurement result by inputting paint data into the estimation model and outputting an estimated value of the barrier performance measurement result.

[0029] The shielding performance measurement result output unit 230 outputs an estimated value of the shielding performance measurement result output by the estimation model. The estimated value of the shielding performance measurement result output from the shielding performance measurement result output unit 230 is recorded in the storage unit 290, for example, along with the corresponding paint data. The estimated value of the shielding performance measurement result output from the shielding performance measurement result output unit 230 is input to and displayed on an external display device, for example.

[0030] Figure 4 is a flowchart showing the operation of the estimation device 20 according to the first embodiment. The paint data acquisition unit 210 acquires paint data (step S201). The barrier performance measurement result estimation unit 220 estimates the barrier performance measurement result using an estimation model based on the paint data (step S202). The barrier performance measurement result output unit 230 outputs an estimated value of the barrier performance measurement result (step S203).

[0031] As described above, the estimation device 20 can estimate the measurement results of the barrier performance before painting by using an estimation model generated by machine learning to estimate the measurement results of the barrier performance from the paint formulation and / or properties of the paint.

[0032] (Second embodiment) The estimation model generation device 10 according to the second embodiment will be described below. For the sake of clarity, the range of values ​​less than a predetermined value will be referred to as range D1, and the range of values ​​greater than or equal to the predetermined value will be referred to as range D2. In the second embodiment, the estimation model generation unit 120 generates multiple estimation models based on the dataset, depending on the range of values ​​indicating the measured blocking performance. For example, the estimation model generation unit 120 generates an estimation model for when the values ​​indicating the measured blocking performance are in range D1 and an estimation model for when they are in range D2, based on the dataset. The estimation model generation unit 120 generates an estimation model corresponding to range D1 based on the data included in the dataset where the values ​​indicating the measured blocking performance are within range D1.

[0033] Furthermore, the estimation model generation unit 120 generates an estimation model corresponding to range D2 based on the data in the dataset whose values ​​indicating the measurement results of blocking performance fall within range D2. However, if the dataset contains only a small number of data whose values ​​indicating the measurement results of blocking performance fall within range D1 or D2, the estimation model generation unit 120 may generate an estimation model corresponding to range D1 or D2 based on all the data in the dataset.

[0034] For example, if the measurement result of the barrier properties is an opacity film thickness indicating black and white opacity, the estimation model generation unit 120 generates an estimation model used when the opacity film thickness is less than a predetermined opacity film thickness α, based on the data included in the dataset where the opacity film thickness is less than a predetermined opacity film thickness α. Alternatively, the estimation model generation unit 120 generates an estimation model used when the opacity film thickness is α or greater, based on all the data included in the dataset, regardless of the range of opacity film thickness. The estimation model generation unit 120 may also generate an estimation model used when the opacity film thickness is α or greater, based on the data included in the dataset where the opacity film thickness is α or greater.

[0035] Figure 5 shows the configuration of the estimation model generation device 10 according to the second embodiment. The estimation model generation device 10 according to the second embodiment includes, in addition to the estimation model generation device 10 according to the first embodiment, a model determination unit 140 and a model determination unit 150. The model determination model generation unit 140 generates a model determination model based on the dataset. The dataset used here includes information as variables about the range in which the measured values ​​of the paint formulation and / or paint properties and barrier performance are included. This dataset may be generated by modifying a dataset acquired by the dataset acquisition unit 110, which includes information about the paint formulation and / or paint properties and barrier performance as variables, and determining the range in which the measured values ​​are included from the information about the measurement results. Alternatively, this dataset may be newly acquired by the dataset acquisition unit 110.

[0036] The range in which the measured values ​​of the interruption performance are included corresponds to the range corresponding to the generated estimation model. In this explanation, the information regarding the range in which the measured values ​​of the interruption performance are included is whether the measured values ​​of the interruption performance are included in range D1 or range D2. The model selection model determines the estimation model to be used when information regarding the paint formulation and / or properties of the paint is input. When information regarding the paint formulation and / or properties of the paint is input, the model selection model outputs estimation results for a range that includes the values ​​shown in the measurement results of the barrier performance, and determines the estimation model corresponding to the range shown in the estimation results.

[0037] The model used for determining the model does not necessarily have to be generated using machine learning. The model used for determining the model used may determine the estimation model to be used using a predetermined formula. For example, the model used for determining the model used may determine the estimation model based on information about the amount of black pigment and / or the amount of aluminum flakes included in the information about the color and amount of the coloring pigment. For example, the model used for determining the model used may determine the estimation model corresponding to a range of small opacity film thickness values ​​when the amount of black pigment is greater than or equal to a predetermined amount. For example, the model used for determining the model used may determine the estimation model corresponding to a range of small opacity film thickness values ​​when the amount of aluminum flakes is greater than or equal to a predetermined amount.

[0038] The model determination output unit 150 outputs the generated model determination model.

[0039] In the estimation device 20 according to the second embodiment, the storage unit 290 stores two estimation models and a model determination model generated by the estimation model generation device 10. Figure 6 is a diagram showing the configuration of the estimation device 20 according to the second embodiment. The estimation device 20 according to the second embodiment includes a model determination unit 240 in addition to the estimation device 20 according to the first embodiment.

[0040] The model determination unit 240 determines the estimation model to be used to estimate the barrier performance by inputting paint data into the model determination model stored in the memory unit 290. The barrier performance measurement result estimation unit 220 estimates the measurement result of the coating film performance by inputting paint data into the estimation model determined by the model determination unit 240 and outputting an estimated value of the measurement result of the coating film performance.

[0041] Figure 7 is a flowchart showing the operation of the estimation model generation device 10 according to the second embodiment. The operation of steps S301 to S303 is the same as the operation of steps S101 to S103 shown in Figure 2. The model generation unit 140 modifies the dataset and generates a dataset that includes information on the range in which the measured values ​​of the paint formulation and / or paint properties and barrier performance are included as variables (step S304). At this point, the dataset acquisition unit 110 may acquire a dataset that newly includes information on the range in which the measured values ​​of the paint formulation and / or paint properties and barrier performance are included as variables.

[0042] The model determination model generation unit 140 generates a model determination model based on the dataset (step S305). The model determination model output unit 150 outputs the generated model determination model (step S306). Note that the generation and output of the model determination model may be performed independently of the generation and output of the estimation model, and the model determination model does not necessarily have to be generated after the estimation model has been generated.

[0043] Figure 8 is a flowchart showing the operation of the estimation device 20 according to the second embodiment. The paint data acquisition unit 210 acquires paint data (step S401). The model determination unit 240 determines the estimation model to be used to estimate the shielding performance based on the paint data (step S402). The shielding performance measurement result estimation unit 220 estimates the measurement result of the shielding performance based on the paint data using the determined estimation model (step S403). The shielding performance measurement result output unit 230 outputs an estimated value of the measurement result of the shielding performance (step S404).

[0044] In the above explanation, for the sake of clarity, the values ​​representing the measurement results of the blocking performance were divided into two ranges, and the explanation described the case where one of the two estimation models is used. However, the values ​​representing the measurement results of the blocking performance may be divided into three or more ranges, and one of the three or more estimation models may be used.

[0045] Of the multiple estimation models generated by the estimation model generation device 10 and stored in the storage unit 290, at least one estimation model may be an estimation model that does not use elements from the paint data that have a small impact on the barrier performance. More specifically, among multiple estimation models, at least one estimation model may estimate the measurement results of the barrier performance without using information on the amount of at least one pigment from among brown, red, and yellow. The amounts of brown, red, and yellow pigments are factors that have little effect on the barrier performance. Therefore, even if information on the amount of at least one pigment from among brown, red, and yellow is not used, the barrier performance can be estimated with good accuracy. If at least one estimation model among multiple estimation models estimates the measurement results of the barrier performance without using information on the amount of at least one pigment from among brown, red, and yellow, it is preferable that another estimation model, different from that estimation model, estimates the measurement results of the barrier performance using information on the amount of at least one pigment from among brown, red, and yellow.

[0046] Furthermore, the multiple estimation models generated by the estimation model generation device 10 and stored in the storage unit 290 may include three or more estimation models that differ from multiple elements among the elements included in the paint data, and the usage model determination unit 240 may determine which estimation model to use based on these multiple elements. More specifically, the model determination model preferably determines the estimated model based on information regarding the amount of black pigment and / or aluminum flakes included in the information regarding the color and amount of coloring pigments. The reason why the model determination model determines the estimated model based on information regarding the amount of black pigment and / or aluminum flakes included in the information regarding the color and amount of coloring pigments will be explained below. Figure 9 is a scatter plot with the amount based on the amount of white pigment in the paint on the horizontal axis and the opacity film thickness of the paint on the vertical axis. Figure 9(a) is a scatter plot when the paint does not contain aluminum flakes and black pigment, with the amount of white pigment in the paint on the horizontal axis and the opacity film thickness of the paint on the horizontal axis. Figure 9(b) is a scatter plot when the paint contains aluminum flakes, with the product of the amount of white pigment and the amount of aluminum flakes in the paint on the horizontal axis and the opacity film thickness of the paint on the horizontal axis. Figure 9(c) is a scatter plot for paint containing black pigment, where the horizontal axis represents the product of the amount of white pigment and the amount of black pigment in the paint, and the horizontal axis represents the opacity film thickness of the paint. It can be seen that when the paint contains aluminum flakes or black pigment, the opacity film thickness is smaller compared to when the paint does not contain aluminum flakes or black pigment.

[0047] Therefore, for example, the estimation model generation device 10 generates four estimation models: a first estimation model used when the amount of aluminum flakes in the paint is less than a first predetermined amount and the amount of black pigment is less than a second predetermined amount; a second estimation model used when the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount and the amount of black pigment is less than a second predetermined amount; a third estimation model used when the amount of aluminum flakes in the paint is less than a first predetermined amount and the amount of black pigment is equal to or greater than a second predetermined amount; and a fourth estimation model used when the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount and the amount of black pigment is equal to or greater than a second predetermined amount. In this case, the model used for estimation is determined as follows: if the amount of aluminum flakes in the paint is less than the first predetermined amount and the amount of black pigment is less than the second predetermined amount, the first estimated model is to be used for estimation; if the amount of aluminum flakes in the paint is equal to or greater than the first predetermined amount and the amount of black pigment is less than the second predetermined amount, the third estimated model is to be used for estimation; and if the amount of aluminum flakes in the paint is equal to or greater than the first predetermined amount and the amount of black pigment is equal to or greater than the second predetermined amount, the fourth estimated model is to be used for estimation.

[0048] In this case, the first estimation model is generated using data from the dataset where the amount of aluminum flakes in the paint is less than a first predetermined amount and the amount of black pigment is less than a second predetermined amount. The second estimation model is generated using data from the dataset where the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount and the amount of black pigment is less than a second predetermined amount. The third estimation model is generated using data from the dataset where the amount of aluminum flakes in the paint is less than a first predetermined amount and the amount of black pigment is equal to or greater than a second predetermined amount. The fourth estimation model is generated using data from the dataset where the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount and the amount of black pigment is equal to or greater than a second predetermined amount. In another example, the estimation model generation device 10 generates two estimation models: a first estimation model used when the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount and / or the amount of black pigment is equal to or greater than a second predetermined amount, and a second estimation model used when the amount of aluminum flakes in the paint is less than the first predetermined amount and the amount of black pigment is less than the second predetermined amount. In this case, the first estimation model is generated using data from the dataset in which the amount of aluminum flakes in the paint is equal to or greater than a first predetermined amount, or the amount of black pigment is equal to or greater than a second predetermined amount. The second estimation model is generated using data from the dataset in which the amount of aluminum flakes in the paint is less than a first predetermined amount, and the amount of black pigment is less than a second predetermined amount.

[0049] (Example of experiment) The experiment conducted is described below. In this experiment, the opacity film thickness of the paint film was estimated based on paint data. First, a dataset was created that included paint data and data on the measurement results of the opacity film thickness of the paint. Paint formulation data was used as the paint data. The paint formulation data used included data on the type of glossing agent contained in the paint, the particle size of the glossing agent contained in the paint, the amount of glossing agent contained in the paint, the type of coloring pigment contained in the paint, and the amount of coloring pigment contained in the paint. In this experiment, the estimation model generation device 10 generates a model that determines the model to be used by estimating whether the opacity film thickness is less than 17 micrometers based on the paint data, and determining the estimation model based on whether the estimated opacity film thickness is less than 17 micrometers. In this experiment, the estimation model generation device 10 generates two estimation models: an estimation model used when the opacity film thickness is less than 17 micrometers (a model for concealment) and an estimation model used when the opacity film thickness is 17 micrometers or more (a model for non-concealment). Here, the model for concealment is generated using data from the dataset in which the opacity film thickness is less than 17 micrometers. The non-hiding model is generated using all the data in the dataset. The estimation accuracy was examined by performing 5-fold cross-validation on the dataset data. The dataset contained 1245 data points.

[0050] Figure 10 shows the experimental results. The estimation accuracy of the estimation model was verified using 5-fold cross-validation (5FoldCV). In the graph shown in Figure 10, the vertical axis shows the estimation results by the estimation model, and the horizontal axis shows the measured values. The training data (train) and test data (test) are shown at different points in the graph. The mean absolute percentage error (MAPE) calculated for the training data and test data for each validation is shown in the graph. In this experiment, the average MAPE for the test data was 11.7%.

[0051] <Other Embodiments> Although one embodiment of this invention has been described in detail above with reference to the drawings, the specific configuration is not limited to that described above, and various design changes can be made without departing from the spirit of this invention.

[0052] In the embodiments described above, the case in which the estimation model generation device 10 generates a model for estimating barrier performance and the estimation device 20 estimates the barrier performance was explained, but it is not limited to this. The estimation model generation device 10 may generate a model for estimating coating performance, and the estimation device 20 may estimate coating performance other than barrier performance. In addition to barrier performance, coating performance also includes coating hardness, color, surface characteristics, adhesion, glass transition temperature of the coating, coating properties (viscoelasticity, smoothness, flexibility, etc.), processability, corrosion resistance of processed parts, abrasion resistance, biodegradability, rust prevention, tactile feel, moisture resistance, salt water resistance, low-temperature impact resistance, low-temperature flexibility, low-temperature brittleness, cold resistance, heat resistance, weathering resistance, scratch resistance, non-flammability, flame retardancy, thermal shock resistance, suitability for recoating, crack resistance, and water whitening resistance, and these can also be estimated in the same way as barrier performance by the embodiments described above. The estimation model generation device 10 may generate multiple estimation models that differ depending on the range of values ​​indicated by the coating performance, and the estimation device 20 may determine which estimation model to use for estimating the coating performance based on the paint data and estimate the coating performance. In addition, the candidate estimation models used for estimating the coating performance may include estimation models that do not use elements from the paint data that have little influence on the coating performance. Furthermore, the candidate estimation models used for estimating the coating performance may include three or more estimation models that differ from several elements from the elements included in the paint data.

[0053] All or part of the functions of the estimation model generation device 10 and estimation device 20 in the above-described embodiment may be implemented by a computer. In that case, the functions may be implemented by recording a program for implementing these functions on a computer-readable recording medium, loading the program recorded on this recording medium into a computer system, and executing it. Here, "computer system" includes the OS and peripheral hardware. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and recording devices such as hard disks built into a computer system. In addition, "computer-readable recording medium" may include those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside a computer system that acts as a server or client in such a case. Furthermore, the above-mentioned program may be for implementing some of the functions described above, and may also be able to implement the above-mentioned functions in combination with a program already recorded in the computer system. Furthermore, all or part of the functions of the estimation model generation device 10 and estimation device 20 may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array). [Explanation of Symbols]

[0054] 10 Estimation model generation device, 110 Data set acquisition unit, 120 Estimation model generation unit, 130 Estimation model output unit, 140 Model used determination model generation unit, 150 Model used determination model output unit, 20 Estimation device, 210 Paint data acquisition unit, 220 Barrier performance measurement result estimation unit, 230 Barrier performance measurement result output unit, 240 Model used determination unit, 290 Storage unit

Claims

1. Using a coating performance estimation model learned from variables including information on the paint formulation and / or properties of the paint and the measurement results of the coating performance, the coating performance estimation model outputs an estimated value of the measurement results of the coating performance by inputting information on the paint formulation and / or properties of the paint. Estimation method.

2. An estimation method using a model for determining the model to be used and a plurality of the coating performance estimation models, wherein the model for determining the model to be used is a model that, upon input of information regarding the formulation and / or properties of the paint, determines which of the plurality of coating performance estimation models will be used for estimation. By inputting information regarding the paint formulation and / or properties of the paint into the aforementioned model determination model, the model used for estimation is determined from among the multiple coating performance estimation models. By inputting information regarding the paint formulation and / or properties of the paint into the determined coating performance estimation model, an estimated value of the measured coating performance is output. The estimation method according to claim 1.

3. The aforementioned model selection model calculates estimated values ​​of the measured coating performance and determines which of the multiple coating performance estimation models will be used for the estimation based on the range of the estimated values. The estimation method according to claim 2.

4. The aforementioned multiple coating performance estimation models are coating performance estimation models that are generated based on data included in the dataset, where the values ​​indicating the measured coating performance fall within the corresponding ranges. The estimation method according to claim 2.

5. The aforementioned plurality of coating performance estimation models include a coating performance estimation model generated based on data in the dataset whose values ​​indicating the measured results of coating performance fall within a corresponding range, and a coating performance estimation model generated based on all the data in the dataset. The estimation method according to claim 2.

6. The aforementioned model used for determining the model is a model generated by machine learning. The estimation method according to claim 2.

7. The information regarding the formulation of the aforementioned paint is, The information includes at least one piece of information selected from the group consisting of: the type of glossing agent contained in the paint, the amount of glossing agent contained in the paint, the color of the glossing agent contained in the paint, the particle size of the glossing agent contained in the paint, the thickness of the glossing agent contained in the paint, the type of surface treatment of the glossing agent contained in the paint, the type of coloring pigment contained in the paint, the amount of coloring pigment contained in the paint, the color of the coloring pigment contained in the paint, the chemical structure of the coloring pigment contained in the paint, the particle size of the coloring pigment contained in the paint, the type of resin contained in the paint, the amount of resin contained in the paint, the chemical structure of the resin contained in the paint, the type of dispersant contained in the paint, the amount of dispersant contained in the paint, the chemical structure of the dispersant contained in the paint, the amount of thickener contained in the paint, and the chemical structure of the thickener contained in the paint. The estimation method according to any one of claims 1 to 6.

8. The information relating to the properties of the aforementioned paint is, The paint includes at least one selected from the group consisting of the viscosity of the paint, the elasticity of the paint, the solids content of the paint, the pH of the paint, the color of the paint, the transmittance of the paint, the surface tension of the paint, and the drying conditions of the paint. The estimation method according to any one of claims 1 to 6.

9. The aforementioned coating performance refers to the electromagnetic wave shielding performance of the coating. The estimation method according to any one of claims 1 to 6.

10. The electromagnetic wave shielding performance of the aforementioned coating film includes at least one selected from the group consisting of black and white opacity, process opacity, light transmittance, and heat shielding properties. The estimation method according to claim 9.

11. The aforementioned coating performance refers to the electromagnetic wave shielding performance of the coating. The information relating to the formulation of the aforementioned paint includes information on the color of the coloring pigment and the amount of coloring pigment used. The aforementioned model determination model determines the coating performance estimation model based on the information regarding the amount of black pigment and / or aluminum flakes included in the information regarding the color and amount of the coloring pigment. The estimation method according to claim 2.

12. The aforementioned coating performance refers to the electromagnetic wave shielding performance of the coating. The information relating to the formulation of the aforementioned paint includes information on the color of the coloring pigment and the amount of coloring pigment used. At least one of the multiple estimation models estimates the measurement results of the electromagnetic wave shielding performance of the coating film without using information on the proportion of at least one pigment from brown, red, and yellow. The estimation method according to claim 2.

13. By acquiring and training information on the paint formulation and / or properties of the paint and the measurement results of the coating performance, This system generates a machine learning model that takes information on the paint's composition and / or properties as input and outputs an estimated value of the measured coating performance. A method for generating a coating performance estimation model.

14. By acquiring and training information on the paint formulation and / or the properties of the paint and the measurement results of the coating performance, Based on information on the paint formulation and / or properties of the paint, a model determination model is generated to determine the estimation model used to estimate the measurement results of the coating performance. Method for generating a model for determining which model to use.

Images