Coating film characteristic estimation method and coating film characteristic estimation model generation method

Abstract

Coating properties can be estimated before coating is applied. [Solution] A coating film characteristic estimation method using a machine learning model in which information regarding painting conditions, paint properties, paint formulation, coating film characteristic measurement conditions and coating film characteristic measurement results is learned as variables, and by inputting information regarding the painting conditions, paint properties, paint formulation and coating film characteristic measurement conditions into the machine learning model, an estimated value of the coating film characteristic measurement results is output.

Description

[Technical field]

[0001] The present invention relates to a coating property estimation method and a coating property estimation model generation method. [Background technology]

[0002] When painting automobiles and their parts, a multi-layer coating is usually formed by sequentially applying electrodeposition coating, undercoating, base coating, and clear coating to the substrate. As a result of painting, the painted object has film properties such as smoothness and gloss. Understanding these properties is very important because they are closely related to the appearance and function of the coating. However, it is difficult to estimate the coating film properties before painting because the properties change in a complex manner depending on the combination of paints used, the type of substrate, the painting conditions, etc. Therefore, in order to know the coating film properties, it is necessary to actually prepare the paint and paint it, which is a problem that requires a huge amount of work time. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2021-140420 A Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present invention is to provide a coating film characteristic estimation method and a coating film characteristic estimation model generation method that can estimate coating film characteristics before painting is performed. [Means for solving the problem]

[0005] One aspect of the present invention is a coating film characteristic estimation method, which uses a machine learning model in which information regarding painting conditions, paint properties, paint formulation, coating film characteristic measurement conditions, and coating film characteristic measurement results is learned as variables, and outputs an estimated value of the coating film characteristic measurement results by inputting information regarding the painting conditions, paint properties, paint formulation, and coating film characteristic measurement conditions into the machine learning model. Effect of the Invention

[0006] According to the present invention, it is possible to estimate the coating film characteristics before coating is performed. [Brief description of the drawings]

[0007] [Figure 1] 1 is a diagram showing an example of the configuration of a coating film characteristic estimation model generating device 10 according to a first embodiment. [Diagram 2] 4 is a flowchart showing the operation of the coating film characteristic estimation model generating device 10 according to the first embodiment. [Diagram 3] 1 is a diagram showing an example of the configuration of a coating film characteristic estimation device 20 according to a first embodiment. [Figure 4] 3 is a flowchart showing the operation of the coating film property estimation device 20 according to the first embodiment. [Diagram 5] FIG. 11 is a diagram showing a data set in the second to fourth embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. (First embodiment) FIG. 1 is a diagram showing an example of the configuration of a coating film characteristic estimation model generating device 10 according to the first embodiment. The coating film characteristic estimation model generating device 10 includes a data set acquisition unit 110, a machine learning model generating unit 120, and a machine learning model output unit 130.

[0009] The data set acquisition unit 110 acquires a data set. The data set includes, as variables, information on painting conditions, paint properties, paint blends, coating film property measurement conditions, and coating film property measurement results. Hereinafter, the painting conditions and paint properties may be collectively referred to as test conditions.

[0010] The coating properties to be estimated preferably include at least one of smoothness, gloss and color. Smoothness can be measured using a Wave-scan, Wave-scan dual (both manufactured by BYK Gardner), etc. Smoothness is expressed as the amplitude of a specific wavelength range, and when measured using a Wave-scan, it is expressed in six stages, namely Dullness / Wa / Wb / Wc / Wd / We, or in two stages, namely SW / LW. Gloss, also known as luster or gloss, can be measured using a micro-tri-gloss (manufactured by BYK Gardner) or the like. When measuring using a micro-tri-gloss, gloss can be measured at three angles of 20°, 60°, and 85°, and in this embodiment, it is preferable to include information on 20° and 60°. Color is measured using a multi-angle spectrophotometer, and can be measured using BYK-Mac i (manufactured by BYK), MA-68II (manufactured by X-Rite), etc. Color-related data includes data that represents lightness, saturation, and hue, or data that can specify color by calculation. For example, data based on a color system such as the XYZ color system (X, Y, Z values), the RGB color system, the L*a*b* color system (L*, a*, b* values), the Hunter Lab color system (L, a, b values), the L*C*h color system (L*, C*, h values) defined by CIE (1994), and the Munsell color system (H, V, C values) can be used. Among these, data based on the L*a*b* color system or the L*C*h color system is preferably included. When measuring using BYK-Mac i, color can be measured at six angles: -15°, 15°, 25°, 45°, 75°, and 110°, and when measuring using MA-68II, color can be measured at five angles: 15°, 25°, 45°, 75°, and 110°.In this embodiment, it is preferable to include color information from three angles: 25°, 45°, and 75°. The coating film properties to be estimated are preferably the properties of a multi-layer coating film formed by sequentially applying an undercoat coat, a base coat, and a clear coat.

[0011] The coating conditions are conditions related to the coating work. In this embodiment, the information on the coating conditions preferably includes at least one of the following groups, which differ depending on the coating process: paint type, coating film structure, temperature and humidity in the coating booth, discharge amount, gun speed, film thickness, and baking conditions. The painting process includes, for example, at least one of an undercoat paint painting process, a base paint painting process, and a clear paint painting process. The paint type includes information regarding the number of paint types used in each painting process. The coating structure includes information regarding the number of times each paint is applied (number of layers) in each coating process. The paint booth temperature and humidity refers to the temperature and humidity inside the booth used for painting in each painting process, and may include the amount of water vapor inside the booth calculated from the temperature and humidity as a variable. The discharge amount is the amount of paint discharged per unit time from a coating machine used for coating in each coating process. The gun speed is the speed at which the sprayer used for painting in each painting process moves during painting. The film thickness is the thickness of the coating film after the paint is applied and dried in each coating process. Baking conditions refer to whether or not baking is performed after the paint is applied in each painting process, and / or the heating conditions when baking and drying to form a coating film. Baking conditions usually include the setting time before placing the coating in the heating and drying oven after painting, the temperature during setting, the time the coating is placed in the heating and drying oven, the temperature of the heating and drying oven, and the orientation of the coating film surface (horizontal or vertical) during setting and heating and drying as variables. Among these, it is preferable to include at least one of the orientations of the coating film surface (horizontal or vertical) during painting, setting, and heating and drying as variables, it is more preferable to include at least one of the orientations of the coating film surface during setting and heating and drying as variables, and it is particularly preferable to include both the orientations of the coating film surface during setting and heating and drying as variables. The above-mentioned painting conditions are often different for each coating film to be applied, and must be set for each painting process. Specifically, for example, the painting conditions for the undercoat painting process, the base coating process, and the clear coating process are each input as variables.

[0012] The paint properties are specific values ​​related to the paint, and preferably include at least one of viscosity and solids concentration. The viscosity can be measured without restriction as long as it is measured by a conventionally known method. For the undercoat paint and base paint, it is preferable to use a viscosity measured with a B-type rotational viscometer, and for the clear paint, it is preferable to use a viscosity measured with a Ford cup. The viscosity measured using a Brookfield type rotational viscometer is defined as the viscosity 1 minute after the start of measurement under certain conditions. Specifically, the viscosity is defined as the viscosity 1 minute after the start of measurement using a Brookfield type viscometer at a temperature of 20°C and 6 rpm (also referred to as the "B6 value" in this specification). It may also be defined as the viscosity 1 minute after the start of measurement using a Brookfield type viscometer at a temperature of 20°C and 6 rpm (also referred to as the "B60 value" in this specification). The viscometer used in this case is LVDV-I (trade name, Brookfield type viscometer). Viscosity may also include the "Ti value (thixotropic index)" as a variable. The "Ti value (thixotropic index)" is described in JIS K 5101-6-2 (2004), Pigment Testing Method, Section 2, Rotational Viscometer Method, and can be determined by measuring the viscosity (mPa s) at 20°C and rotation speeds of 6 rpm and 60 rpm using a B-type viscometer, and calculating the viscosity measurement value at "viscosity measurement value at 6 rpm / viscosity measurement value at 60 rpm." The solids concentration means the concentration of the residue after removing the volatile components, and the residue may be either solid or liquid at room temperature. For example, it can be calculated by measuring the amount of the remaining components after heating the sample at 105°C for 3 hours to remove the volatile components. In most cases, the above-mentioned paint properties differ for each paint, so in the case of a multi-layer process, they must be set for each painting process. Specifically, for example, the paint properties of the undercoat paint, the paint properties of the base paint, and the paint properties of the clear paint are each input as variables.

[0013] The formulation of a paint refers to the components that make up the paint and their amounts. The components that make up the paint include resins, pigments, solvents, additives, etc. In particular, the information regarding the paint formulation preferably includes information regarding the type of pigment and information related to its amount, and more preferably includes information related to at least one of the size of the lustrous pigment, the color of the lustrous pigment, the color of the color pigment, and the chemical structure of the color pigment. The above-mentioned paint formulations differ for each paint, and therefore need to be set for each painting process in the case of a multi-layer process. Specifically, for example, the information on the paint formulation includes information on the formulation of the undercoat paint, information on the formulation of the base paint, and information on the formulation of the clear paint. The undercoat painting process, base painting process, and clear painting process may involve multi-layer painting of two or more types of paint in one process, and in that case, it is preferable to input information for each paint. In this embodiment, from the viewpoint of improving prediction accuracy, it is preferable that the information on the paint formulation includes information on the formulation of the base paint.

[0014] The information on the coating film characteristic measurement conditions is information on the conditions when measuring the coating film characteristics. When the coating film characteristic to be estimated is smoothness, the equipment used, the wavelength to be measured, etc. are input, and when the coating film characteristic is gloss or color, the equipment used, the measurement angle, etc. are input. The coating film characteristic measurement results are the results of measuring the coating film characteristics of a coating film applied under the above-mentioned coating conditions, paint properties, and paint blend, and the measurement results corresponding to the above-mentioned coating film characteristic measurement conditions are input.

[0015] The information contained in the data set may be processed by a computer or the like.

[0016] The machine learning model generation unit 120 generates a machine learning model based on the data set. The machine learning model is a model that outputs an estimated value of a coating film property measurement result when information on painting conditions, paint properties, paint formulation, and coating film property measurement conditions is input. The machine learning model generation unit 120 generates a machine learning model using information on painting conditions, paint properties, paint formulation, and coating film property measurement conditions as explanatory variables and coating film property measurement results as objective variables. Examples of algorithms for machine learning models include decision trees, decision trees using gradient boosting, linear regression, logistic regression, simple perceptron, MLP, neural networks, support vector machines, random forests, Gaussian processes, Bayesian networks, k-nearest neighbor methods, Ridge regression, Lasso regression, graph neural networks, K-means methods, naive Bayes, CNN, and other algorithms used in machine learning, and these algorithms may be used alone or in combination. Note that in the present invention, it is preferable to use a decision tree using gradient boosting.

[0017] The machine learning model output unit 130 outputs the generated machine learning model.

[0018] 2 is a flowchart showing the operation of the coating film characteristic estimation model generating device 10 according to the first embodiment. The dataset acquisition unit 110 acquires a dataset (step S101). The machine learning model generating unit 120 generates a machine learning model based on the dataset (step S102). The machine learning model output unit 130 outputs the generated machine learning model (step S103).

[0019] 3 is a diagram showing an example of the configuration of the coating film property estimation device 20 according to the first embodiment. The coating film property estimation device 20 includes a coating data acquisition unit 210, a coating film property measurement result estimation unit 220, a coating film property measurement result output unit 230, and a storage unit 290. The storage unit 290 stores the machine learning model output by the coating film property estimation model generation device 10.

[0020] The painting data acquisition unit 210 acquires painting data. The painting data includes painting conditions, paint properties, paint mix, and coating film characteristic measurement conditions, which are explanatory variables of the machine learning model.

[0021] The coating film characteristic measurement result estimation unit 220 estimates the coating film characteristic measurement result by inputting the painting data into a machine learning model and outputting an estimated value of the coating film characteristic measurement result.

[0022] The coating film property measurement result output unit 230 outputs an estimated value of the coating film property measurement result output by the machine learning model. The estimated value of the coating film property measurement result output from the coating film property measurement result output unit 230 is recorded in the memory unit 290 together with the corresponding painting data, for example. The estimated value of the coating film property measurement result output from the coating film property measurement result output unit 230 is input to and displayed on, for example, an external display device.

[0023] 4 is a flowchart showing the operation of the coating film property estimation device 20 according to the first embodiment. The coating data acquisition unit 210 acquires coating data (step S201). The coating film property measurement result estimation unit 220 estimates the coating film property measurement result using a machine learning model based on the coating data (step S202). The coating film property measurement result output unit 230 outputs an estimated value of the coating film property measurement result (step S203).

[0024] As described above, the coating film characteristic estimation device 20 can estimate coating film characteristic measurement conditions from painting conditions, paint properties, and paint formulation using a machine learning model, thereby estimating coating film characteristic measurement conditions before painting.

[0025] Second embodiment The coating film characteristic estimation model generating device 10 according to the second embodiment will be described below. In the second embodiment, the dataset acquired by the dataset acquiring unit 110 includes, in addition to the dataset in the first embodiment, differences in test conditions and differences in coating film characteristic measurement results corresponding to the differences in test conditions when the paint blend and coating film characteristic measurement conditions are the same. Fig. 5 is a diagram showing the dataset in the second embodiment.

[0026] The data set includes test conditions B1 and coating film property measurement results F1, and test conditions B2 and coating film property measurement results F2, when the paint formulation is C1 and the coating film property measurement conditions are E1. The data set includes the difference in test conditions ΔB and the difference in coating film property measurement results ΔF, where ΔB and ΔF are the difference between B1 and B2 and the difference between F1 and F2, respectively.

[0027] The machine learning model generation unit 120 according to the second embodiment generates a machine learning model so as to output a second coating film characteristic measurement result corresponding to the second test condition when a paint formulation, coating film characteristic measurement conditions, first test conditions, a difference between the first test conditions and the second test conditions, and a first coating film characteristic measurement result corresponding to the first test conditions are input. The first test conditions and the second test conditions are different test conditions. To explain using the dataset shown in Figure 5, the machine learning model generation unit 120 according to the second embodiment generates a machine learning model that outputs F2 when C1, E1, B1, ΔB, and F1 are input.

[0028] The machine learning model generation unit 120 of the second embodiment may generate a machine learning model to output the difference between the second coating film characteristic measurement result and the first coating film characteristic measurement result when a paint formulation, coating film characteristic measurement conditions, first test conditions, the difference between the first test conditions and the second test conditions, and a first coating film characteristic measurement result corresponding to the first test conditions are input. To explain using the dataset shown in Figure 5, the machine learning model generation unit 120 of the second embodiment may generate a machine learning model that outputs ΔF when C1, E1, B1, ΔB, and F1 are input.

[0029] The coating film characteristic estimation device 20 according to the second embodiment will be described below. The coating data acquisition unit 210 according to the second embodiment acquires the paint formulation, coating film characteristic measurement conditions, known coating film characteristic measurement results, known test conditions, and differences between the known test conditions and unknown test conditions. Here, the known coating film characteristic measurement results are the coating film characteristic measurement results under the paint formulation, coating film characteristic measurement conditions, and known test conditions acquired by the coating data acquisition unit 210. The known coating film characteristic measurement results are acquired in advance, similar to the data set. The unknown test conditions are test conditions in which the coating film characteristic measurement results under the unknown test conditions, the paint blend, and the coating film characteristic measurement conditions acquired by the coating data acquisition unit 210 are unknown. Hereinafter, the coating film characteristic measurement results under the paint blend, the coating film characteristic measurement conditions, and the unknown test conditions acquired by the coating data acquisition unit 210 are referred to as unknown coating film characteristic measurement results.

[0030] The coating film characteristic measurement result estimation unit 220 of the second embodiment inputs the paint formulation, coating film characteristic measurement conditions, known test conditions, the difference between the known test conditions and unknown test conditions, and known coating film characteristic measurement results into the machine learning model generated by the machine learning model generation unit of the second embodiment, and outputs an estimated value of the unknown coating film characteristic measurement result.

[0031] The coating film characteristic measurement result estimation unit 220 of the second embodiment may input the paint formulation, coating film characteristic measurement conditions, known test conditions, the difference between the known test conditions and unknown test conditions, and the known coating film characteristic measurement results into the machine learning model generated by the machine learning model generation unit of the second embodiment, and output an estimated value of the difference between the known coating film characteristic measurement results and the estimated value of the unknown coating film characteristic measurement results.

[0032] (Third embodiment) The coating film characteristic estimation model generating device 10 according to the third embodiment will be described below. In the third embodiment, the dataset acquired by the dataset acquisition unit 110 includes, in addition to the dataset in the first embodiment, a difference in test conditions and a difference in coating film characteristic measurement results corresponding to the difference in test conditions when the paint blend and coating film characteristic measurement conditions are the same. Fig. 5 is a diagram showing a dataset in the third embodiment. Explanation of symbols in Fig. 5 will be omitted as they are the same as those in the second embodiment.

[0033] The machine learning model generation unit 120 according to the third embodiment generates a machine learning model so as to output a second coating property measurement result corresponding to the second test condition when a paint formulation, coating property measurement conditions, second test conditions, a difference between the second test conditions and the first test conditions, and a first coating property measurement result corresponding to the first test conditions are input. The first test conditions and the second test conditions are different test conditions. To explain using the dataset shown in Figure 5, the machine learning model generation unit 120 according to the second embodiment generates a machine learning model that outputs F2 when C1, E1, B2, ΔB, and F1 are input.

[0034] The machine learning model generation unit 120 of the third embodiment may generate a machine learning model to output the difference between the second coating characteristic measurement result and the first coating characteristic measurement result when a paint formulation, coating film characteristic measurement conditions, second test conditions, the difference between the second test conditions and the first test conditions, and the first coating film characteristic measurement result are input. To explain using the dataset shown in Figure 5, the machine learning model generation unit 120 of the third embodiment may generate a machine learning model that outputs ΔF when C1, E1, B2, ΔB, and F1 are input.

[0035] The coating film characteristic estimation device 20 according to the third embodiment will be described below. The coating data acquisition unit 210 according to the third embodiment acquires paint formulation, coating film characteristic measurement conditions, unknown test conditions, differences between the unknown test conditions and known test conditions, and known coating film characteristic measurement results.

[0036] The coating film characteristic measurement result estimation unit 220 of the third embodiment inputs the paint formulation, coating film characteristic measurement conditions, unknown test conditions, the difference between the unknown test conditions and the known test conditions, and known coating film characteristic measurement results into the machine learning model generated by the machine learning model generation unit of the third embodiment, and outputs an estimated value of the unknown coating film characteristic measurement result.

[0037] The coating film characteristic measurement result estimation unit 220 of the third embodiment may input the paint formulation, coating film characteristic measurement conditions, unknown test conditions, the difference between the unknown test conditions and the known test conditions, and the known coating film characteristic measurement results into the machine learning model generated by the machine learning model generation unit of the third embodiment, and output the difference between the known coating film characteristic measurement results and the estimated value of the unknown coating film characteristic measurement results.

[0038] (Fourth embodiment) Hereinafter, a coating film property estimation device 20 according to the fourth embodiment will be described. The coating film property estimation device 20 according to the fourth embodiment calculates an estimated value of a second coating film property measurement result by using both the machine learning model according to the second embodiment and the machine learning model according to the third embodiment.

[0039] The coating film characteristic measurement result estimation unit 220 of the fourth embodiment inputs the paint formulation, coating film characteristic measurement conditions, known test conditions, the difference between the known test conditions and unknown test conditions, and known coating film characteristic measurement results into the machine learning model of the second embodiment, and outputs an estimate of the unknown coating film characteristic measurement result (hereinafter referred to as the second estimate).

[0040] In addition, the coating film characteristic measurement result estimation unit 220 of the fourth embodiment inputs the paint formulation, coating film characteristic measurement conditions, unknown test conditions, the difference between the unknown test conditions and the known test conditions, and the known coating film characteristic measurement results into the machine learning model of the third embodiment, and outputs an estimated value of the unknown coating film characteristic measurement result (hereinafter referred to as the third estimated value).

[0041] The coating property measurement result estimating unit 220 according to the fourth embodiment calculates an estimate of the unknown coating property measurement result based on the second estimate value and the third estimate value. For example, the coating property measurement result estimating unit 220 according to the fourth embodiment calculates an average value of the second estimate value and the third estimate value as the estimate of the unknown coating property measurement result.

[0042] The coating film characteristic measurement result estimation unit 220 of the fourth embodiment may input the paint formulation, coating film characteristic measurement conditions, known test conditions, the difference between the known test conditions and unknown test conditions, and the known coating film characteristic measurement results into the machine learning model of the second embodiment, and output the difference between the known coating film characteristic measurement results and the estimated value of the unknown coating film characteristic measurement results (hereinafter referred to as the second difference estimated value).

[0043] In addition, the coating film characteristic measurement result estimation unit 220 of the fourth embodiment may input the paint formulation, coating film characteristic measurement conditions, unknown test conditions, the difference between the unknown test conditions and the known test conditions, and the known coating film characteristic measurement results into the machine learning model of the third embodiment, and output the difference between the known coating film characteristic measurement results and the estimated value of the unknown coating film characteristic measurement results (hereinafter referred to as the third difference estimated value).

[0044] The coating film characteristic measurement result estimating unit 220 according to the fourth embodiment calculates an estimate of the difference between the known coating film characteristic measurement result and the unknown coating film characteristic measurement result based on the second difference estimate value and the third difference estimate value. For example, the coating film characteristic measurement result estimating unit 220 according to the fourth embodiment may calculate an average value of the second difference estimate value and the third difference estimate value as an estimate of the difference between the known coating film characteristic measurement result and the unknown coating film characteristic measurement result.

[0045] In the second to fourth embodiments, the coating film characteristic measurement result estimation unit 220 may calculate a correction value for the estimated value of the unknown coating film characteristic measurement result or a correction value for the difference between the known coating film characteristic measurement result and the estimated value of the unknown coating film characteristic measurement result, and correct the difference between the known coating film characteristic measurement result and the estimated value of the unknown coating film characteristic measurement result or the estimated value of the unknown coating film characteristic measurement result.

[0046] In the second to fourth embodiments, the coating film property measurement result estimating unit 220 inputs the paint formulation, the coating film property measurement conditions, the known test conditions, and the known coating film property measurement results into a machine learning model to output an estimate of the known coating film property measurement results. At this time, the coating film property measurement result estimating unit 220 inputs 0 as the difference in the test conditions.

[0047] In the second to fourth embodiments, the coating film characteristic measurement result estimating unit 220 calculates a correction value based on the known coating film characteristic measurement result and the estimated value of the known coating film characteristic measurement result. For example, the coating film characteristic measurement result estimating unit 220 calculates the difference between the known coating film characteristic measurement result and the estimated value of the known coating film characteristic measurement result as the correction value.

[0048] Other Embodiments Although one embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to the above, and various design changes, etc. are possible within the scope that does not deviate from the gist of the present invention.

[0049] The coating film characteristic estimation model generating device 10 and the coating film characteristic estimating device 20 in the above-mentioned embodiment may be entirely or partially realized by a computer. In this case, a program for realizing the function may be recorded in a computer-readable recording medium, and the program recorded in the recording medium may be read into a computer system and executed to realize the function. The term "computer system" as used herein includes an OS and hardware peripherals. The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and recording devices such as hard disks built into a computer system. The term "computer-readable recording medium" may also include a medium that dynamically holds a program for a short period of time, such as a communication line when a program is transmitted via a network such as the Internet or a communication line such as a telephone line, or a medium that holds a program for a certain period of time, such as a volatile memory inside a computer system that is a server or client in such a case. The above-mentioned program may be a program for realizing a part of the above-mentioned function, or may be a program that can realize the above-mentioned function in combination with a program already recorded in the computer system. In addition, all or part of the functions of the coating film property estimation model generating device 10 and the coating film property estimation device 20 may be realized using a programmable logic device such as an FPGA (Field Programmable Gate Array). [Explanation of symbols]

[0050] 10 Coating film characteristic estimation model generating device, 110 Data set acquisition unit, 120 Machine learning model generating unit, 130 Machine learning model output unit, 20 Coating film characteristic estimating device, 210 Painting data acquisition unit, 220 Coating film characteristic measurement result estimating unit, 230 Coating film characteristic measurement result output unit, 290 Memory unit

Claims

1. Using a machine learning model trained with variables including information on painting conditions, paint properties, paint formulation, coating film property measurement conditions, and coating film property measurement results, inputting information on the painting conditions, paint properties, paint formulation, and coating film property measurement conditions into the machine learning model, outputting an estimated value of the coating film property measurement result; A coating film characteristic estimation method, comprising: The variables include information regarding differences in test conditions, including painting conditions and paint properties, when the paint formulation and coating film property measurement conditions are the same, and differences in coating film property measurement results corresponding to the differences in the test conditions, by inputting information on a first test condition, a difference between the first test condition and a second test condition, the second test condition being different from the first test condition, the difference, a paint formulation, coating film characteristic measurement conditions, and a first coating film characteristic measurement result corresponding to the first test condition, into the machine learning model, an estimated value of a second coating film characteristic measurement result corresponding to the second test condition and / or a difference between the first coating film characteristic measurement result and an estimated value of the second coating film characteristic measurement result; Paint film property estimation method.

2. A coating film characteristic estimation method using a machine learning model trained with variables including information on coating conditions, paint properties, paint formulation, coating film characteristic measurement conditions, and coating film characteristic measurement results, and inputting information on the coating conditions, paint properties, paint formulation, and coating film characteristic measurement conditions into the machine learning model to output an estimated value of a coating film characteristic measurement result, The variables include information regarding differences in test conditions, including painting conditions and paint properties, when the paint formulation and coating film property measurement conditions are the same, and differences in coating film property measurement results corresponding to the differences in the test conditions, by inputting information on a second test condition, a difference between the second test condition and the first test condition, the difference being a test condition different from the second test condition, a paint formulation, coating film characteristic measurement conditions, and a first coating film characteristic measurement result corresponding to the first test condition, into the machine learning model, an estimated value of the second coating film characteristic measurement result corresponding to the second test condition and / or a difference between the estimated value of the second coating film characteristic measurement result and the first coating film characteristic measurement result; Paint film property estimation method.

3. inputting information regarding the second test conditions, the difference between the first test conditions and the second test conditions, the paint formulation, the coating film characteristic measurement conditions, and the first coating film characteristic measurement result into the machine learning model, thereby outputting an estimated value of the second coating film characteristic measurement result and / or a difference between the first coating film characteristic measurement result and the estimated value of the second coating film characteristic measurement result; calculating the estimated value of the second coating film characteristic measurement result and / or the difference between the estimated values ​​of the first coating film characteristic measurement result and the second coating film characteristic measurement result based on the estimated value of the second coating film characteristic measurement result and / or the difference between the estimated values ​​of the first coating film characteristic measurement result and the second coating film characteristic measurement result outputted to the machine learning model by inputting the first test conditions, and the estimated value of the second coating film characteristic measurement result and / or the difference between the estimated values ​​of the first coating film characteristic measurement result and the second coating film characteristic measurement result outputted to the machine learning model by inputting the second test conditions; The coating film characteristic estimating method according to claim 1 .

4. inputting information on the first test conditions, a paint formulation, coating film characteristic measurement conditions, and the first coating film characteristic measurement results into the machine learning model, thereby outputting an estimated value of the first coating film characteristic measurement result corresponding to the first test conditions; Calculating a correction value based on the first coating film characteristic measurement result and an estimated value of the first coating film characteristic measurement result; correcting the estimated value of the second coating film characteristic measurement result and / or the difference between the first coating film characteristic measurement result and the estimated value of the second coating film characteristic measurement result using the correction value; The coating film characteristic estimating method according to any one of claims 1 to 3.

5. A coating film characteristic estimation method described in any one of claims 1 to 3, wherein the coating film characteristics include at least one of smoothness, gloss and color.

6. A method for estimating coating film characteristics described in any one of claims 1 to 3, wherein the painting conditions include at least one of a group that differs depending on the painting process, consisting of painting booth temperature and humidity, discharge amount, gun speed, film thickness and baking conditions.

7. A method for estimating coating film characteristics described in any one of claims 1 to 3, wherein the paint properties include at least one of viscosity and solids concentration.

8. A method for estimating coating film characteristics described in any one of claims 1 to 3, wherein the paint formulation includes at least one of the size of the lustrous material, the color of the lustrous material, the color of the coloring pigment, and the chemical structure of the coloring pigment.

9. Information on painting conditions, paint properties, paint formulation, coating film characteristic measurement conditions, and coating film characteristic measurement results are acquired as variables. Using the variables, a machine learning model is generated that outputs an estimated value of a coating film characteristic measurement result when information on painting conditions, paint properties, paint formulation, and coating film characteristic measurement conditions is input. A coating film characteristic estimation model generation method, comprising the steps of: The variables include information regarding differences in test conditions, including painting conditions and paint properties, and differences in coating film characteristic measurement results corresponding to the differences in test conditions, when the paint formulation and coating film characteristic measurement conditions are the same, By inputting information regarding a first test condition, a difference between the first test condition and a second test condition, the second test condition being different from the first test condition, the difference, a paint formulation, a coating film characteristic measurement condition, and a first coating film characteristic measurement result corresponding to the first test condition, a machine learning model is generated that outputs an estimated value of a second coating film characteristic measurement result corresponding to the second test condition and / or a difference between the first coating film characteristic measurement result and the estimated value of the second coating film characteristic measurement result. A method for generating a model to estimate coating properties.

10. Information on painting conditions, paint properties, paint formulation, coating film characteristic measurement conditions, and coating film characteristic measurement results are acquired as variables. Using the variables, a machine learning model is generated that outputs an estimated value of a coating film characteristic measurement result when information on painting conditions, paint properties, paint formulation, and coating film characteristic measurement conditions is input. A coating film characteristic estimation model generation method, comprising the steps of: The variables include information regarding differences in test conditions, including painting conditions and paint properties, when the paint formulation and coating film property measurement conditions are the same, and differences in coating film property measurement results corresponding to the differences in the test conditions, By inputting information regarding a second test condition, a difference between the second test condition and a first test condition, the first test condition being different from the second test condition, the difference, a paint formulation, a coating film characteristic measurement condition, and a first coating film characteristic measurement result corresponding to the first test condition, a machine learning model is generated that outputs an estimated value of the second coating film characteristic measurement result corresponding to the second test condition and / or a difference between the estimated value of the second coating film characteristic measurement result and the first coating film characteristic measurement result. A method for generating a model to estimate coating properties.

Images