Method and system for predicting durability of coating film under actual exposure

Abstract

A prescribed artificial intelligence model has inputs that include (a) paint information data relating to a paint composition, and property information data relating to a coating film obtained by applying the paint composition, before actual exposure, and (b) data relating to at least one of an exposure time of the coating film, environmental information relating to the exposure of the coating film, an installation angle and installation azimuth of a coated object on which the coating film has been formed, and an evaluation result of an accelerated test of the coating film, and outputs a result of the durability of the coating film under actual exposure, and a computer calculates and predicts the durability of the coating film under actual exposure by inputting the data (a) and (b) into the artificial intelligence model.

Description

Method and system for predicting durability of coating film under actual exposure

[0001] The present invention relates to a method and system for predicting the durability of a coating film obtained by applying a coating composition under actual exposure.

[0002] Various coating compositions are applied to the walls and roofs of buildings such as houses and offices in order to maintain the quality and appearance of the walls and roofs under conditions where they are exposed to wind, rain, and direct sunlight. Such coating compositions are required to be durable against wind, rain, and sunlight.

[0003] Therefore, various methods for predicting the degree of aging deterioration of the properties of a coating film have been proposed. For example, Patent Document 1 proposes using a machine learning technique to predict the results of an accelerated test as one of the evaluations of an article using an artificial intelligence model with paint information as an explanatory variable and the evaluation of the article as a target variable. Such machine learning techniques are said to enable rapid prediction.

[0004] Japanese Patent Application Laid-Open No. 2023-021558

[0005] However, accelerated testing involves exposing test specimens with a coating on their surface to conditions that artificially amplify the strength of factors such as ultraviolet rays, heat, and water, which cause deterioration of the coating's properties over time, and evaluating the changes in their condition.However, there is often a significant discrepancy between the results and the actual exposure results, and there are cases in which the accelerated testing is not sufficient as an indicator of deterioration over time.

[0006] Therefore, the present invention aims to provide a method for quickly and accurately predicting the changes over time in the properties of a coating film on an object at any location by using a machine learning technique to predict the durability of the coating film under actual exposure.

[0007] The gist and configuration of the present invention are as follows. (1) A method for predicting the durability of a coating film obtained by applying a paint composition under actual exposure using a machine learning technique, the method comprising: a step of predicting durability under actual exposure in a predetermined artificial intelligence model that includes as input (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any one of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and the evaluation results of an accelerated test of the coating film, and outputs the results of the durability of the coating film under actual exposure, by inputting (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any one of data on the exposure time, environmental information on the exposure, the installation angle and installation orientation of the coated object, and the evaluation results of the accelerated test, to calculate and predict the durability of the coating film under actual exposure.

[0008] (2) The input of the artificial intelligence model includes (a) paint information data of the paint composition and property information data of the paint film before actual exposure obtained by applying the paint composition, and (b) at least the exposure time of the paint film, and in the durability prediction step of actual exposure, (a) paint information data of the paint composition and property information data of the paint film before actual exposure obtained by applying the paint composition, and (b) at least the exposure time are input, in the method for predicting the durability of a paint film after actual exposure described in (1).

[0009] (3) The input of the artificial intelligence model includes (a) paint information data of the paint composition and property information data of the paint film before actual exposure obtained by applying the paint composition, and (b) at least data on the exposure time of the paint film and environmental information on the exposure of the paint film, and in the durability prediction step of the actual exposure, (a) paint information data of the paint composition and property information data of the paint film before actual exposure obtained by applying the paint composition, and (b) at least data on the exposure time and environmental information on the exposure, is input.

[0010] (4) A method for predicting the durability of a coating film under actual exposure according to any one of (1) to (3), wherein the paint information data includes at least information regarding the formulation of the paint composition.

[0011] (5) The method for predicting durability of a coating film under actual exposure according to any one of (1) to (4), wherein the environmental information on the exposure of the coating film includes at least one of information on latitude, amount of solar radiation, amount of ultraviolet light, temperature, amount of precipitation, relative humidity, wind speed, amount of sea salt particles attached, and amount of sulfur oxides attached.

[0012] (6) A system for predicting the durability of a coating film obtained by applying a paint composition under actual exposure using a machine learning technique, comprising: (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any one of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and the evaluation results of an accelerated test of the coating film, as inputs; and the system is characterized by comprising an actual exposure durability prediction unit that calculates and predicts the durability of the coating film under actual exposure by inputting, by a computer, (a) paint information data for the paint composition and property information data for the coating film before actual exposure, and (b) at least any one of data on the exposure time, environmental information on the exposure, the installation angle and installation orientation of the coated object, and the evaluation results of the accelerated test.

[0013] According to the present invention, by using a machine learning technique to predict the durability of a coating film under actual exposure, it is possible to provide a method for quickly and accurately predicting the changes over time in the properties of a coating film on an object at any location.

[0014] 1 is a flowchart of a method for predicting the durability of a coating film under actual exposure according to one embodiment of the present invention.

[0015] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0016] <Method for predicting durability of coating film after actual exposure> Figure 1 is a flowchart of a method for predicting durability of coating film after actual exposure according to one embodiment of the present invention. Hereinafter, one embodiment of a method for predicting durability of coating film after actual exposure obtained by applying a paint composition using a machine learning technique will be described by way of example with reference to Figure 1. Note that the method for predicting durability of coating film after actual exposure according to this embodiment can be executed, for example, using a system for predicting durability of coating film after actual exposure according to one embodiment of the present invention described below.

[0017] As shown in FIG. 1 , in this embodiment, first, (a) paint information data of the paint composition and property information data of the coating film obtained by applying the paint composition before actual exposure, and (b) relational data showing the relationship between at least any of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and the evaluation results of the accelerated test of the coating film and the results of the durability of the coating film after actual exposure are prepared (step S101).

[0018] The "paint information data of the paint composition" in (a) above is paint information data of the paint composition in its initial state. Here, the "paint information data of the paint composition" in (a) above preferably includes at least information on the formulation of the paint composition. Specifically, this information can be the names of raw materials (also called compound names) of the paint composition and their blending amounts. Data on the product name (product number) can be used instead of or in addition to the raw material names. In addition, the color (L ＊ Value, a ＊ value, b ＊ The data can also include the colorant's shape (spherical, scaly, fibrous, etc.), average primary particle size, average secondary particle size, average dispersed particle size, particle size distribution, aspect ratio, thickness, etc. Chemical properties can include molecular weight, molecular weight distribution, discoloration temperature, reactivity, etc.

[0019] Furthermore, the "property information data before actual exposure" in (a) above can be data on one or more of the color and gloss of the resulting coating film before actual exposure.

[0020] The color is, for example, L ＊ a ＊ b ＊ L in color space ＊ Value, a ＊ value, b ＊ Value (JIS Z_8781-4 (2013)), X-Y-Z color system, RGB color system, Yxy color system, Hunter L-a-b color system, L ＊ -C ＊ -h ＊ The color can be measured using a known color measurement method, for example, by using a CM-512m3 commercially available from Konica Minolta, Inc., and measuring the L value by irradiating a light source at angles of 25°, 45°, and 75° when the light receiving part perpendicular to the coating film is set to 0°. ＊ Value, a ＊ value, b ＊ The value can be measured. Alternatively, the measurement can be performed using an X-Rite MA68II (manufactured by X-Rite). The measurement angle can be adjusted appropriately depending on the purpose or the equipment used. Any other index can be used. Furthermore, any index can be used, such as reflection spectrum data, which is an index in which the reflection spectrum intensity is expressed as a color in 5 nm increments from 380 nm to 780 nm.

[0021] The gloss is not particularly limited, but can be used as an index. The gloss can be measured using a known gloss measurement method. As an example, the 60° gloss of a coating film formed on a test plate can be measured using a specular gloss meter (VG-7000 gloss meter (manufactured by Nippon Denshoku Industries Co., Ltd.)) in accordance with JIS K 5600-4-7 (specular gloss).

[0022] The "property information data before actual exposure" may include data on the smoothness of the coating film before actual exposure. It is preferable to use the wave scan value as an indicator of smoothness. The wave scan value is preferably one or more of du (wavelength 0.1 mm or less), Wa (wavelength 0.1 to 0.3 mm), Wb (wavelength 0.3 to 1.0 mm), Wc (wavelength 1.0 to 3.0 mm), Wd (wavelength 3.0 to 10.0 mm), We (wavelength 10.0 to 30.0 mm), Lw (wavelength 1.2 to 12 mm), and Sw (wavelength 0.3 to 1.2 mm). The smaller the wave scan value, the less unevenness at that wavelength on the surface and the better the appearance quality of the coating film. These indicators may also be calculated by digital analysis using 3D scanning. Flip-flop properties can be measured using a variable angle color difference meter or the like. Coating abnormalities specifically include unevenness, blisters, cracks, sagging, pinholes, and frame defects. These can also be measured by known methods. "Property information data before actual exposure" can also include other properties such as hiding power, flip-flop property, micro-brilliance, visual appearance, electrical resistance, contact angle, staining property, solar reflectance, ultraviolet transmittance, and viscoelasticity.

[0023] Furthermore, the "property information data of the coating film before actual exposure" in (a) above can also include information on the coating film thickness. Furthermore, if the coating film is made up of multiple layers, it can also include information on the composition and film thickness of each layer. It can also include information on the substrate (substrate) on which the coating film is applied.

[0024] The "exposure time of the coating film" in (b) above is information on the time (period) for which the coating film is exposed to the environment. For example, it may be information on the start date (year / month / day) of exposure and a subsequent date (year / month / day), or it may be information on the difference in time (period) calculated from these.

[0025] The "environmental information on the exposure of the coating film" in (b) preferably includes at least one of information on latitude, amount of solar radiation, amount of ultraviolet light, temperature, amount of precipitation, relative humidity, wind speed, amount of sea salt particles attached, and amount of sulfur oxides attached.

[0026] "Latitude" can be expressed in any unit system, but as an example, "°" can be used. "Solar radiation" is expressed in SRAD (J / cm ２ ) (average over a period or cumulative over a period), etc. can be used. The "ultraviolet ray amount" is UVA / UVB / UV (J / cm ２ ) (period average or period cumulative), etc. can be used. "Temperature" can be expressed in any unit system, but in one example, "°C" (period average or period cumulative) can be used. "Precipitation" can be expressed in any unit system, but in one example, "mm" (period average or period cumulative) can be used. "Relative humidity" can be expressed in any unit system, but in one example, "%" (period average or period cumulative) can be used. "Wind speed" can be expressed in any unit system, but in one example, "m / s" (period average or period cumulative) can be used. "Sea salt particle adhesion amount (chloride)" can be expressed in any unit system, but in one example, "mgNaCl / (m ２ ·day) (average or cumulative) can be used. ２ " etc.)" can be in any unit system, but as an example, "mgSO ２ / (m ２ ·day) (average over a period or cumulative over a period) can be used.

[0027] Regarding the "installation angle and installation orientation of a coated article on which a coating film is formed" in (b), the "installation angle" may be an angle based on any direction, and as an example, the unit may be "°". The "installation orientation" may be an orientation based on any direction, and as an example, the unit may be "°". The "installation angle and installation orientation of a coated article on which a coating film is formed" may be the installation angle and / or installation orientation of a coated article on which a coating film is formed.

[0028] The "evaluation results of accelerated testing of coating film" in (b) above can be evaluated (tested) by SWOM, QUV-A, QUV-B, SUV, Xe, SST, CCT, SDT, hot water resistance test, etc. Here, "SWOM" refers to a sunshine carbon arc lamp weather resistance test, "QUV" refers to a test using an ultraviolet fluorescent lamp weather tester QUV (ASTM G154, JIS K 7350-3), "SUV" is an abbreviation for Super UV and refers to a weather resistance test using a metal halide lamp, and "Xe" refers to a test using a xenon arc. Furthermore, "SST" is an abbreviation for salt spray test, which means a durability test in which sprayed salt water is constantly attached to the test piece; "CCT" is an abbreviation for salt water combined cycle test, which means a durability test in which cycle tests such as salt spray, drying, and wetting are repeated; "SDT" is an abbreviation for salt water hot water test, which means a durability test in which a test piece is immersed in heated salt water hot water; and "hot water resistance test" means a durability test in which a test piece is immersed in hot water maintained at a constant temperature. The "evaluation results of the accelerated test of the coating film" can be deterioration information at any test time (variation values ​​of one or more of color difference, gloss, rust width, blister width, chalking level, peeling area, film thickness, etc.).

[0029] Furthermore, the "durability of the coating film under actual exposure" can be evaluated using one or more of color difference, gloss, rust width, blister width, chalking level, peeling area, and film thickness.

[0030] It is preferable that the relationship data is constantly, timely, or periodically updated to enable more accurate predictions.

[0031] As shown in FIG. 1 , in this embodiment, next, (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) relational data indicating the relationship between at least any of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and the evaluation results of the accelerated test of the coating film and the results of durability of the coating film after actual exposure (i.e., the relational data prepared in step S101) are input into a computer (step S102).

[0032] Next, as shown in Figure 1, a predetermined artificial intelligence model is created by computer machine learning, using (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least one of the following data as input: exposure time of the coating film, environmental information for the exposure of the coating film, installation angle and installation orientation of the coated object on which the coating film is formed, and evaluation results of the accelerated test of the coating film, and outputting the results of durability of the coating film after actual exposure (step S103). Note that in this embodiment, an artificial intelligence model is created in this step, but it is not necessarily required to create an artificial intelligence model; the created artificial intelligence model can also be obtained externally. For example, the artificial intelligence model can be obtained via a communication unit of the computer, or it can be obtained by a human being through transfer or the like.

[0033] Any known algorithm can be used for the machine learning algorithm, such as a decision tree, linear regression, partial least squares regression, lasso regression, ridge regression, polynomial regression, Gaussian process regression, support vector machine, random forest, gradient boosting, K-nearest neighbor method, neural network, Bayesian estimation, or a prediction algorithm based on ensemble learning of these. It is particularly preferable that the machine learning uses a prediction algorithm based on a neural network. The neural network may be a convolutional neural network or deep learning that forms three to several hundred layers while performing dropout. Furthermore, machine learning frameworks such as TensorFlow, Keras, Caffe, PyTorch, Chainer, and Scikit-learn may be used to create the artificial intelligence.

[0034] Here, the machine learning can be supervised learning (including reinforcement learning). In this embodiment, step S103 is performed by a computer performing machine learning (supervised learning) using the relational data input in step S102 as learning data (supervised data in this example).

[0035] Here, the algorithm is preferably determined by the following method. That is, multiple artificial intelligence models are created using various machine learning algorithms, and the accuracy of each artificial intelligence model is confirmed by comparing the predicted output value with the answer using data for which the answer is known in advance. Then, for an algorithm with high accuracy among the multiple algorithms, optimal hyperparameters are determined, for example, by creating a model using cross-validation or using Bayesian optimization. By relearning all data with the determined hyperparameters, an artificial intelligence model can be created that inputs (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any of data among the exposure time of the coating film, environmental information on the exposure of the coating film, installation angle and installation orientation of the coated object on which the coating film is formed, and evaluation results of the accelerated test of the coating film, and outputs results of the durability of the coating film under actual exposure.

[0036] In this embodiment, the prepared relational data is preprocessed. Here, the prepared relational data is subjected to processing such as data normalization / standardization, taking logarithms, multiplying by powers (power processing), and other processing such as data processing to generate new data, and data filtering to remove inappropriate data. For example, data showing a large discrepancy between the actual paint property values ​​and the predicted values ​​can be closely examined, and data suspected of input errors or entry errors can be deleted. As described above, in this embodiment, various machine learning algorithms can be used, and the preprocessing can be applied commonly to multiple or all algorithms (e.g., removing outlier data that is deemed inappropriate by any machine learning algorithm), or can be applied exclusively to the machine learning algorithm actually used (e.g., removing data that is likely to cause errors in a specific algorithm).

[0037] In the pre-processing, it is preferable to add zero-point relation data (zero point addition) that indicates that when the exposure time of the coating film is 0, the durability result of the coating film after actual exposure remains the same as before the actual exposure (amount of change is 0). This can further improve the accuracy of the prediction. Here, it is also possible to change it to zero, or to a number close to zero.

[0038] As a preprocessing step, estimated data can be added between data. For example, an approximate line or approximate curve can be calculated from the data, and the points on the approximate line or approximate curve can be used as estimated data. This increases the amount of data and can further improve the accuracy of prediction. Alternatively, past data can be connected to the current data by a line or curve, and any number of data can be added to the line.

[0039] Such preprocessing and normalization enable more accurate prediction of durability results for real-world exposure in the machine learning described below. However, in the present disclosure, preprocessing is not essential, and this step can be omitted. A predetermined artificial intelligence model can be created using the preprocessed data in the same manner as described above.

[0040] Next, in this embodiment, the computer calculates and predicts the durability of the coating film after actual exposure by inputting (a) paint information data of the coating composition and property information data of the coating film obtained by applying the paint composition before actual exposure, and (b) at least any of data on the exposure time, exposure environment information, installation angle of the coated object, installation orientation, and evaluation results of the accelerated test into the artificial intelligence model (step S104 (durability prediction process for actual exposure)).

[0041] The predicted value of the durability of the coating film under actual exposure may be a single value or may consist of a group of multiple candidates. When the calculated predicted value consists of a group of multiple candidates, it is preferable to further include a step of appropriately selecting a single value from the group of candidates using a predetermined criterion.

[0042] According to the method for predicting the durability of a coating film after real-world exposure of this embodiment, a predicted value of the durability of a coating film after real-world exposure can be obtained by inputting data using an artificial intelligence model, thereby achieving efficiency. Furthermore, by learning the relationship between the (a) paint information data, the coating film's property information data before real-world exposure, and the (b) exposure conditions and accelerated test results, and the durability results of "real-world exposure," the aforementioned problem of discrepancies between accelerated tests and real-world exposure can be avoided. This enables highly accurate prediction of the durability results of a coating film after real-world exposure, as shown in the examples described below. This makes it possible to predict in advance whether or not there is a risk of coating film deterioration, or the level of risk, and take measures to avoid that risk, for example, by adjusting the formulation. As described above, according to the method for predicting the durability of a coating film after real-world exposure of this embodiment, by predicting the durability of a coating film after real-world exposure using a machine learning technique, it is possible to quickly and accurately predict the changes over time in the properties of a coating film at an object at any location.

[0043] Furthermore, according to the method of predicting the durability of a coating film under actual exposure in this embodiment, it is possible to predict the durability under actual exposure even for a coating film produced using a paint composition with a formulation that has not been used in exposure experiments.

[0044] The inputs to the artificial intelligence model include (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least the exposure time of the coating film, and in the actual exposure durability prediction step (step S104), it is preferable to input (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least the exposure time. This is because including at least the exposure time of the coating film in the input makes it possible to predict the time-series deterioration behavior of the coating film and obtain the prediction error.

[0045] Furthermore, it is also preferable that the (b) includes the installation angle and / or installation orientation of the coated object on which the coating film is formed, because this makes it possible to predict the deterioration behavior of the coating film at various installation angles and / or installation orientations and obtain the prediction error.

[0046] The inputs to the artificial intelligence model include (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least data on the exposure time of the coating film and environmental information on the exposure of the coating film, and in the actual exposure durability prediction step (step S104), it is preferable to input (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least data on the exposure time and environmental information on the exposure. This is because the inclusion of the exposure time of the coating film and environmental information on the exposure of the coating film in (b) makes it possible to obtain predictions of the time-series deterioration behavior of the coating film that take environmental information into consideration, as well as the prediction error.

[0047] <System for predicting the durability of a coating film after actual exposure> A system for predicting the durability of a coating film after actual exposure according to one embodiment of the present invention is a system for predicting the durability of a coating film after actual exposure obtained by applying a coating composition using a machine learning technique. The system of this embodiment includes as input (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any one of data on the exposure time of the coating film, exposure environment information for the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and the evaluation results of the accelerated test of the coating film, and outputs the results of the durability of the coating film after actual exposure. The system includes an actual exposure durability prediction unit that calculates and predicts the durability of the coating film after actual exposure by inputting (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before actual exposure, and (b) at least any one of data on the exposure time, exposure environment information, the installation angle and installation orientation of the coated object, and the evaluation results of the accelerated test. The durability prediction unit may be any known processor. The above (a) and (b) regarding the system are the same as those described above for the above (a) and (b) in the method embodiment, and therefore will not be described again.

[0048] The system for predicting the durability of a coating film after real-world exposure of this embodiment can also achieve efficiency by inputting data using an artificial intelligence model to obtain predicted values ​​for the durability of a coating film after real-world exposure. Furthermore, by learning the relationship between the (a) paint information data, the coating film's property information data before real-world exposure, and the (b) exposure conditions and accelerated test results, and the durability results of "real-world exposure," the problem of discrepancies between the accelerated test and real-world exposure can be avoided. This enables highly accurate prediction of the durability results of a coating film after real-world exposure, as shown in the examples described below. This makes it possible to predict in advance whether or not there is a risk of coating film deterioration, or the level of that risk, and take measures to avoid that risk, for example, by adjusting the formulation. As described above, the system for predicting the durability of a coating film after real-world exposure of this embodiment uses machine learning techniques to predict the durability of a coating film after real-world exposure, thereby quickly and accurately predicting the changes over time in the properties of a coating film on an object at any location. Examples of the present invention are described below, but the present invention is not limited to these examples.

[0049] (Method for preparing coating composition) Method for preparing polyester resin (A) 155 parts by mass of neopentyl glycol, 222 parts by mass of 1,6-hexanediol, 50 parts by mass of trimethylolpropane, 441 parts by mass of isophthalic acid, 132 parts by mass of adipic acid, and 26 parts by mass of xylene were mixed in a reaction vessel equipped with a thermometer, a condenser, and a stirrer, and the temperature was gradually raised to 230°C in a nitrogen stream. The water produced was distilled off, and an esterification reaction was carried out until the amount of dehydration reached 126 parts by mass and the reaction product reached a predetermined viscosity. Thereafter, the temperature of the reaction vessel was lowered to 50°C, and then 90 parts by mass of xylene, 379 parts by mass of T-SOL 150, and 87 parts by mass of propylene glycol monomethyl ether acetate were mixed to prepare a coating film-forming resin (A) (solids concentration: 60% by mass, number average molecular weight: 3,000, hydroxyl value: 55 mgKOH / g).

[0050] Method for Preparing Colorants Method for Preparing Yellow Colorant 1 59 parts by mass of polyester resin (A), 16 parts by mass of crosslinker 1, 6 parts by mass of organic solvent 1, and 1 part by mass of organic solvent 2 were uniformly mixed using a disper, and the mixture was dispersed using a sand mill (dispersion medium: glass beads) until the maximum particle size of the pigment coarse particles became 10 μm or less, thereby preparing Yellow Colorant 1.

[0051] A colorant was prepared in the same manner as for yellow colorant Y1, except that the type and / or amount of each component was changed as shown in Table 1.

[0052]

[0053] Preparation method of coating composition A Coating composition A was prepared by adding 73 parts by mass of white colorant 1, 1 part by mass of blue colorant 1, 11 parts by mass of black colorant 1, 4 parts by mass of yellow colorant 1, 8 parts by mass of aggregate 1, and 3 parts by mass of surface conditioner and mixing them uniformly using a disper.

[0054] Details of the materials used in preparing coating composition A are as follows: Acrylic resin (A): Paraloid B44 (manufactured by Rohm & Haas), solid content: 100% by mass Fluororesin (A): KYNAR500 (manufactured by ARKEMA), solid content: 100% by mass Crosslinker 1: Super Beckamin L-155-70 (manufactured by DIC), butylated melamine resin; solid content: 70% by mass Aggregate 1: Nipsil E-200A (manufactured by Tosoh Silica Corporation), silica fine particles; average particle size: 3 μm Aggregate 2: GASIL HP395 (manufactured by INEOS SILICAS), synthetic silica; average particle size: 15 μm Aggregate 3: Sylysia 435 (manufactured by Fuji Silysia Chemical Ltd.), silicon dioxide; average particle size: 4 μm Surface conditioner: Disparlon OX-70 (manufactured by Kusumoto Chemicals Co., Ltd.), acrylic surface conditioner; solid content: 30% by mass White pigment 1: Typeque CR-97 (manufactured by Ishihara Sangyo Kaisha), titanium dioxide White pigment 2: TI-PURE R-706 (manufactured by DuPont), titanium dioxide Black pigment 1: Black 6350 (manufactured by Asahi Chemical Industries Co., Ltd.), chromium iron oxide Black pigment 2: Mitsubishi Carbon Black MA-100 (manufactured by Mitsubishi Chemical Corporation), carbon black Yellow pigment 1: TAROX synthetic iron oxide HY-100 (manufactured by Titanium Kogyo Co., Ltd.), yellow iron oxide Yellow pigment 2: TAROX synthetic iron oxide LL-XLO (manufactured by Titanium Kogyo Co., Ltd.), yellow iron oxide Red pigment 1: TODA COLOR KN-V (manufactured by Toda Pigment Co., Ltd.), iron oxide Red pigment 2: TODA COLOR 140ED (manufactured by Toda Kogyo Co., Ltd.), iron oxide. Organic solvent 1: T-SOL 150 (manufactured by JXTG Nippon Oil & Energy Corporation), aromatic solvent. Organic solvent 2: Cyclohexanone (manufactured by Shoei Chemical Industry Co., Ltd.), ketone solvent. Organic solvent 3: Isophorone (manufactured by ARKEMA), ketone solvent.

[0055] Coating compositions B to F were prepared in the same manner as coating composition A, except that the type and / or amount of each component was changed as shown in Table 2.

[0056]

[0057] (Method for preparing a coating film of coating composition A) After alkaline degreasing of a 0.4 mm thick hot-dip galvanized steel sheet, a phosphate treatment agent Surfcoat EC2310 (manufactured by Nippon Paint Surf Chemicals Co., Ltd.) was applied to the front and back surfaces of the steel sheet for chemical conversion treatment, followed by drying. Flexicoat 612WW Yo Primer (manufactured by Nippon Paint Industrial Coatings Co., Ltd., polyester resin-based primer) was applied to the surface of the obtained steel sheet as a primer paint using a bar coater to a dry film thickness of 5 μm, and baked for 25 seconds under conditions of a maximum material temperature of 215 ° C. to form a primer coating film. The coating composition A obtained above was then applied using a bar coater to a dry film thickness of 15 μm, and baked for 25 seconds under conditions of a maximum material temperature of 220 ° C. to form a surface topcoat coating film, thereby obtaining a coating film of the coating composition.

[0058] (Method of Preparing Coating Films of Coating Compositions B, C, and D) Coating films of coating compositions B, C, and D were obtained in the same manner as in the case of coating composition A.

[0059] (Method for preparing a coating film of coating composition E) After alkaline degreasing of a 0.4 mm thick hot-dip galvanized steel sheet, a phosphate treatment agent Surfcoat EC2310 (manufactured by Nippon Paint Surf Chemicals Co., Ltd.) was applied to the front and back surfaces of the steel sheet for chemical conversion treatment, followed by drying. Next, Fine Tough G Primer (epoxy resin primer: manufactured by Nippon Paint Industrial Coatings Co., Ltd.) was applied as a primer coating using a bar coater to a dry film thickness of 5 μm, and baked for 60 seconds under conditions of a maximum substrate temperature of 210 ° C. to form a primer coating film. Thereafter, the coating composition E obtained above was applied using a bar coater to a dry film thickness of 18 μm, and baked for 60 seconds under conditions of a maximum substrate temperature of 250 ° C. to form a surface topcoat coating film, thereby obtaining a coating film of coating composition E.

[0060] (Method of Preparing a Coating Film of Coating Composition F) A coating film of coating composition F was obtained in the same manner as in the case of coating composition E above.

[0061] (Method of measuring color difference) For each of the coating films obtained above, the color (L before test) was measured using a spectrophotometer color meter SM-T45 (manufactured by Suga Test Instruments Co., Ltd.). ＊ , a ＊ , b ＊ The value and the color difference (ΔE) before and after the test were measured.

[0062] (Method of Measuring Glossiness) The 60° glossiness of each of the coating films obtained above was measured in accordance with JIS K 5600-4-7 (specular glossiness) using a glossmeter VG7000 (manufactured by Nippon Denshoku Industries Co., Ltd.), and the gloss retention before and after the test was calculated using the following formula: Gloss retention (%) = 60° glossiness after the test / 60° glossiness before the test × 100

[0063] (Accelerated Weathering Test: SWOM) The various test plates obtained above were subjected to an accelerated weathering test for up to 5,000 hours using a Sunshine Weather Meter S80 (manufactured by Suga Test Instruments Co., Ltd.), which is a sunshine carbon arc lamp type accelerated weathering tester specified in JIS B 7753. The operating conditions were as follows: irradiance: 255 W / m ２ Black panel temperature: 63°C Water spray time: 18 minutes out of 120 minutes

[0064] (Exposure test) Test panels were installed for 10 years facing south (equatorial plane) at an angle of 20° to the horizontal in accordance with JIS K 5600-7-6 at locations of various latitudes, such as Miyakojima and Iioka. Exposure location 1: Nippon Paint Miyakojima Weathering Center: 3742 Karimata, Hirara, Miyakojima City, Okinawa Prefecture (24°51'N) Exposure location 2: Nippon Paint Iioka Weathering Center: 3-1351 Nagai, Asahi City, Chiba Prefecture (35°69'N)

[0065] 854 paint colors (property information data before actual exposure; film thickness: 6 to 28 μm, L ＊ Value: 7-91, a ＊ Value: -30 to 48, b ＊ Exposure tests were conducted at six locations, including the pre-exposure site, for the following materials: gloss (value: -49 to 41, 60° gloss: 1 to 93), and a total of 20,381 data points were prepared for exposure times of 0 to 240 months.

[0066] Examples 1 to 4, Comparative Example 1 (a) paint information and property information data before exposure, (b) exposure time, exposure environment information, installation angle of the coated object, installation direction, and evaluation results of the accelerated weather resistance test (SWOM), the combinations of variables used as explanatory variables are shown in Table 3. The paint information includes the formulation of the paint composition, and the property information data before exposure includes the film thickness, initial L ＊ , initial a ＊ , initial b ＊ and initial glossiness were used, and latitude, UV radiation amount, and precipitation were used as exposure environment information. Furthermore, the color difference (ΔE) and glossiness after the exposure test were used as the objective variables. Gaussian process regression was used for learning. Unlearned data was predicted by inputting the input information shown in Table 3 into the program obtained by the learning, and the predicted results were compared with the exposure results of the unlearned data. When confirming the prediction accuracy, the leave-group-out method was used to verify that unlearned data was not included in the learning data, and the root mean square error (RMSE) was calculated to compare the accuracy (degree of difference between the predicted value and the exposure result). The results are shown in Table 3. The degree of difference between the predicted value and the actual measured value was evaluated using the root mean square error (RMSE). The root mean square error is an index that measures the degree of discrepancy between the predicted value and the actual measured value; the closer the value is to zero, the smaller the difference. In Table 3, "◯" indicates that the variable was used as an explanatory variable or a dependent variable, and "×" indicates that the variable was not used as an explanatory variable or a dependent variable.

[0067] Comparative Example 2: Based on the results of the accelerated weathering test (SWOM), the RMSE was calculated when a test time of 250 hours was estimated as one year of actual exposure, and the prediction accuracy was compared. The evaluation results are shown in Table 3 below.

[0068]

[0069] Example 1 is an example in which all of the above-mentioned explanatory variables were used. Example 2 is an example in which the results of accelerated weathering evaluation were not included as explanatory variables. Example 3 is an example in which environmental exposure information was not included as explanatory variables. Example 4 is an example in which only exposure time was included as an explanatory variable. In all examples, the RMSE values ​​for color difference and glossiness were small, indicating that the aging change in the properties of the coating film could be predicted with high accuracy. On the other hand, Comparative Example 1 is an example in which none of the explanatory variables (b) were included, and the RMSE values ​​for color difference and glossiness were large, resulting in an insufficient prediction accuracy for the aging change in the properties of the coating film. Furthermore, Comparative Example 2 is an example in which the results of an exposure test were predicted from the results of an accelerated weathering (SWOM) evaluation, and the RMSE values ​​for color difference and glossiness were large, resulting in an insufficient prediction accuracy for the aging change in the properties of the coating film.

[0070] According to the present invention, by using a machine learning technique to predict the durability of a coating film under actual exposure, it is possible to provide a method for quickly and accurately predicting the aging of the properties of a coating film on an object at any point. For example, this method can be applied to predicting the deterioration of plastics.

Claims

1. A method for predicting the durability of a coating film obtained by applying a paint composition under real-world exposure using a machine learning technique, comprising: (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before real-world exposure; and (b) at least any one of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and an evaluation result of an accelerated test of the coating film, as inputs, and the results of the durability of the coating film under real-world exposure are output by a computer in a predetermined artificial intelligence model, comprising a step of predicting durability of the coating film under real-world exposure, in which (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before real-world exposure, and (b) at least any one of data on the exposure time, environmental information on the exposure, the installation angle and installation orientation of the coated object, and an evaluation result of the accelerated test, to calculate and predict the durability of the coating film under real-world exposure.

2. The method for predicting the durability of a coating film upon actual exposure described in claim 1, wherein the input of the artificial intelligence model includes (a) paint information data of the paint composition and property information data before actual exposure of the coating film obtained by applying the paint composition, and (b) at least the exposure time of the coating film; and in the actual exposure durability prediction step, (a) paint information data of the paint composition and property information data before actual exposure of the coating film obtained by applying the paint composition, and (b) at least the exposure time are input.

3. The method for predicting the durability of a coating film upon actual exposure described in claim 1 or 2, wherein the input of the artificial intelligence model includes (a) paint information data of the paint composition and property information data of the coating film obtained by applying the paint composition before actual exposure, and (b) at least data on the exposure time of the coating film and environmental information of the exposure of the coating film, and in the actual exposure durability prediction step, (a) paint information data of the paint composition and property information data of the coating film before actual exposure obtained by applying the paint composition, and (b) at least data on the exposure time and environmental information of the exposure are input.

4. A method for predicting durability of a coating film under actual exposure as described in claim 1 or 2, wherein the paint information data includes at least information regarding the formulation of the paint composition.

5. A method for predicting the durability of actual exposure of a coating film as described in claim 1 or 2, wherein the environmental information of the exposure of the coating film includes at least one of the following information: latitude, amount of solar radiation, amount of ultraviolet light, temperature, precipitation, relative humidity, wind speed, amount of sea salt particles attached, and amount of sulfur oxide attached.

6. A system for predicting the durability of a coating film obtained by applying a paint composition under real-world exposure using a machine learning technique, comprising: (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before real-world exposure; and (b) at least any one of data on the exposure time of the coating film, environmental information on the exposure of the coating film, the installation angle and installation orientation of the coated object on which the coating film is formed, and an evaluation result of an accelerated test of the coating film, as inputs, and the system outputs the results of the durability of the coating film under real-world exposure, characterized in that the system includes an actual-world exposure durability prediction unit that calculates and predicts the durability of the coating film under real-world exposure by inputting, by a computer, (a) paint information data for the paint composition and property information data for the coating film obtained by applying the paint composition before real-world exposure, and (b) at least any one of data on the exposure time, environmental information on the exposure, the installation angle and installation orientation of the coated object, and an evaluation result of the accelerated test.

Images