Machine learning model for coating film and coating film color difference prediction device

Abstract

To provide a machine learning model for a coating film and a coating film color difference prediction device, capable of highly accurately predicting color difference of a coating film after a predetermined period of time has passed.SOLUTION: A machine learning model for a coating film causes a computer to function so as to predict color difference after a predetermined period of time has passed for a coating film having at least a base layer and a clear layer. In the machine learning model for a coating film, information on a base paint used to form the base layer, information on a clear paint used to form the clear layer, a thickness of the clear layer, an amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film, and the color difference between the initial state of the coating film and after the predetermined period of time have passed are set as explanatory variables.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a machine learning model for a coating film and a coating film color difference prediction device. [Background technology]

[0002] As a conventional example of such a technical field, a coating quality prediction device is known, for example, as described in Patent Document 1. This coating quality prediction device calculates the smoothness of a coating surface using a machine learning model in which information on the coating material used in painting a vehicle, the name of the coating material manufacturer, coating data, coating conditions, and the smoothness of the coating surface are used as explanatory variables. [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] In the above-mentioned coating quality prediction device, the characteristics of the coating are further composed of the type of intermediate coating, the color of intermediate coating, the thickness of intermediate coating film, information on the base coating, the type of clear coating, the thickness of the base coating film, the thickness of the clear coating film, the total amount of various pigments, etc. Therefore, there are many explanatory variables used to construct the machine learning model, and there is still room for improvement. In addition, recently, there is a demand for the development of a device that can predict the color difference of a coating film after a predetermined period of time has passed in order to evaluate the weather resistance of the coating film. However, the above-mentioned coating quality prediction device cannot predict the color difference of the coating film.

[0005] The present invention has been made to solve such technical problems, and aims to provide a machine learning model for coating films and a coating film color difference prediction device that can predict the color difference of a coating film after a specified period of time has passed with high accuracy. [Means for solving the problem]

[0006] The machine learning model for coating films according to the present invention is a machine learning model for coating films for causing a computer to function to predict the color difference after a predetermined period of time has elapsed for a coating film having at least a base layer and a clear layer, and is characterized in that the explanatory variables are information on the base paint used to form the base layer, information on the clear paint used to form the clear layer, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue and saturation of the coating film, and the color difference between the initial state of the coating film and after the predetermined period of time has elapsed.

[0007] In the machine learning model for coating films according to the present invention, explanatory variables are information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial brightness, hue, and saturation of the coating film, and the color difference between the initial state and after a predetermined period of time has elapsed. The base paint, the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint are elements that have a relatively large effect on the color difference of the coating film, and by using such elements as explanatory variables in the machine learning model for coating films, it is possible to construct a machine learning model for coating films that can predict the color difference of the coating film after a predetermined period of time has elapsed with high accuracy.

[0008] The coating color difference prediction device of the present invention is a coating color difference prediction device that predicts the color difference after a predetermined period of time for a coating film having at least a base layer and a clear layer, and is characterized in that it comprises a machine learning model for coating films in which information on the base paint used to form the base layer, information on the clear paint used to form the clear layer, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film, and the color difference between the initial state of the coating film and after the predetermined period of time are used as explanatory variables, and a prediction unit that uses the machine learning model for coating films to predict the color difference of the coating film after the predetermined period of time based on the initial lightness, hue, and saturation of the coating film to be predicted, information on the base paint related to the coating film to be predicted, information on the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint.

[0009] The coating color difference prediction device according to the present invention includes a coating machine learning model in which the explanatory variables are information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating, and the color difference between the initial state of the coating and after a predetermined period of time has elapsed. The base paint, the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint are elements that have a relatively large effect on the color difference of the coating, and by using these elements as explanatory variables of the coating machine learning model and using the coating machine learning model, it becomes possible to predict the color difference of the coating after a predetermined period of time has elapsed with high accuracy. Effect of the Invention

[0010] According to the present invention, the color difference of a coating film after a predetermined period of time can be predicted with high accuracy. [Brief description of the drawings]

[0011] [Figure 1] FIG. 2 is a schematic cross-sectional view showing a steel plate and a coating film formed on the steel plate. [Diagram 2] 2 is a schematic cross-sectional view showing a resin plate and a coating film formed on the resin plate. FIG. [Diagram 3] FIG. 2 is a diagram showing the structure of a learning device. [Figure 4] FIG. 1 is a flow diagram for explaining generation of a machine learning model for a coating film according to an embodiment. [Diagram 5] 1 is a diagram showing a structure of a coating color difference prediction device according to an embodiment; [Figure 6] FIG. 11 is a flow diagram showing prediction of color difference of a coating film after a predetermined period of time has elapsed. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Below, we will explain embodiments of the machine learning model for coating films and the coating color difference prediction device of the present invention with reference to the drawings. Before that, we will briefly explain the coatings related to the machine learning model for coating films and the coating color difference prediction device.

[0013] The coating films related to the coating film machine learning model and coating film color difference prediction device are coating films formed on the outer panels and exterior parts of vehicles such as passenger cars, trucks, buses, and railway cars.

[0014] Taking a passenger car as an example, the exterior plate refers to an exterior panel that is a member constituting the vehicle body and is visible from the outside, and includes, for example, a hood panel, a roof panel, a trunk lid, a fender panel, an outer door, a pillar, etc. The material of the exterior plate is, for example, a metal plate, specifically, a hot-rolled steel plate or a cold-rolled steel plate. On the other hand, the exterior part refers to an outfitting part that is attached to the vehicle body and is visible from the outside, and includes, for example, a front bumper cover, a rear bumper cover, a side mirror, a roof molding, etc. A resin material (for example, a resin plate) is mainly used as the material of the exterior part.

[0015] Hereinafter, the structure of the coating film formed on the steel plate and the structure of the coating film formed on the resin plate will be described with reference to Figs. 1 and 2, respectively.

[0016] Fig. 1 is a schematic cross-sectional view showing a steel plate and a coating film formed on the steel plate, where the steel plate is, for example, a hood panel 100. In Fig. 1, the upper side of the paper is the front side (also called the outside) of the hood panel 100, and the lower side is the back side (also called the inside) of the hood panel 100. As shown in Fig. 1, the hood panel 100 comprises a steel plate 200 processed into the shape of a hood panel, and a coating film 300 formed on the steel plate 200. The coating film 300 has an electrodeposition layer 310 formed so as to cover the entire steel plate 200, and an intermediate coating layer 320, a base layer 330, and a clear layer 340 formed in this order on the electrodeposition layer 310 on the front side of the hood panel 100.

[0017] The electrodeposition layer 310 is a layer having a function of enhancing the rust prevention of the steel sheet 200, and has a thickness of, for example, 20 μm. The intermediate coating layer 320 is a layer having a function of improving the chipping resistance and appearance of the coating film 300, and has a thickness of, for example, 35 μm. The base layer 330 is a layer for expressing the color of the coating film 300 (so-called body color), and has a thickness of, for example, 15 μm. The paint used for the base layer 330 contains coloring pigments such as red and white, as well as aluminum powder and mica pigments for expressing metallic colors. The clear layer 340 is a transparent coating layer for protecting the base layer 330, and is a layer that enhances the weather resistance and smoothness of the coating film 300 while enhancing the gloss and luxurious feel. The thickness of the clear layer 340 is, for example, 35 μm.

[0018] 2 is a schematic cross-sectional view showing a resin plate and a coating film formed on the resin plate. The resin plate here is, for example, a front bumper cover 400. In FIG. 2, the upper side of the paper surface is the front side (also called the outside) of the front bumper cover 400, and the lower side is the back side (also called the inside) of the front bumper cover 400. The front bumper cover 400 includes a resin member 500 integrally molded in the shape of a bumper, and a coating film 600 formed on the resin member 500.

[0019] The coating film 600 has a primer layer 610, a base layer 620, and a clear layer 630 formed in this order on the resin member 500. The primer layer 610 is a layer that has a function of improving the adhesion between the base layer 620 and the resin member 500. The base layer 620 is a layer that expresses the color of the coating film 600, and the paint used for the base layer 620 contains color pigments such as red and white, as well as aluminum powder and mica pigments for expressing metallic colors. The clear layer 630 is a transparent coating layer that protects the base layer 620, and is a layer that enhances the weather resistance and smoothness of the coating film 600 while enhancing the gloss and luxury feel.

[0020] Here, the circumstances that led to the present invention will be described.

[0021] Conventionally, the method of calculating the color difference ΔE between the initial state of the coating film and the state after a certain period of time is as follows: first, a test sample having the above-mentioned coating film 300, 600 is prepared in a factory or a testing site, and the initial lightness (L), hue (a) and chroma (b) of the coating film (in other words, a new coating film) of the test sample are measured, and then the test sample is exposed to a natural environment in a region with strong ultraviolet rays for a certain period of time. After the certain period of time has passed, the lightness (L), hue (a) and chroma (b) of the coating film are measured again for the exposed test sample. Then, the differences (i.e., ΔL, Δa, Δb) between the initial lightness, hue and chroma of the coating film and the lightness, hue and chroma of the coating film after the certain period of exposure are calculated, and ΔE is calculated using the following formula (1).

[0022]

number

[0023] The lightness (L), hue (a) and chroma (b) of the coating film are standards established by the International Commission on Illumination to represent colors, and are measured using a device such as a spectrophotometer. In the present embodiment, the initial state of the coating film refers to a new coating film before exposure to light, that is, the body coating of a new car shipped from a factory.

[0024] The above-mentioned method of calculating color difference ΔE requires a long test period, which affects the development of cars. In addition, the calculation result of color difference is greatly affected by the weather at the test site, so it is difficult to perform tests that match the weather conditions, in other words, there is a problem that it is not possible to perform reproducible tests.

[0025] Therefore, as a result of extensive research, the inventors of the present application have constructed a machine learning model for coating films using data including the above-mentioned color difference ΔE obtained from previous findings, and have discovered that the constructed machine learning model for coating films can be used to predict the color difference of a coating film after a predetermined period of time has passed, thereby completing the present invention.

[0026] Specifically, the inventors first focused on the base layer, which greatly affects the color of the paint film, and the clear layer, which greatly affects the weather resistance, among the multiple layers that make up the paint film, and selected information on the base paint used to form the base layer, the amount of pigment contained in the base paint, information on the clear paint used to form the clear layer, and the thickness of the clear layer, and used these as explanatory variables for the machine learning model for the paint film.

[0027] The information on the base paint used to form the base layer is, in other words, information on the composition of the base paint. The base paint is a paint made by mixing pigments and adding resins and additives, and forms the color-developing layer of the coating film 300. For example, a solvent-based colored base paint or a water-based colored base paint is preferably used as the base paint. Examples of the water-based colored base paint include paints containing a pigment, a resin soluble or dispersible in water, a crosslinking agent as necessary, and water as a solvent. Examples of the resin soluble or dispersible in water include resins containing a hydrophilic group such as a carboxyl group and a crosslinkable functional group such as a hydroxyl group in one molecule, and specifically, acrylic resins, polyester resins, polyurethane resins, and the like. Examples of the crosslinking agent include hydrophobic or hydrophilic alkyl ether melamine resins, blocked isocyanate compounds, and the like. On the other hand, examples of the solvent-based colored base paint include paints containing a pigment, the same resin as above, a crosslinking agent as necessary, and a solvent.

[0028] The amount of pigment contained in the base paint refers to the amount of pigment contained in the base paint to create the actual body color, and is based on 100 parts by mass of paint solids such as resin in the paint.

[0029] The pigment is not particularly limited, and may be a normal color pigment or a metallic pigment. Examples of the color pigment include titanium dioxide, zinc oxide, basic lead sulfate, calcium plumbate, zinc phosphate, aluminum phosphate, zinc molybdate, calcium molybdate, Prussian blue, ultramarine blue, cobalt blue, copper phthalocyanine blue, indanthrone blue, yellow lead, synthetic yellow iron oxide, transparent red iron oxide (yellow), bismuth vanadate, titanium yellow, zinc yellow, strontium chromate, cyanamide lead, monoazo yellow, disazo, isoindolinone yellow, metal complex azo yellow, quinophthalone yellow, isoindoline iridium yellow, and the like. Examples of the pigment include yellow, benzimidazolone yellow, red iron oxide, transparent red iron oxide (red), minium, monoazo red, unsubstituted quinacridone red, azo lake (Mn salt), quinacridone magenta, anthanthrone orange, dianthraquinonyl red, perylene maroon, perylene red, diketopyrrolopyrrole chrome vermilion, basic lead chromate, chromium oxide, chlorinated phthalocyanine green, brominated phthalocyanine green, pyrazolone orange, benzimidazolone orange, dioxazine violet, perylene violet, etc. Examples of the metallic pigment include aluminum powder, flake aluminum oxide, pearl mica, flake mica, etc.

[0030] These pigments can be used alone or in combination of two or more. For example, when two or more types are used in combination, the pigment amount refers to the content of each type of pigment.

[0031] The information on the clear coating used to form the clear layer is, in other words, information on the composition of the clear coating. Examples of the clear coating include a thermosetting resin capable of forming a transparent coating film, an organic solvent, and, if necessary, an ultraviolet absorber. Examples of the thermosetting resin include resins such as acrylic resins, polyester resins, alkyd resins, fluororesins, urethane resins, and silicon-containing resins having crosslinkable functional groups such as hydroxyl groups, carboxyl groups, silanol groups, and epoxy groups, and crosslinkers such as melamine resins, urea resins, (blocked) polyisocyanate compounds, epoxy resin compounds or resins, carboxyl group-containing compounds or resins, acid anhydrides, and alkoxysilane group-containing compounds or resins that can react with these crosslinkable functional groups.

[0032] The thickness of the clear coat layer is also called the clear coat thickness, and is a film that protects the coating from ultraviolet rays, dust, etc., and generally the thicker it is, the stronger the weather resistance of the coating. The thickness of the clear coat layer is preferably within the range of, for example, 25 to 45 μm, and is measured using a device such as a film thickness meter.

[0033] Furthermore, the inventors of the present application generated a machine learning model for a coating film using the learning device 1 shown in Figure 3, while adding the color difference ΔE actually measured for the above test sample (i.e., the color difference between the initial state of the coating film and after exposure for a specified period of time) and the actual measured values ​​of the initial lightness (L), hue (a), and saturation (b) of the coating film as explanatory variables.

[0034] The learning device 1 is configured, for example, by a microcomputer that combines a CPU (Central Processing Unit) that executes calculations, a ROM (Read Only Memory) as a secondary storage device that records programs for the calculations, and a RAM (Random Access Memory) as a temporary storage device that saves the calculation progress and temporary control variables, and generates a machine learning model for coating films by executing the stored program.

[0035] Specifically, the learning device 1 includes an input unit 2, a memory unit 3, a generation unit 4, and an output unit 5 (see FIG. 3). The input unit 2 is configured to be able to input information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness (L), hue (a), and chroma (b) of the paint film, and the color difference ΔE between the initial state of the paint film and after a predetermined period of time has elapsed (for example, the actual measured value of the above-mentioned color difference ΔE). The memory unit 3 stores a program for executing machine learning of the learning device 1. The memory unit 3 also has a base paint DB (database) 6 that stores information on the base paint, a clear paint DB 7 that stores information on the clear paint, a clear thickness DB 8 that stores the thickness of the clear layer, a pigment amount DB 9 that stores the amount of pigment contained in the base paint, a color difference DB 10 that stores the actual measured value of the color difference ΔE, and an initial lightness, hue, and chroma DB 16 of the paint film that stores the initial lightness (L), hue (a), and chroma (b) of the paint film.

[0036] Then, when information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film, and the actual measured value of the color difference ΔE described above are input via the input unit 2, the memory unit 3 stores this information and data in each of the DBs described above.

[0037] The generating unit 4 generates teacher data for machine learning and a machine learning model for a coating film. The method of machine learning performed by the generating unit 4 is not particularly limited as long as it is supervised learning. Examples of models or algorithms used in supervised learning include regression analysis, decision trees, support vector machines (SVMs), neural networks, ensemble learning, random forests, and the like, and in this embodiment, a support vector machine is used.

[0038] Fig. 4 is a flow diagram for explaining the generation of a machine learning model for a coating film according to an embodiment. As shown in Fig. 4, first, the generation unit 4 of the learning device 1 acquires information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film, and the color difference ΔE stored in the storage unit 3 (step S11).

[0039] Next, the generation unit 4 generates training data for machine learning by associating the base paint information, the clear paint information, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film with the color difference ΔE (step S12), and generates a machine learning model for the coating film by performing machine learning using deep learning on the generated training data (step S13). Here, Python is used as the programming language, and a support vector machine is used to generate the machine learning model for the coating film.

[0040] This generates a machine learning model for coating films for causing a computer to function. The generated machine learning model for coating films is then output via the output unit 5 (step S14). The generated machine learning model for coating films is implemented in a general-purpose computer or terminal, downloaded as software or an application, or distributed in a state stored in a storage medium, and used in the coating film color difference prediction device 11, etc.

[0041] Hereinafter, the coating film color difference prediction device 11 according to this embodiment will be described with reference to Figures 5 and 6. The coating film color difference prediction device 11 of this embodiment is a device for predicting the color difference of a coating film to be predicted after a predetermined period of time has elapsed, using the above-mentioned machine learning model for coating films. That is, in this embodiment, the color difference ΔE of the coating film to be predicted after a predetermined period of time has elapsed is the objective variable.

[0042] The coating film color difference prediction device 11 is configured, for example, by a microcomputer that combines a CPU (Central Processing Unit) that executes calculations, a ROM (Read Only Memory) as a secondary storage device that records programs for the calculations, and a RAM (Random Access Memory) as a temporary storage device that saves the calculation progress and temporary control variables.

[0043] Specifically, the coating color difference prediction device 11 includes an input unit 12, a storage unit 13, a prediction unit 14, and an output unit 15. The input unit 12 is configured to allow each piece of data to be input. For example, actual measured values ​​of initial lightness, hue, and saturation of the coating film to be predicted, information on the base paint related to the coating film to be predicted, information on the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint are input via the input unit 12.

[0044] The storage unit 13 stores the above-mentioned machine learning model for a coating film. The machine learning model for a coating film is downloaded from the learning device 1 via a communication network, for example. The storage unit 13 also stores various programs related to the operation of the coating film color difference prediction device 11. Furthermore, the storage unit 13 may store various pieces of information and data input via the input unit 12.

[0045] The prediction unit 14 predicts the color difference of the coating film after a predetermined period of time has elapsed based on the initial lightness, hue, and saturation of the coating film to be predicted, information on the base paint related to the coating film to be predicted, information on the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint, using the machine learning model for the coating film stored in the memory unit 13. The output unit 15 is configured to be capable of outputting the prediction result of the color difference of the coating film after a predetermined period of time has elapsed.

[0046] 6 is a flow diagram showing a color difference prediction of a coating film after a predetermined period of time has passed. For example, a user of the coating film color difference prediction device 11 measures the initial lightness, hue, and saturation of the coating film to be predicted, and inputs the measurement results via the input unit 12. The user also determines information on the base paint related to the coating film to be predicted (in other words, used in forming the coating film to be predicted), information on the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint as design values, and inputs them to the coating film color difference prediction device 11 via the input unit 12 (step S21).

[0047] Next, the prediction unit 14 predicts the color difference ΔE of the coating film to be predicted after a predetermined period of time has elapsed based on the information and data input in step S21 using the machine learning model for the coating film stored in the storage unit 13 (step S22). The predicted color difference ΔE is output via the output unit 15 (step S23).

[0048] In the machine learning model for coating films according to this embodiment, the explanatory variables are information on the base paint, information on the clear paint, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial brightness, hue and saturation of the coating film, and the color difference ΔE of the coating film after a predetermined period of time has passed. The base paint, the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint are elements that have a relatively large effect on the color difference of the coating film, and by using such elements as explanatory variables of the machine learning model for coating films, it is possible to construct a machine learning model for coating films that can predict the color difference of the coating film after a predetermined period of time has passed with high accuracy.

[0049] In addition, since the machine learning model for coatings according to this embodiment has fewer explanatory variables compared to learning models used in conventional painting quality prediction devices, the time required to construct the learning model can be reduced.

[0050] Furthermore, in the coating film color difference prediction device 11 according to this embodiment, by using the above-mentioned machine learning model for coating films, it is possible to predict with high accuracy the color difference of the coating film to be predicted after a predetermined period of time has elapsed.

[0051] Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various design modifications can be made without departing from the spirit of the present invention as described in the claims.

[0052] For example, the coating film color difference prediction device 11 may be configured to include a learning device 1. Specifically, for example, the coating film color difference prediction device 11 further includes the above-mentioned generation unit 4, and the storage unit 13 further includes a base paint DB6, a clear paint DB7, a clear thickness DB8, a pigment amount DB9, a color difference DB10, and a coating film initial lightness, hue, and saturation DB16, and information stored in each DB is input via the input unit 12. In addition, the generation unit 4 causes the storage unit 13 to store the generated coating film machine learning model. [Explanation of symbols]

[0053] 1: learning device, 2, 12: input section, 3, 13: memory section, 4: generation section, 5, 15: output section, 6: base paint DB, 7: clear paint DB, 8: clear thickness DB, 9: pigment amount DB, 10: color difference DB, 11: coating color difference prediction device, 14: prediction section, 16: coating initial lightness, hue and saturation DB, 100: hood panel, 200: steel plate, 300, 600: coating, 310: electrocoat layer, 320: intermediate coat layer, 330, 620: base layer, 340, 630: clear layer, 400: front bumper cover, 500: resin member, 610: primer layer

Claims

1. A machine learning model for a coating film for causing a computer to function so as to predict a color difference after a predetermined period of time for a coating film having at least a base layer and a clear layer, A machine learning model for a coating film, characterized in that explanatory variables include information on a base paint used to form the base layer, information on a clear paint used to form the clear layer, the thickness of the clear layer, the amount of pigment contained in the base paint, the initial lightness, hue, and saturation of the coating film, and the color difference between the initial state of the coating film and after the specified period of time has passed.

2. A coating color difference prediction device for predicting a color difference after a predetermined period of time has elapsed for a coating film having at least a base layer and a clear layer, A machine learning model for a coating film in which explanatory variables are information on a base paint used to form the base layer, information on a clear paint used to form the clear layer, a thickness of the clear layer, an amount of pigment contained in the base paint, an initial brightness, hue, and saturation of the coating film, and a color difference between the initial state of the coating film and after the predetermined period of time has elapsed; A prediction unit that predicts the color difference of the coating film after the predetermined period of time has elapsed based on the initial lightness, hue, and saturation of the coating film to be predicted, information on the base paint related to the coating film to be predicted, information on the clear paint, the thickness of the clear layer, and the amount of pigment contained in the base paint, using the machine learning model for the coating film; A coating color difference prediction device comprising:

Images