Systems, methods, and interfaces for predicting coating aging

Abstract

The invention discloses a computer-implemented method comprising: identifying initial color data for an initial coating applied to a target object, where the initial color data identifies (i) a coating having a plurality of components and (ii) colorimetric data of the coating at the time of application; identifying aged color data for the initial coating on the target object; determining changed colorimetric data between the initial color data and the aged color data; training a machine learning algorithm using the changed colorimetric data; receiving an end user selection of a new coating having a plurality of different compositions provided through the graphical user interface; using one or more machine learning algorithms to predict a form of aging of the new coating upon application of a plurality of aging exposure conditions; and displaying data corresponding to the predicted form of aging of the new coating.

Description

Technical Field

[0001] The present invention relates to an apparatus, computer-implemented method and system for predicting aging-based changes in coatings over time. Background Art

[0002] Modern coatings provide several important functions in industry and society. Coatings can protect the coated material from corrosion, such as rust. Coatings can also provide aesthetic functions by providing a particular color and / or texture to an object. For example, most assets, such as cars, are coated with paint and various other coatings in order to protect the metal body of the car from the elements and also provide an aesthetically pleasing visual effect.

[0003] In view of the wide range of uses of different coatings, it is often desirable to identify a target coating composition that can withstand certain aging, such as to maintain a specific color in a specific location over a specific amount of time. For example, it may be desirable to identify a target coating composition that can maintain a blue metallic hue in a hot, humid location, while in another case, it may be desirable to maintain the color in a cold, dry location. However, one or more pigments used in such coating formulations may be more or less stable in various environments. For example, a pigment may perform particularly well in a humid environment, but may denature more quickly in a cold environment without a stabilizing component.

[0004] Some conventional systems attempt to address these issues by monitoring how certain formulations perform over time in a given actual use environment. For example, a conventional system may monitor a target object (e.g., an automobile) that has been returned for repainting and monitors relative fading or other aging damage, and save such reports for future reference. In other conventional systems, coating manufacturers may attempt to predict the performance of such formulations based on the chemical properties of the known underlying formulation ingredients. Systems that only monitor the target object over time cannot provide relevant real-time compositional information for future coatings, as these coatings are constantly updated over time with different hues and effect pigments that may not have been available previously. In addition, relying primarily or entirely on the known chemical properties of individual components may ignore the protective or degrading effects of other ingredients in the final coating, especially when placed in a given physical environment.

[0005] As a result, there are many opportunities to quickly identify how a particular coating formulation is likely to perform in different environments. Summary of the invention

[0006] The present invention provides systems, methods and computer program products for efficiently and accurately estimating the expected aging of a coating based on individual and / or combined contributions from individual components. For example, the present invention includes a computerized system that employs a method for identifying the aging of a real-world coating and its constituent components and determining the individual contributions thereto. The system may also employ artificial intelligence to use one or more previously measured components to predict colorimetric changes resulting from future weather-based degradation.

[0007] For example, a computer-implemented method for automatically determining and displaying expected aging after aging exposure of a coating formulation includes: identifying initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having multiple components, and (ii) colorimetric data of the coating when applied; identifying aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to multiple aging exposure conditions, wherein the aging exposure conditions include at least heat, light, and water exposure; determining altered colorimetric data between the initial color data and the aged color data, wherein the altered colorimetric data is related to (i) the initial color data and the aged color data; ) each of the plurality of coating components and (ii) a plurality of aging exposure conditions; processing the changed colorimetric data through one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of aging exposure conditions are applied; receiving an end-user selection of a new coating having a different plurality of components provided through a graphical user interface; using the one or more machine learning algorithms, applying the changed colorimetric data, determining a predicted aging pattern of the new coating when the plurality of aging exposure conditions are applied; and displaying data corresponding to the predicted aging pattern of the new coating on the graphical user interface.

[0008] Additionally, a computerized system may include: a database, and one or more computer processors; computer executable instructions stored on the database, which when executed, cause the computerized system to: identify initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having a plurality of components, and (ii) colorimetric data of the coating when applied; identify aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, wherein the aging exposure conditions include at least exposure to heat, light, and precipitation; determine altered colorimetric data between the initial color data and the aged color data, wherein the altered colorimetric data is associated with (i) each of the coating components and (ii) a plurality of aging exposure conditions; and process the altered colorimetric data through one or more machine learning algorithms, wherein the one or more machine learning algorithms store in the database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of aging exposure conditions are applied.

[0009] Additional features and advantages will be described in the following description, and in part will be clear from the description, or can be understood through practice. The features and advantages can be realized and obtained by the instruments and combinations particularly pointed out in the appended claims and aspects. These and other features will become more apparent from the following description and the appended claims, or can be learned through the practice of the examples set forth below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to describe the manner in which the above and other advantages and features can be obtained, a more particular description of the above brief description will be rendered by reference to specific examples of the invention, which are illustrated in the accompanying drawings. It should be understood that these drawings are only illustrative and are not therefore to be considered as limiting the scope thereof. The invention will be described and explained in more detail and detail through the use of the accompanying drawings, in which:

[0011] Figure 1A A schematic overview of a system trained using multiple formulation colorimetric measurements prior to aging is shown;

[0012] Figure 1B shows a schematic overview of the system 100 as it continues to receive learning input after the panel has been subjected to aging;

[0013] Figure 1C shows a schematic overview of a client and server system that can be used to identify individual recipe ingredients and the individual impact of each ingredient on recipe aging;

[0014] Figure 1Dshows a chart comparing machine learning predictions to actual data after the system has been trained and validated;

[0015] Figure 2 A schematic diagram showing an end user accessing the system to mix and match ingredients and identify expected aging data in real time;

[0016] Figure 3 A method is shown that includes a plurality of acts in a method for automatically determining and displaying expected degradation of a coating formulation after degradation exposure; and

[0017] Figure 4 An additional or alternative method performed by a system configured to automatically determine and display expected degradation of a coating formulation after weathering exposure is shown. DETAILED DESCRIPTION

[0018] The present invention provides systems, methods and computer program products for efficiently and accurately estimating the expected aging of a coating based on individual and / or combined contributions from various components. For example, the present invention includes a computerized system that employs a method for identifying the true aging of a coating and its constituent components and determining the individual contributions to the aging.

[0019] Thus, as will be more fully understood herein, the present invention can provide numerous benefits to end users, particularly those who wish not only to optimize the appropriate color for a refurbished article, but also those who wish to maintain color consistency over time given local weather patterns and / or durability requirements. Furthermore, end users such as asset maintenance operators and even end customers can be confident that the custom colors designed via the graphical user interface will age as expected on the finished product.

[0020] Figure 1A A schematic overview of a system employing one or more machine learning algorithms trained using multiple formulation colorimetric measurements prior to aging is shown. For example, Figure 1A As shown, the system 100 may include a plurality of coating recipes 140a, 140b, 140c, etc., one or more target objects 145(a, etc.), and a server 105 configured to operate in the context of input from one or more color measurement steps 150(a).

[0021] Figure 1A As further shown, each coating recipe 140a, 140b, and 140c specifies the composition and / or subcomponents of the coating. For example, Figure 1ACoating formulation 140a is shown to include ingredients "A," "B," and "C." As used herein, the terms "subcomponent" and "element" are used interchangeably with the term ingredient. Furthermore, these terms may be understood to refer to specific ingredients of specific names (e.g., a specific formulation of a specific shade of blue pigment), as well as broad categories of such ingredients (pigments or toners of a specific chemical class, e.g., organic / inorganic pigments, binders, solvents, special effect pigments, etc.), and even different concentrations of the same initial ingredient. Along these lines, Figure 1A Coating formulation 140b is shown to include ingredients "B," "C," and "D," and coating formulation 140c includes ingredients "C," "D," and "E," and so on. These ingredients may include a basecoat and / or one or more colorants, toners, pigments, stabilizers, effect pigments, and / or combinations thereof. At least some exemplary ingredients may include pigments such as titanium dioxide, and basecoat components such as polyester polyols, and solvents such as propylene glycol methyl ether and ethylene glycol hexyl ether.

[0022] An operator (or end user) may apply each coating formulation 140a, 140b, 140c to a given one or more target objects 145 (e.g., 145a or other objects not shown). Figure 1A As shown, the target object 145a is represented by a flat plate. However, it will be understood that the target object 145a can alternatively include a bent or curved metal sheet. In addition, other target objects 145a of various materials and / or suitable compositions that are closest to the end-use application are contemplated herein.

[0023] Figure 1A Further shown is that target object 145a is sprayed with at least formulation 140a (i.e., formulation "ABC"). As described above, the operator can apply one or more coating formulations directly to target object 145a. For example, although Figure 1A One panel 145a is shown with one formulation 140a (i.e., formulation "ABC") applied, but the operator may also spray other panels (not shown) with corresponding formulations 140b and 140c, etc., or may spray the same individual panel with multiple different formulations. In any case, after allowing a given panel to cure, Figure 1A Display, the operator can be painted / coated part of the initial (ie, before aging) color and / or reflectivity measurement steps. As discussed more fully below, the color measurement results can be used to train one or more machine learning modules 110 of the server 105.

[0024] For example, Figure 1AAlso shown is a color measurement step 150a of providing data to the server 105. That is, the operator can use one or more computerized color or reflectance recognition devices, such as a colorimeter, a spectrometer, a spectrophotometer, and / or an image capture device (e.g., a camera) to measure the target object 145 (a). Using the relevant measurement device, the operator can then generate the initial color (and / or reflectance) data of the initial coating of each coating formula 140a, 140b, 140c, etc. The operator can then pass the received measurement data of each coating formula 140a, 140b to the server 105. The server 105 then performs various learning and storage functions to associate and maintain the measurement data related to the formula ingredients. In one example, the server 105 uses the one or more machine learning modules 110 shown to coordinate the color measurement results 150a data for each sprayed coating formula 140a, 140b, 140c, and further associates each component of each sprayed coating formula with the color measurement data. Thus, the color data, reflectance data, and / or image data (if applicable) obtained by the received measurement step 150a can be understood to comprise a “training” set for training one or more machine learning modules 110 on what the formulation looks like when applied before aging.

[0025] Figure 1A Also shown, the server 105 includes one or more storage devices 120, which in turn include various components 125, 130, etc. for storing and correlating recipe and color measurement data for use in conjunction with the one or more machine learning modules 110. For example, Figure 1A As shown, the storage device 120 includes various storage components, in this case, in the form of a recipe component 125 and a colorimetric data component 130. The recipe component 125 can be understood as a data structure that stores the recipe name, the corresponding ingredient list, and its various properties (or concentrations) before and after aging. In contrast, the colorimetric data component 130 can be understood as a data structure that stores the measurement results obtained in the color measurement step 150a, which may include any one or more of the image, color, and reflectance data corresponding to the recipe in the recipe component 125.

[0026] However, such formal distinctions between components and data structures are not required, and such distinctions are adopted here mainly for ease of description. For example, each storage component 125, 130 can cooperate to store information about the ingredients included in each formula 140 (ac, etc.), and further associations of images and color / reflectivity values ​​of each formula and ingredient applied. In either case, and no matter how it is configured or divided, one or more machine learning modules 110 coordinate with the storage device 120 to learn the user association of the formula, colorimetric data, and the relevant image of the spray panel 145.

[0027] As an explanation, the term "module" used herein means a computer executable code, which, when executed by one or more processors at a given computer system (e.g., computer system 100 or server 105), enables the given computer system to perform a specific function. In contrast, the term "component" used herein means a passive instruction set or data structure that stores, manages and / or otherwise provides information disposed or otherwise processed by a given module. However, it will be appreciated by those skilled in the art that the difference between different modules or components is at least partially arbitrary, and modules or components can be combined and / or divided in other ways and can still remain within the scope of the present disclosure. Therefore, a given element is described as a "module" or "component" only for clarity and explanation, and unless otherwise explicitly stated, it should not be interpreted as indicating any specific structure requiring a computer executable code and / or computer hardware. In this specification, the terms "component", "agent", "manager", "service", "engine", "virtual machine", etc. may also be used similarly.

[0028] Reference again Figure 1A , the recipe component 125 may include current and previous information determined and / or stored regarding any number of given ingredients (or combinations thereof). Such information includes that recipe 140a includes ingredients A, B, C, while recipe 140b includes ingredients B, C, D, and so on. As previously described, the replacement ingredients B, D, C may include the original ingredients A, B, C, etc. with adjusted concentrations. The recipe component further includes analytics associated with each recipe and corresponding ingredient. Such analytics may include correlations made by the machine learning module 110 (or an operator applying training to the module 110) to one or more aging effects (e.g., Figure 1B ).

[0029] The one or more aging effects may include not only the type of degradation (or "failure type") associated with each formulation and ingredient, but may also include information about various aging sites, various facility types, and / or about devices used to recreate these environmental conditions (e.g., weathering testers), which can be used to help identify defects or other forms of degradation. For the purposes of this specification and claims, "degradation" means the presence of one or more defects or processes that lead to defects, such as lack of pigment durability, fading, gloss change, chemical, electrochemical, biological and physical decomposition or decomposition processes, scaling (including biofouling produced by the growth of biological matter), pitting, rotting, rusting, delamination, pits, cracking or other surface or coating defects and processes that promote these defects. Therefore, the terms "defect", "degradation" and "failure type" can be understood as somewhat interchangeable terms with respect to aging effects in at least some applications herein, particularly when referring to the time point at which the defect exists.

[0030] As described above, the recipe component 125 may include storage and / or reference to a wide range of recipes, ingredients, and other related data. For example, in addition to recipes and measurement data, the storage device 120 (via the recipe component 125 and / or the colorimetric data component 130) may store hundreds, thousands, or even millions or more images of the sprayed panel 145 corresponding to each recipe representing the state before, during, and / or after weathering, as well as the ingredients corresponding to each recipe in the system. Similarly, as discussed more fully below, any of the stored data may be collected before, during, and after weathering is applied (see Figure 1B ). The storage device 120 components 125, 130 further include individual human or machine determinations regarding qualitative issues in the applied formulation. Such determinations may include a broad range of acceptability designations (e.g., acceptable or "OK", and unacceptable - not OK or "NOK"), or similar such determinations made by one of the one or more machine learning processors 110 after sufficient training.

[0031] For example, the system 100 may employ one or more learning algorithm modules 110 to learn from various training sets and improve analytical expertise over time. Such machine learning module 110 algorithms may include, but are not limited to, algorithms understood as supervised learning algorithms, unsupervised learning algorithms, semi-supervised learning algorithms, reinforcement learning algorithms, reinforcement learning algorithms, self-learning algorithms, feature learning algorithms, anomaly detection algorithms, robot learning algorithms, and / or composite versions thereof. In one example, the supervised learning algorithm used herein may include (i) random forest techniques, (ii) XGBoost tree techniques, or (iii) support vector machine techniques. Any one or more of the above algorithms may be employed in a deep learning neural network to store, learn, and process panel images into similar surface features in future images. In order to improve the class imbalance between the "NOK" and "OK" categories, the analysis method may employ a combination of an oversampling technique (synthetic minority oversampling technique, or SMOTE) followed by an undersampling technique (edited nearest neighbor, ENN) - cumulatively referred to as SMOTEENN. Typically, SMOTE is an oversampling technique that generates synthetic samples for a minority class in an unbalanced classification data set. Using ENN after SMOTE provides the additional advantage of removing noise in most of the data points.

[0032] Table I below provides the test results showing the improvement in prediction and recognition before and after using the SMOTEENN technique with a random forest classifier.

[0033] Table I

[0034]

[0035] An exemplary use of artificial intelligence and machine learning related to object and surface recognition in accordance with one or more aspects of the present invention is outlined below. As part of training and / or final processing of data provided during the color measurement step 150a, one or more machine learning modules 110 may perform image analysis on user captured images using a "Look Only Once" (YOLO) type object detection algorithm. To achieve this use, an operator or user of the system 100 first trains the machine learning module 110 using various types of image segmentation on training images, such as by using object segmentation, instance segmentation, and semantic segmentation to process panels 145 processed before and after aging. In general, "object segmentation" can be used to characterize images from a broad perspective, and the system 100 may additionally or alternatively employ "intersection over union" (IOU) related to object detection.

[0036] In contrast, “instance segmentation” can involve more specifically defining boundaries or masks around various aging defects. Figure 1B As understood, the machine learning module 110 in conjunction with the aging data component 135 can use instance segmentation to identify specific areas of different aging effects on the painted panel. In turn, "semantic segmentation" involves characterizing each pixel within the boundaries identified in the instance segmentation step. For example, using semantic segmentation, an aged panel can be mapped by pixels and associated color values.

[0037] When ongoing training begins, the system 100 (i.e., the server 105 and the machine learning module 110) can continue to learn and characterize the identified surfaces. For example, the system 100 can employ training on both previously characterized images and new images fed into the system. As previously described, one or more human experts can label the segmented images from a previously aged training set (not shown) to identify surface defects and the type of surface defects, and further provide any other characterizations of the spray recipes associated therewith. The expert user can then feed the characterization of each image back into the neural network of the system 100, enabling the system 100 to optimize itself for future applications.

[0038] Along these lines, the human expert can label the processed image to include various geometric features, such as labels for cone, rectangle, circle, ellipse, etc., and further provide color labels to the extent such labels have not yet been recognized by the system. Other geometric features can include more specific inputs and corresponding rules, such as certain types of fading or discoloration are associated with thermal factors compared to exposure to water, which will be discussed more fully below. Thus, the human expert can train one or more machine learning modules 110 to understand not only how to recognize generalized shapes, but also the context of their object recognition, which can help ensure that future analysis is contextually accurate.

[0039] Figure 1B 1 shows a schematic overview of the system 100 as it continues to receive learning input after the panels 145 (a, b, c, etc.) have been subjected to aging. For example, Figure 1B As shown, the system 100 further applies aging to each of the panels (145a, 145b, 145c, etc.) in step 160. The aging step 160 can be natural, for example, by allowing the panels to sit in an outdoor environment for several months or years, or the operator can apply aging artificially through a weathering tester (not shown). Typically, the weathering tester includes a laboratory-scale device into which the operator inserts the target object (e.g., 145 (a, b, c)) and applies simulated aging phenomena, such as ultraviolet light, precipitation, cloud cover, heat index, dew point and other changes in heat, and "water exposure", etc. As used herein, "water exposure" means precipitation and humidity (naturally occurring or through simulation) and applied water, such as water applied by a water sprayer or injector (e.g., via a weathering tester). The weathering tester can enable the operator to generate aging effects on the target object 145 over a period of hours, days, or weeks, rather than over the more common time range of months or years used to observe natural aging.

[0040] After or even during the aging step 160, the operator may again collect images and / or color measurements / reflectivity data for each panel / target object 145. For example, Figure 1BA second color measurement step 150b is shown, in which the operator / user again measures the color and / or reflectance data, such as using a colorimeter, spectrometer, or spectrophotometer. The operator transmits the second color measurement results 150b to the server 105, which includes a machine learning processor 110 and a storage device 120, in the same manner as previously used for the initial measurement data in step 150a. As discussed in more detail below, the operator also provides various indications of color acceptability, such as indicating various panels as "OK" or "Not OK" (i.e., "NOK"). Such determinations may include, but are not limited to, determining "ΔE" (or "dE") tolerance levels at different measurement angles, where a ΔE of 0 is ideal at any given angle.

[0041] Figure 1B As further shown, storage device 120 may include one or more additional data structures in the form of aging data components 135 that may be used to store or otherwise coordinate the data measured from step 150b. Again, the specific distinctions between components 125, 130, and 135 are somewhat arbitrary and are made for ease of description. In any case, Figure 1B It is shown that the aging data 135 includes at least colorimetric data for each component of the coating formulations 140a, 140b, 140c and any other aging data for each component, such as OK / NOK, specific degradation points. The aging data for the formulation components in component 135 is represented as "[A']" for component "[A]". Similarly, the measured data after aging of component "[B]" is shown as "[B']", the aging data for component "[C]" is shown as "[C']", and so on. As described above, in at least one case, the difference between "[A]" and "[A']" can be represented by a ΔE colorimetric difference. The differences shown between [A'], [B'], and [C'] can correspond to assessments based on images showing various defects, such as pigment durability, gloss changes, delamination, rust or other forms of degradation, and any measurement results (color, reflectivity, etc.) that can be related to these effects.

[0042] For example, as described above, an operator may provide a qualitative assessment for each aged panel 145a', 145b', and 145c'. The qualitative assessment may include a broad characterization based on a specific tolerance for ΔE, such as "good" or "not good" (i.e., "NOK"), or observations regarding the type of fading, gloss change, color intensity, delamination, or other defects observed. The qualitative assessment may also include a specific identification of degradation on the panel, such as pointing out fading, blistering, rusting, or other forms of degradation observed through manual identification. The operator's qualitative assessment of each panel 145a', 145b', 145c', etc. may be passed through one or more machine learning modules 110 and stored in the aging data component 135 with other data. If the determination of OK and NOK results in an unbalanced data set, the server 105 may further employ SMOTEENN to oversample the minority data set. In summary, these techniques enable one or more machine learning modules 110 to also learn to correlate specific changes across the data set in image data, colorimetric (e.g., ΔE), and / or reflectance data for any given aged panel 145 (a'-c', etc.) surface with qualitative and numerical assessments to help identify the contributions of specific components.

[0043] Determinations made by one or more machine learning modules 110 may include determinations of rates of change of gloss, fading, or other appearance metrics, including the formation of a coating on an object in one form or another due to aging. For example, one or more machine learning modules 110 may determine that ingredient or element C of formulation 140a has a tendency to fade, cause gloss changes, or delaminate at a higher or lower rate than another ingredient or element in the presence of humidity, or in the presence of ultraviolet light, or some combination thereof with other aging factors. Similarly, one or more machine learning modules 110 may determine that ingredient B of formulation 140b appears to have a stabilizing effect on element C of the same formulation, such that the combination of ingredients B and C tends to cause a formulation including both ingredients B and C to resist degradation in the same type of environment, or may accelerate degradation in another type of environment. The same analysis can be applied by simply changing the concentration of ingredient B or the concentrations of B and / or C.

[0044] Figure 1C A schematic overview of a client and server system is shown that can be used to identify individual recipe ingredients and each ingredient's relevance or contribution to aging data. For example, Figure 1C As shown, server 105 communicates with one or more client systems 200 via network 205. Network 205 may include any local, wide area, regional, or global area network connection, ranging from a direct wired connection to a connection via the Internet. Network 205 may also or additionally include local near field networks, including Bluetooth and WIFI networks. In any case, Figure 1CAs shown, client system 200 interacts with server 105 data to help train one or more machine learning modules, evaluate their progress, and obtain predictions corresponding to certain recipe ingredients and their responses to aging step 160.

[0045] For example, Figure 1C As shown, the client system 200 communicates with the server 105, such as via a network 205. The client system 200 may include several different types of end-user terminals, including but not limited to traditional personal computing devices, or mobile devices such as laptops, tablets, mobile phones, or other forms of network-connected computing devices. However, a network-connected client system is not required, and the operator can directly interface with the server 105. In other words, the server 105 and the database 120 can be interpreted as independent or other database applications installed on the client system 200. In any case, the operator can interact with the server 105 via a variety of different means to input data, train one or more machine learning modules 110 using pre- and post-aging measurements (e.g., analysis 215). Such operator interaction (e.g., via one or more analysis messages 215) can also include monitoring the progress of one or more machine learning modules 110 in terms of the ability of one or more machine learning modules 110 to provide consistent or accurate predictions and / or adjustments to the predictions (e.g., after applying the SMOTEENN technology).

[0046] Along these lines, Figure 1D Various example result graphs (225a, 225b) are provided that show exemplary progress of machine learning (in this case using a random forest classifier) ​​predictions based on various training sets provided to the server 105 by an operator. Figure 1D The graph of 225a plots the percentage of correctly identified "OK" (225a) and "NOK" (225b) versus the measured angle (e.g., steps 150a, 150b) using the "ΔE" measurements for all angles. As shown, the one or more machine learning modules 110 can use realistic, supervised class distributions to generate analysis and predictions that perform better than a random baseline (i.e., the lower line of graph 225a).

[0047] For example, charts 225a and 225b show test data results used to validate how well the machine learning algorithms outlined herein can perform in real-world scenarios using aging data. The data presented in the charts describe recall and precision metrics that are traditionally used to evaluate the accuracy of computer predictions. For example, chart 225a shows the percentage of "not OK" (NOK) at each measurement angle, where "NOK predictions" refers to the proportion of correct NOK predictions (computer-generated data using one or more machine learning algorithms) among all NOK computer-generated predictions (i.e., precision), and "NOK identified" refers to the actual number of true NOK samples that the model was able to correctly label as NOK (i.e., recall). Along these lines, chart 225a shows that the computer predictions exceeded the accuracy of random control. In addition, chart 225a shows that the line showing the percentage of correct NOK identifications has roughly similar shapes and magnitudes to the line showing the percentage of correct NOK identifications, indicating the overall high predictive power / accuracy of the model.

[0048] In contrast, graph 225b compares the percentage of correctly identified OKs (OK recall) and the percentage of correct OK predictions, i.e., predictions generated using a computer based on one or more machine learning algorithms (OK precision). As with the NOK measurements in graph 225a, the OK measurements and predictions in graph 225b show that the computer predictions perform well for most angles and fail only at low angles where variations in the measurements create noise on the model. In other words, the graph shows that one or more machine learning algorithms are able to very closely approximate real-world observations using ΔE data to predict aging degradation. That is, the fact that both the computer prediction precision and recall metrics for both OK and NOK are high (typically exceeding the 75% accuracy threshold) confirms that the computer-aided models and algorithms outlined herein are able to correctly distinguish between OK and NOK characteristics in most situations, thereby confirming the effectiveness of the component-specific approach used to generate predictions.

[0049] Figure 2 A schematic diagram is shown of an end user using the client system 200 to mix and match ingredients for a new recipe and identify expected aging data in real time. For example, Figure 2An overview of a client system 200 is shown, which includes a graphical user interface 250 in various input and output interface sequences (220a / 220b and 230a / 230b, respectively). Through the first input interface 220, a user inputs a set of ingredients based on, for example, recipe 140a, although replacing ingredient D with ingredient C or changing its concentration. For example, ingredient D can be an alternative concentration of ingredient C. The computer system 200 sends the new recipe to the server 105 via one or more messages 260a. For example, the user may be interested in a recipe using a specific color and a new effect pigment (element D, such as "mica"), while the system 100 only has prior measurements of the color using a different effect pigment (e.g., element C, such as "aluminum"), and the client 200 submits the input to the server 105 via the network 205. One or more machine learning modules 110 at the server 105 can process the data from the one or more messages 260a. In one example, one or more machine learning algorithms 110 extrapolate the stored aging data associated with each ingredient to determine their respective contributions to degradation in one or more other / new formulations (e.g., as measured from formulations 140b and 140c).

[0050] That is, the one or more machine learning modules 110 may then prepare a predicted aging effect that combines data from measurements associated with recipe 140a with contributions from element D from pre / post measurements of recipes 140b and 140c (and possibly other recipes). Thus, in response to an input request in interface 220a, user interface 250 may provide output 230a (sent via one or more messages 260b) indicating that the new recipe ABD is predicted to succeed, improve, or fail by a certain amount or a certain rate of change relative to the original baseline recipe 140a in response to a particular type of aging. For example, output interface 230a may indicate that the requested recipe ABD will fade, change adhesion or gloss, or deteriorate by 5% in 9 years compared to 11 years for the original recipe, or otherwise deteriorate by 20% in different types of accelerated tests (e.g., different weathering machine conditions). Alternatively, output interface 230a may indicate that by changing one of the ingredients to a new formulation, such as changing formulation ABD to ABE, the input formulation ABD will last longer (e.g., 12 years) than the expected duration (11 years) without such defects with respect to one or more of those same factors.

[0051] Along these lines, in the next illustrated example, the user similarly enters another new recipe BCE in the input interface 220b, where the new recipe BCE is a variation of the original recipe 140b (i.e., recipe BCD is changed with ingredient E). The computer system 200 sends the user's input to the server 105 via one or more input messages 260b, indicating that the input is an adjustment made to the recipe 140b by inserting ingredient E in place of the original ingredient D. Figure 2 It is then shown that the server 105 generates one or more predictions for the changed recipe and returns one or more aging predictions 265b. The computer system 200 then displays the prediction information in the output screen 230b.

[0052] Thus, Figures 1 to 2 provide a plurality of components, modules and schematic diagrams as part of a system for automatically determining and predicting the expected aging effects of individual components of a coating / composition. The present invention may also be described in terms of one or more methods for achieving a particular result. Along these lines, Figure 3 and 4 Various methods are shown for automatically determining and displaying the expected degradation of a coating formulation after aging exposure. Figure 1A-2 Discussion of system components and modules Figure 3 and 4 actions and steps.

[0053] For example, Figure 3 , method 300 for automatically determining and displaying expected aging after aging exposure of a coating formulation may include an act 310 of identifying initial color data of an initial coating applied to a target object. Act 310 includes identifying initial color data of an initial coating applied to the target object, wherein the initial color data identifies (i) a coating having multiple components, and (ii) colorimetric data of the coating when applied. For example, Figure 1A As shown, the various recipes, their corresponding ingredients, and their initial color measurements when applied to target object 145 ( a ) (via step 150 a ) may be communicated to server 105 via storage device 120 and stored therein.

[0054] Figure 3 Also shown, method 300 may include an act 320 of identifying aged color data for the initial coating. Act 320 includes identifying aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, the aging exposure conditions including at least heat, light, and water exposure. For example, Figure 1BThe operator is shown taking second color measurements 150b of the various target objects after applying the weathering step 160. As previously described, weathering exposure conditions may include natural and synthetically applied (e.g., via a weatherometer) weathering in the form of heat, light, and water exposure in the form of precipitation and humidity, as well as water applied by water spraying or jetting, and the like.

[0055] in addition, Figure 3 As shown, method 300 may include act 330 of determining altered colorimetric data between the initial coating and the aged color data. Act 330 includes determining altered colorimetric data between the initial color data and the aged color data, wherein the altered colorimetric data is associated with (i) each of the coating components and (ii) a plurality of aged exposure conditions. For example, Figure 1B As shown, server 105 can identify ingredient-specific associations (eg, stored in component 135) with various aging (applied or natural) at least in part by comparing color measurements taken in steps 150a (initial measurement) and 150b (measurement after aging).

[0056] also, Figure 3 As shown, method 300 may include an act 340 of processing the changed colorimetric data by artificial intelligence. Act 340 may include processing the changed colorimetric data by one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of aging exposure conditions are applied. For example, Figure 1A-1B As shown, one or more machine learning algorithms process the pre- and post-aging data associated with various formulations and store the associations in the storage device 120. As part of the processing, the one or more machine learning algorithms may identify various associations between individual ingredients and identified changes in color measurements (e.g., the delta E or "dE" between steps 150a and 150b).

[0057] Going further, Figure 3 As shown, method 300 may include action 350 of receiving an end-user selection of a new coating having a new composition. Action 350 may include receiving an end-user selection of a new coating having a different plurality of compositions provided via a graphical user interface. For example, Figure 2 1. The user is shown providing input for adjustments to the recipe 140a through the input interface 220a, such as by swapping ingredient C with ingredient D in the original recipe 140a.

[0058] Figure 3It is also shown that method 300 may include an act 360 of using a machine learning algorithm to predict the aging pattern of the new composition. Act 360 may include using one or more machine learning algorithms to apply the changed colorimetric data to determine the predicted aging pattern of the new coating when a plurality of aging exposure conditions are applied. For example, Figure 2 26. In response to a user input requesting predicted aging of the adjusted formulation 140a (i.e., message 260a), the server 105 sends a response via message 265a that includes an aging prediction for the requested adjusted formulation. The message 265a may include, for example, ΔE predictions obtained by one or more machine learning modules 110 for the coating formulation at various angles, and given certain assumptions about expected aging.

[0059] in addition, Figure 3 Display, method 300 may include an action 370 of displaying the data of the predicted aging form. Action 370 may include displaying data corresponding to the predicted aging form of the new coating on a graphical user interface. For example, Figure 2 Display, input and output interfaces (respectively 220a / b, 230a / b) may be provided and may be displayed on the display device 250 of the client computer system 200. As previously described, the predicted aging pattern of the new coating may include predicted changes to the new coating based on multiple components of the new coating and multiple aging exposure conditions. The predicted changes may include changes in color, adhesion, gloss, and / or other appearance metrics, such as dirt.

[0060] The predicted aging pattern of the new coating may also include a prediction of whether the new coating will fail (e.g., a change in gloss or adhesion exceeding a predetermined value) after being exposed to the aging exposure conditions for a certain amount of time. The predicted aging pattern of the new coating may also include a prediction of what type of failure the new coating will experience; for example, a change in gloss and / or adhesion. The prediction may also include expected aging changes based on specific climate conditions in a specific geographic location. For example, a response from server 205 may include that the new formulation will degrade to outside of acceptable tolerances within 10 years in a cold, dry climate, and will degrade to outside of acceptable tolerances within 8 years in a humid, hot climate.

[0061] Figure 4 As shown, an additional or alternative method 400 for automatically determining and displaying expected aging after aging exposure of a coating formulation may include an act 410 of identifying initial color data of an initial coating applied to a target object. Act 410 includes identifying initial color data of an initial coating applied to the target object, wherein the initial color data identifies (i) a coating having multiple components, and (ii) colorimetric data of the coating when applied. For example, Figure 1AAn operator is shown applying one or more formulations including a plurality of ingredients to a panel and then performing a color measurement step 150a to identify at least an initial colorimetric measurement (ie, before aging) of the applied formulation.

[0062] Figure 4 Also shown, method 400 may include an act 420 of identifying aged color data for the initial coating. Act 420 includes identifying aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, the aging exposure conditions including at least exposure to heat, light, and precipitation. For example, Figure 1B The application of an aging step 160 is shown (i.e., naturally or by a weathering test machine). The aging step may include other forms of applying light, heat, water, humidity, and various combinations as outlined herein. The operator may then perform a second or subsequent color measurement step 150b to at least identify any differences in the colorimetric data due to the applied aging.

[0063] in addition, Figure 4 As shown, method 400 may include an act 430 of determining changed colorimetric data between the initial coating and the aged color data. Act 430 includes determining changed colorimetric data between the initial color data and the aged color data, wherein the changed colorimetric data is associated with (i) each of the coating components and (ii) a plurality of aged exposure conditions. For example, Figure 1B It is shown that a second color measurement 150b is taken after applying the aging step 160. The operator may communicate the second color measurement data for a given panel to the server 105 where the color measurement results may be tied to various recipes.

[0064] also, Figure 4As shown, method 400 may include an action 440 of processing the changed colorimetric data by artificial intelligence. Action 440 includes processing the changed colorimetric data by one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the multiple components with the rate of change of the colorimetric data determined when applying multiple aging exposure conditions. For example, server 105 includes one or more machine learning modules 110, which apply any number of one or more supervised learning algorithms, such as random forest techniques, XGBoost tree techniques, and support vector machine techniques. As previously described, one or more machine learning modules 110 can determine the various contributions of the individual components of the formula by analyzing one or more formulas containing the component and how the various formulas respond to the aging step 160. One or more machine learning modules 110 can process the changes in the color measurement results from step 150b relative to step 150a, and determine the changes in the colorimetric data, reflectance, etc., applied to each individual component of each formula.

[0065] Although not shown, the computer-implemented methods 300 and 400 may also include suggesting one or more ingredient substitutions. The one or more ingredient substitutions may be optimized by the one or more machine learning modules 110 to minimize changes in appearance in response to multiple aging exposure conditions. That is, the one or more ingredient substitutions may be optimized by the one or more machine learning algorithms employed by the modules 110 to resist predicted changes due to multiple aging exposure conditions.

[0066] therefore, Figure 1A-4 Several systems, components and modules are provided that can be used to significantly improve the speed and accuracy of the contribution of individual ingredients to the aging response. Part of this improvement can be provided by using different types of aging, including using natural and synthetic aging conditions. Part of the reason for this improvement may also be due to the use of one or more machine learning algorithms to process and associate changes in instrument measurement data, and associate these changes with individual ingredients. Therefore, the principles outlined in this disclosure can provide end users with significantly more efficient and accurate tools to obtain instant feedback on the aging degradation results of new formulas with new ingredients.

[0067] The present invention may include or utilize a special or general-purpose computer system, which includes computer hardware, such as, for example, one or more processors and system memory, as discussed in more detail below. The scope of the present invention further includes physical and other computer-readable media for carrying or storing computer-executable instructions and / or data structures. Such computer-readable media can be any available media that can be accessed by a general or special-purpose computer system. Computer-readable media that store computer-executable instructions and / or data structures are computer storage media. Computer-readable media that carry computer-executable instructions and / or data structures are transmission media. Therefore, by way of example and not limitation, the present invention may include at least two distinct types of computer-readable media: computer storage media and transmission media.

[0068] Computer storage media are physical storage media that store computer-executable instructions and / or data structures. Physical storage media include computer hardware, such as RAM, ROM, EEPROM, solid-state drive ("SSD"), flash memory, phase-change memory ("PCM"), optical disk storage, magnetic disk storage or other magnetic storage, or any other hardware storage that can be used to store program code in the form of computer-executable instructions or data structures that can be accessed and executed by a general-purpose or special-purpose computer system to implement the disclosed functions of the present invention.

[0069] Transmission media may include networks and / or data links that may be used to carry computer executable instructions or program codes in the form of data structures and may be accessed by general or special computer systems. "Network" is defined as one or more data links that can be used to transmit electronic data between computer systems and / or modules and / or other electronic devices. When information is transmitted or provided to a computer system via a network or another communication connection (hardwired, wireless, or a combination of hardwired or wireless), the computer system may view the connection as a transmission medium. Combinations of the above should also be included within the scope of computer-readable media.

[0070] In addition, upon reaching various computer system components, program code in the form of computer-executable instructions or data structures can be automatically transferred from transmission media to computer storage media (or vice versa). For example, computer-executable instructions or data structures received via a network or data link can be cached in RAM within a network interface module (e.g., "NIC"), and then ultimately transferred to computer system RAM and / or low-volatility computer storage media at the computer system. Thus, it should be understood that computer storage media can be included in computer system components that also (or even primarily) utilize transmission media.

[0071] Computer executable instructions include, for example, instructions and data, which, when executed at one or more processors, cause a general purpose computer system, a special purpose computer system, or a special purpose processing device to perform a certain function or group of functions. Computer executable instructions may be, for example, binary, intermediate format instructions such as assembly language, or even source code.

[0072] Those skilled in the art will appreciate that the present invention can be practiced in a network computing environment with various types of computer system configurations (including personal computers, desktop computers, laptop computers, message processors, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network PCs, microcomputers, mainframe computers, mobile phones, PDAs, tablet computers, pagers, routers, switches, etc.). The present invention can also be practiced in a distributed system environment where both local and remote computer systems perform tasks, and the local and remote computer systems are linked by a network (by a hardwired data link, a wireless data link, or by a combination of a hardwired and wireless data link). Therefore, in a distributed system environment, a computer system can include a plurality of constructed computer systems. In a distributed system environment, a program module can be located in both local and remote memory storage devices.

[0073] Those skilled in the art will also appreciate that the present invention may be practiced in a cloud computing environment. A cloud computing environment may be distributed, but this is not required. When distributed, a cloud computing environment may be distributed internationally within an organization and / or have components owned across multiple organizations. In this specification and the appended claims, "cloud computing" is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage devices, applications, and services). The definition of "cloud computing" is not limited to any of the other numerous advantages that may be derived from such a model when properly deployed.

[0074] The cloud computing model can consist of various characteristics, such as on-demand self-service, broad network access, resource pooling, rapid elasticity, measurable services, etc. The cloud computing model can also appear in the form of various service models, such as, for example, software as a service ("SaaS"), platform as a service ("PaaS"), and infrastructure as a service ("IaaS") The cloud computing model can also be deployed using different deployment models, such as private cloud, community cloud, public cloud, hybrid cloud, etc.

[0075] A cloud computing environment or cloud computing platform may include a system that includes one or more hosts, each of which is capable of running one or more virtual machines. During operation, the virtual machine simulates an operating computing system, thereby supporting an operating system and possibly one or more other applications. Each host may include a hypervisor that simulates virtual resources for the virtual machine using physical resources that are abstract from the perspective of the virtual machine. The hypervisor also provides appropriate isolation between virtual machines. Therefore, from the perspective of any given virtual machine, the hypervisor provides an illusion that the virtual machine is docked with physical resources, even if the virtual machine is only docked with the appearance of physical resources (e.g., virtual resources). Examples of physical resources include processing power, memory, disk space, network bandwidth, media drives, etc.

[0076] In view of the foregoing, the present invention may be embodied in a number of different configurations, as summarized above and as exemplified below.

[0077] For example, one configuration may include a computer-implemented method for automatically determining and displaying expected aging after aging exposure of a coating formulation, comprising: identifying initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having a plurality of components, and (ii) colorimetric data of the coating when applied; identifying aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, the aging exposure conditions including at least heat, light, and water exposure; determining colorimetric data of a change between the initial color data and the aged color data, wherein the changed colorimetric data associating (i) each of the coating components and (ii) a plurality of aging exposure conditions; processing the changed colorimetric data through one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of aging exposure conditions are applied; receiving an end-user selection of a new coating having a different plurality of components provided through a graphical user interface; applying the changed colorimetric data using the one or more machine learning algorithms to determine a predicted aging pattern of the new coating when the plurality of aging exposure conditions are applied; and displaying data corresponding to the predicted aging pattern of the new coating on a graphical user interface.

[0078] In an additional or alternative configuration, the aging condition includes subjecting the target object to natural aging in a natural environment for a period of multiple months. A computer-implemented method according to any of the preceding claims, wherein the aging condition includes a simulation of the aging condition by a weathering test machine. In an additional or alternative configuration, the method may also include identifying a historical data set of the initial coating; wherein the historical data set includes colorimetric data of changes associated with multiple aging exposure conditions and one or more different applications of the initial coating on different target objects subjected to different multiple aging exposure conditions. In an additional or alternative configuration, the method may also include applying the historical data set using one or more machine learning algorithms to determine, for each of the components, the expected colorimetric data changes associated with multiple aging exposure conditions and each of the different multiple aging exposure conditions. In an additional or alternative configuration, the method may also include determining a failure type of the initial coating after applying multiple aging exposure conditions; wherein the failure type includes a color change.

[0079] In additional or alternative configurations, the method may further include determining a failure type of the initial coating after applying the plurality of aging exposure conditions; wherein the failure type includes a change in gloss or a change in adhesion. In additional or alternative configurations, the method may further include receiving a user selection reflecting one or more changes in the composition of the new coating; determining an aging change of the new coating when any of the plurality of aging exposure conditions are applied. In additional or alternative configurations, the method may further include displaying on a graphical interface a predicted change of the new coating after a plurality of time intervals of simulating the application of the plurality of aging exposure conditions.

[0080] In an additional or alternative configuration, the predicted change of the new coating shows a change in adhesion. In an additional or alternative configuration, the predicted change of the new coating shows a change in colorability. In an additional or alternative configuration, the one or more machine learning algorithms use a supervised learning algorithm to calculate the expected aging of the new coating formulation. In an additional or alternative configuration, the supervised learning algorithm includes one or more of (i) a random forest technique, (ii) an XGBoost tree technique, or (iii) a support vector machine technique. In an additional or alternative configuration, the method may also include displaying a predicted colorimetric value for the selected coating formulation; and displaying one or more changes in the predicted colorimetric value of the selected coating formulation after one or more future time intervals.

[0081] In additional or alternative configurations, the present disclosure may provide a computerized system having a database and one or more computer processors; computer executable instructions stored on the database, which, when executed, cause the computerized system to perform the following operations: identifying initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having multiple components, and (ii) colorimetric data of the coating when applied; identifying aged color data for the initial coating on the target object, wherein the aged color data includes colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to multiple aging exposure conditions, wherein the aging exposure conditions include exposure to heat, light, and precipitation; determining changed colorimetric data between the initial color data and the aged color data, wherein the changed colorimetric data is associated with (i) each of the coating components and (ii) multiple aging exposure conditions; and processing the changed colorimetric data through one or more machine learning algorithms, wherein the one or more machine learning algorithms store in the database an association between each of the multiple components and a rate of change of the colorimetric data determined when the multiple aging exposure conditions are applied.

[0082] In an additional or alternative configuration, the aged color data is measured at least by a spectrophotometer. In an additional or alternative configuration, the aged color data represents a plurality of measurements of an initial coating on a target object over a time range of years. In an additional or alternative configuration, the aged color data represents a plurality of measurements of an initial coating on a target object over a time range of months. In an additional or alternative configuration, the aged color data represents a plurality of measurements of colorimetric data of the initial coating at a plurality of different measurement angles. In an additional or alternative configuration, the system is further configured to: receive an end-user selection of a new coating having a different plurality of compositions provided via a graphical user interface, wherein the different plurality of compositions differs from any composition of any basecoat or one or more compositions in the initial coating.

[0083] In additional or alternative configurations, the system is further configured to: apply the altered colorimetric data using one or more machine learning algorithms to determine a predicted aging pattern of the new coating when a plurality of aging exposure conditions are applied; and display data corresponding to the predicted aging pattern of the new coating on a graphical user interface. In additional or alternative configurations, the system is further configured to: display one or more ingredient substitution suggestions for the new coating; wherein the one or more ingredient substitution suggestions are optimized by the one or more machine learning algorithms to minimize changes in appearance in response to the plurality of aging conditions. In additional or alternative configurations, the one or more ingredient substitution suggestions displayed include changes in concentration of any ingredient in the plurality of ingredients.

[0084] Although the subject matter has been described in language specific to structural features and / or methodological acts, and although specific examples of the disclosure have been described above for illustrative purposes, it will be apparent to those skilled in the art that many changes may be made to the details of the disclosure without departing from the disclosure as defined in the appended claims. Specifically, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the features or the acts described above, or the order of the acts described above. In fact, the features and acts described are disclosed as example forms of implementing the claims.

Claims

1. A computer-implemented method for automatically determining and displaying expected degradation of a coating formulation after aging exposure, the computer-implemented method comprising: identifying initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having a plurality of components and (ii) colorimetric data of the coating as applied; identifying aged color data for the initial coating on the target object, wherein the aged color data comprises colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, the plurality of aging exposure conditions comprising at least heat, light, and water exposure; determining altered colorimetric data between the initial color data and the aged color data, wherein the altered colorimetric data is associated with (i) each of the coating components and (ii) the plurality of aged exposure conditions; processing the changed colorimetric data through one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of weathering exposure conditions are applied; receiving an end-user selection of a new coating having a different plurality of compositions provided via a graphical user interface; determining, using the one or more machine learning algorithms applying the altered colorimetric data, a predicted degradation pattern of the new coating when the plurality of weathering exposure conditions are applied; as well as Data corresponding to the predicted degradation pattern of the new coating is displayed on the graphical user interface. 2 . The computer-implemented method of claim 1 , wherein the aging condition comprises subjecting the target object to natural aging in a natural environment over a period of multiple months.

3. The computer-implemented method of any one of the preceding claims, wherein the aging condition comprises a simulation of an aging condition by a burn-in tester.

4. The computer-implemented method of any preceding claim, further comprising: identifying a historical data set for the initial coating; Wherein the historical data set includes the changed colorimetric data associated with the plurality of weathering exposure conditions and one or more different applications of the initial coating on different target objects subjected to different plurality of weathering exposure conditions.

5. The computer-implemented method of any preceding claim, further comprising: An expected change in colorimetric data associated with each of the plurality of aging exposure conditions and the different plurality of aging exposure conditions for each of the components is determined using the one or more machine learning algorithms applying the historical data set.

6. The computer-implemented method of any preceding claim, further comprising: determining a failure type for the initial coating after applying the plurality of aging exposure conditions; The failure types include color change.

7. The computer-implemented method of any preceding claim, further comprising: determining a failure type for the initial coating after applying the plurality of aging exposure conditions; The failure types described include gloss change or adhesion change.

8. The computer-implemented method of any preceding claim, further comprising: receiving a user selection reflecting one or more changes to a composition of the new coating; An aging change of the new coating upon application of any of the plurality of aging exposure conditions is determined.

9. The computer-implemented method of any preceding claim, further comprising: The predicted changes of the new coating after a plurality of time intervals of simulating application of the plurality of aging exposure conditions are displayed on a graphical interface.

10. The computer-implemented method of claim 9, wherein: The predicted change in the new coating shows a change in adhesion.

11. A computer-implemented method according to any one of claims 9 or 10, wherein The predicted change of the new coating shows a change in coloration.

12. A computer-implemented method according to any one of the preceding claims, wherein: The one or more machine learning algorithms employ a supervised learning algorithm to calculate the expected aging of a new coating formulation.

13. The computer-implemented method of claim 12, wherein the supervised learning algorithm comprises one or more of (i) a random forest technique, (ii) an XGBoost tree technique, or (iii) a support vector machine technique.

14. The computer-implemented method of any preceding claim, further comprising: Displays predicted colorimetric values ​​for the selected coating formulation; as well as One or more changes in predicted colorimetric values ​​for a selected coating formulation after one or more future time intervals are displayed.

15. A computerized system, comprising: a database and one or more computer processors; Computer executable instructions stored on the database, which, when executed, cause the computerized system to: identifying initial color data for an initial coating applied to a target object, wherein the initial color data identifies (i) a coating having a plurality of components and (ii) colorimetric data of the coating as applied; identifying aged color data for the initial coating on the target object, wherein the aged color data comprises colorimetric data corresponding to the initial coating after the coating applied to the target object has been subjected to a plurality of aging exposure conditions, the plurality of aging exposure conditions comprising exposure to heat, light, and precipitation; determining altered colorimetric data between the initial color data and the aged color data, wherein the altered colorimetric data is associated with (i) each of the coating components and (ii) the plurality of aged exposure conditions; as well as The changed colorimetric data is processed by one or more machine learning algorithms, wherein the one or more machine learning algorithms store in a database an association of each of the plurality of components with a rate of change of the colorimetric data determined when the plurality of aging exposure conditions are applied.

16. The computerized system of claim 15, wherein: The aged color data is measured at least via a spectrophotometer.

17. A computerized system according to any one of claims 15 or 16, wherein: The aged color data represents a plurality of measurements of the initial coating on the target object over a period of years.

18. A computerized system according to any one of the preceding claims 15 to 17, wherein: The aged color data represents a plurality of measurements of the initial coating on the target object over a period of a plurality of months.

19. A computerized system according to any one of the preceding claims 15 to 18, wherein: The aged color data represents a plurality of measurements of colorimetric data for the initial coating at a plurality of different measurement angles.

20. The computerized system of any one of the preceding claims 15 to 19, wherein the system is further configured to: An end-user selection of a new coating having a different plurality of ingredients is received, provided via a graphical user interface, wherein the different plurality of ingredients differs from any basecoat or any of the one or more ingredients in the initial coating.

21. The computerized system of claim 20, wherein the system is further configured to: using the one or more machine learning algorithms applying the altered colorimetric data to determine a predicted degradation pattern of the new coating when the plurality of weathering exposure conditions are applied; and Data corresponding to the predicted degradation pattern of the new coating is displayed on the graphical user interface.

22. The computerized system of any one of claims 20 or 21, wherein the system is further configured to: Display one or more ingredient substitution suggestions for the new coating; Wherein the one or more ingredient replacement recommendations are optimized by the one or more machine learning algorithms to minimize changes in appearance in response to multiple aging conditions.

23. The computerized system of claim 22, wherein the one or more ingredient substitution suggestions displayed include changes in concentration of any ingredient in the plurality of ingredients.

Images