Deterioration estimation system

Abstract

The degradation estimation system of the present invention can estimate the degradation of a product having contents sealed in a metal container with high reliability. The system comprises: a prediction model, which is formed by machine learning using data obtained by performing a storage test on a product as teacher data; an input unit, which inputs data for deterioration estimation; and an output unit, which outputs the degree of deterioration of the product obtained by using the prediction model based on the data input to the input unit, wherein the teacher data includes: container data on an actual metal container, content data on the contents sealed in the actual metal container, environmental data on the environment in which the actual product is stored, and degradation data on the degree of deterioration caused by the actual product, and the data input to the input unit includes: container data on a container of a target product for which deterioration is estimated, content data on the contents of the target product, and environmental data on the environment in which the target product is scheduled to be stored.

Description

Technical Field

[0001] The present invention relates to a system for estimating temporal degradation of a product obtained by sealing a predetermined content in a metal container, and in particular, to a system for estimating degradation such as corrosion (rust) using a machine learning model. Background Art

[0002] For canned beverages, canned foods, etc., containers (or cans) made of metals such as steel and aluminum alloys are used due to their superiority in strength, durability, or cost. In this regard, not only does the metal ionize and dissolve into liquids such as water, but also, since fruit juice, lactic acid beverages, or cooked foods are often acidic or alkaline, in order to prevent direct contact between the contents and the metal, at least the inner surface of the metal container is coated with a synthetic resin or an adhesive film. In this way, corrosion (rust) on the inner surface of the container and changes in the taste and aroma of the contents caused by this are suppressed.

[0003] However, it is impossible to completely and permanently prevent the contact between the contents and the container material by the inner surface coating, and corrosion (rust) may occur relatively early due to differences in the coating material, the composition of the contents, or the thickness of the coating. Therefore, in the past, storage tests were often conducted to confirm corrosion (rust) and other degradation of the metal surface. The storage test is a test in which the intended product is actually manufactured and sold, and the test product is stored under specified storage conditions for several months or more than ten months, and then the corrosion (rust) and other degradation are investigated and measured.

[0004] New products also require such so-called safety or reliability confirmation and assurance, but when new products are developed one by one according to market demand, there is a large discrepancy between the development cycle of new products and the time required for storage testing, and there is a possibility that the required storage testing will become an obstacle to the development of new products.

[0005] In addition, as a method for inspecting defects of the inner surface film of a metal container, an inner coating measurement method (Japanese: エナメルレーター) is known. This method is a method of applying a voltage between the content and the metal can through the inner surface film, finding the value of the flowing current, and detecting defects of the inner surface film based on the current value. Methods or devices that improve this inspection method are described in Patent Documents 1 and 2.

[0006] The device described in Patent Document 1 is a device that is configured to measure the resistance of the inner surface film after applying an impact to the can, thereby predicting whether a film defect occurs when the can is impacted, the degree of film damage, the progress of corrosion of the metal plate, etc. In addition, the method described in Patent Document 2 is a method for evaluating corrosion resistance by passing an electric current through the inner surface coating of the can molded body, and a voltage of 50mV to 200mV is applied between an electrode immersed in the content and the can body for 6 to 48 hours, and the corrosion resistance is evaluated based on the sequence of the accumulated current during this period.

[0007] Patent Document 1: Japanese Patent No. 2934164

[0008] Patent Document 2: Japanese Patent No. 5830910

[0009] As for the deterioration of the product such as corrosion of the inner surface of the metal container, changes in flavor and taste caused by the contact between the contents and the metal surface, it can be considered that the physical properties of the contents, the inconsistency of the thickness of the inner surface film caused by the characteristics of the production line of the container, the damage of the coating film due to deformation of the container by external force, or the environmental conditions of the storage location of the product are interrelated and occur over time. Therefore, in the past, an actual product was made and used as a test product, and a storage test was performed in which the test product was stored in an environment simulating the actual storage location environment. The test period is long as described above, and if the test results are not satisfied, an improved product is made and the storage test is performed again. Therefore, it takes a longer time to obtain a product that will not deteriorate even if it is stored for a specified period of time. As a result, it is difficult to produce products that meet market demand in a timely manner.

[0010] In contrast, according to the device described in Patent Document 1, if there is a defect in the inner surface film of the metal container, a large current flows, and therefore the inner surface film defect can be quickly detected based on the current value. However, the inner surface defect is limited to defects caused by the application of impact, and if it is not stored for a long period of time like the storage test, it is impossible to detect the corrosion of the inner surface, which is an example of the time-dependent degradation of the product. In addition, it is necessary to use an actual product as the inspection product, which has the disadvantage of requiring troublesome operations and time-consuming work, such as its manufacture.

[0011] In addition, similar to the invention described in Patent Document 1, the method described in Patent Document 2 also has the following problems: although the voltage is applied to detect defects, it takes more than 6 hours and less than 48 hours because it is a method for solving the sequence of the accumulated current amount, and it cannot be performed quickly. In addition, the product for its test needs to be an actual product, so there is a disadvantage that its manufacturing requires troublesome operations and time-consuming. In addition, the method described in Patent Document 2 cannot know the deterioration caused by the storage environment and the influence of time on the contents. Summary of the invention

[0012] The present invention was made under the above technical background, and an object of the present invention is to provide a system that can estimate the time-dependent degradation of a product obtained by sealing a content in a metal container without producing an actual product, thereby speeding up the development or production of the product.

[0013] In order to achieve the above-mentioned purpose, the present invention is a degradation estimation system for a product formed by sealing contents in a metal container, characterized in that it has: a prediction model, which is obtained by machine learning using data obtained by performing storage tests on the above-mentioned product as teacher data; an input unit, which inputs data for deterioration estimation; and an output unit, which outputs the degree of deterioration of the above-mentioned product obtained by using the above-mentioned prediction model based on the above-mentioned data input to the above-mentioned input unit, the above-mentioned teacher data includes: container data about the actual above-mentioned metal container, content data about the above-mentioned contents sealed in the above-mentioned actual metal container, environmental data about the environment in which the above-mentioned actual product is stored, and degradation data indicating the above-mentioned degree of deterioration caused by the above-mentioned actual product, and the above-mentioned data input to the above-mentioned input unit includes: container data about the above-mentioned container of the object product for which deterioration is estimated, content data about the above-mentioned contents of the above-mentioned object product, and environmental data about the environment in which the above-mentioned object product is scheduled to be stored.

[0014] In addition, in the present invention, the above-mentioned container is formed by a metal plate having an inner surface coating, and the above-mentioned degradation degree output from the above-mentioned output unit and the degradation data indicating the above-mentioned degradation degree contained in the above-mentioned teacher data include: the depth of corrosion of the inner surface of the above-mentioned metal plate, the contour shape of the location where the above-mentioned corrosion occurs, the dispersion state of the above-mentioned corrosion, and at least any one of the dissolution amount of the above-mentioned metal into the above-mentioned content.

[0015] Furthermore, in the present invention, an evaluation unit may be further provided for evaluating the degree of degradation output from the output unit and classifying the degree of degradation into a plurality of levels.

[0016] Alternatively, in the present invention, the container data may include any one of data indicating dimensions of each portion of the container, data indicating the material of the container, and data regarding a coating provided on the inner surface of the container; the content data may include any one of the type of the content, the amount of the content, and pH; and the environmental data may include the temperature of the environment in which the product is stored.

[0017] In the present invention, the output unit may be configured to output the degree of degradation as data divided into a plurality of degrees.

[0018] Furthermore, in the present invention, the evaluation unit may evaluate the degree of degradation based on the data classified into the plurality of degrees and labeled.

[0019] In the present invention, data on a target product that is estimated to deteriorate over time is input into a prediction model obtained by machine learning using data obtained by storing actual products as teacher data, thereby being able to estimate the degree of degradation of the target product. The teacher data is data related to the metal container of the actual product, data related to the content, data related to the environment in which the actual product is stored, and the degree of degradation of the actual product, and the correlation of these data is set with high accuracy. Regarding the target product that is estimated to deteriorate, data on the metal container, the content, and the storage environment are pre-defined as product items, and the data is prepared in the form of a database or the like and input to an input unit, and is calculated according to the prediction model, and the degree of degradation is output from an output unit. Therefore, according to the present invention, in the case of a newly developed or designed product, at the stage of determining the data on the above-mentioned items of the product, the degree of degradation is determined based on the data, so there is no need to actually manufacture a new product, or there is no need to store across the required storage period, so that the development or manufacture of the product can be accelerated.

[0020] More specifically, if the estimated result is a degree of degradation within the allowable range, the product with the data (specifications) determined in the design can be immediately put into the production line. In addition, in the case where it is estimated that unacceptable degradation has occurred, part of the data (specifications) of the metal container and the contents is changed and the degree of degradation is re-estimated using the above-mentioned prediction model. By repeating such re-estimation based on the prediction model, it is possible to set product specifications that do not cause degradation or make the degree of degradation within the allowable range. In this case, there is no need to manufacture actual products with changed specifications, and there is no need to conduct storage tests for a long time or a long period of time, so that the development or manufacture of products can be accelerated. Therefore, it is easy to estimate the degree of degradation when the specifications of metal containers, contents, etc. are changed, thereby making it easy to select or change metal containers and contents, and as a result, it is easy and quick to develop or create new products.

[0021] In the present invention, the degree of degradation can be output from the output unit as a numerical value or data divided into a plurality of parts as a result of calculation based on the prediction model. In addition, the estimated result represented by the numerical value or the data divided into a plurality of parts can be divided into a plurality of levels by the evaluation unit. The stratification can be a level where the degree of degradation is allowable, a level where it is recommended to actually manufacture the product and conduct a storage test, a level where it is necessary to reconsider the product specifications, and so on. In this way, the degree of degradation obtained by the prediction model can be more easily reflected in product development or design. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic block diagram for explaining one embodiment of the present invention.

[0023] Figure 2 This is a graph simply showing an example of a database of a storage test actually performed.

[0024] Figure 3 This is a diagram showing a simplified example of classification label data according to an embodiment of the present invention.

[0025] Figure 4 This is a table showing in simplified form an example of a necessity determination table according to an embodiment of the present invention.

[0026] Description of Reference Numerals

[0027] 1…teacher data; 2…machine learning; 3…prediction model; 4…output unit; 5…input unit; 6…evaluation unit. DETAILED DESCRIPTION

[0028] The degradation estimation system involved in the present invention is a system for estimating the time-dependent degradation of a product enclosed in a metal container, and is particularly configured as a system for estimating the degree of degradation of a product using a prediction model (learning model) that uses labeled data as teacher data for machine learning. The metal container constituting the product is a so-called two-piece can, a three-piece can, etc., in which the open end of the metal body is sealed with a metal cover, a so-called bottle-shaped can, etc., in which a part of the body is formed in a bottle shape and a cover is installed on its neck to seal it. In addition, the material of the metal container is a metal material known in the past, such as various steels, aluminum alloys, etc. In addition, at least on the inner surface of the metal container, in order to prevent the contents from directly contacting the metal, a coating is applied to at least the inner surface to form an inner surface film. The inner surface film can also be formed by applying a synthetic resin coating to a metal plate, or it can also be formed by pasting a synthetic resin film on a metal plate. Such an inner surface film can also be the same as the synthetic resin film known in the past.

[0029] Furthermore, on the other hand, the content constituting the product may be any of a beverage, cooked solid food, powdered or granular substances such as milk powder, and the like.

[0030] As for product degradation, in products using metal containers, metal corrosion (rust) is the main cause of degradation. In addition, corrosion (rust) may also cause an increase in leached metals, flavor deterioration, and changes in taste and color. For example, in the case of almost water-free contents such as milk powder and protein, it is not easy to consider degradation due to metal corrosion (rust), but there are cases where the product deteriorates over time due to oxidation of the contents themselves, flavor changes caused by synthetic resins and metals constituting the inner surface film, etc.

[0031] The deterioration of the product mentioned here is the change of the product over time, which is the change of each item obtained through the previous storage test. The test sample for the storage test is the same as the actual product. The storage test is performed by storing the test sample in a predetermined environment and opening the test sample after a predetermined period of time to investigate the corrosion (rust) of the metal container, the flavor or color of the content, the taste, and other changes in each item.

[0032] If we list examples of items (data) used to determine the inspection items (products), for metal containers, we can cite the can type name and the cover name. In addition, the can type name is a name given according to the shape, material and size of each metal container, for example, represented by a combination of letters and numbers. Similarly, the cover name is a name given according to the shape, material and size of each cover, for example, represented by a combination of letters and numbers. Therefore, the material, size (dimensions of each part), coating data on the inner surface film, use / non-use of bisphenol A, etc. of the metal container are determined by the can type name and the cover name. Here, for the coating data, more specifically, in the case of a coating, it is the type of the coating, the solvent, the solid content, the coating amount, the film thickness, etc., and in the case of forming the inner surface film by a film, it is the type, thickness, layer structure, crystallinity, initial type, thickness, etc. of the film.

[0033] The items for determining the contents include whether the contents have been distilled, sports drinks, alcoholic drinks, carbonated drinks, fruit juices, dairy products, high-calorie energy drinks, special / functional / quasi-drugs, solids, viscosity, vegetables, meat, seafood, etc. In addition, as chemical property values ​​(data) of the contents, there are Brix (sugar content), alcohol content, gas volume, air volume in the container, hue L as brightness, hue a indicating the intensity of red and green hues, hue b indicating the intensity of yellow and blue hues, color difference ΔE indicating the degree of separation of the linear distance between two colors in the color space, metal elution, taste of the total amount of the contents, internal pressure, pH, etc.

[0034] The test items (data) indicating the degree of change are the depth of corrosion (rust) on the inner surface of the metal container, the form of corrosion (rust) such as plane, spot, crack, blister, etc., the dispersion state of corrosion (rust), flavor, etc. The flavor is the sensory evaluation of the tester.

[0035] Considering that the degradation of products is not only affected by the storage time of the products but also by the storage environment (conditions) to a great extent, the storage time, storage temperature, pressure, brightness of lighting, amount of ultraviolet rays, duration of vibration associated with handling (transportation), etc. are used as items (data) for determining the storage environment (conditions).

[0036] The degradation estimation system of the present invention is configured to perform machine learning using data obtained from past storage tests on actual products as teacher data to create a prediction model (learning model), and estimate the degree of degradation using the prediction model. In addition, the storage test can be performed by setting a variety of storage forms of the product, such as making the product stand upright, stand upside down, and make the body part concave.

[0037] Figure 1 An example of a degradation estimation system is schematically represented by a block diagram. In the example shown here, teacher data 1 is data about a product for actual storage testing, data about a metal container (container data), data about the contents (content data), data about the storage environment (environmental data), and data about the degree of deterioration obtained by opening and investigating after a specified period of time (deterioration data). These data can also be stored in a database.

[0038] As mentioned above, the container data includes data such as shape, size, material, and material of the inner surface coating, and at least includes any one of the material and the size of each part. As mentioned above, the content data also includes the type of materials such as items obtained by distillation or carbonated beverages, and their chemical property values ​​such as pH, alcohol content or color tone, and in particular includes at least any one of the type, amount, and pH. In addition, the environmental data can also be all data about the aforementioned storage environment, but at least includes temperature. Moreover, the data related to the degree of deterioration equivalent to the answer includes at least any one of the depth, morphology, and flavor of corrosion (rust). The more these data are, the higher the accuracy of learning, but on the other hand, there is a situation where the amount of calculation becomes unnecessarily large. Therefore, it is preferred to use a specified judgment criterion for data processing. As an example, when using a neural network, the amount of data can also be reduced by a convolution layer or a pooling layer.

[0039] Machine learning is a conventionally known operation process that uses the above-mentioned teacher data as input data, performs operations using a neural network, and adjusts each parameter in such a way that the degree of degradation output as the operation result is as close as possible to the degree of degradation actually measured in the storage test (i.e., the answer). After the adjustment of each parameter is completed, a prediction model (learning model) 3 is generated. In addition, each parameter can be adjusted sequentially.

[0040] The prediction model 3 is, for example, a neural computer and has an input layer, a single or multiple intermediate layers, and an output layer, and adjusts parameters in such a way that the difference between the output value of a preset item and the data indicating the degree of degradation in the teacher data is as small as possible. The calculation result of the prediction model 3 is output from the output unit 4. The output may also be in the form of indicators such as continuous numerical values ​​for each item of each item, but may also be data obtained by dividing these indicators into multiple (or grouped or hierarchical) parts based on the order in which they are arranged, or by assigning a specified label to each part of the divided part (or group or hierarchy).

[0041] Figure 2 An example of data obtained through actual storage tests (storage test database) is shown. Figure 2 Only the content of "1 data" is recorded, but there is a large amount of accumulated measured data, so "2 data" and below also record test data based on "1 data", which become teacher data.

[0042] in addition, Figure 3An example of data (classification label data) obtained by dividing the calculation results into multiple data and assigning labels to the data and outputting them from the output unit 4 is shown. In the example shown here, the amount of metal dissolution, the depth of corrosion, the amount of corrosion (rust) judged by visual judgment, or the contour shape or degree of dispersion (corrosion morphology) are set as indicators indicating the degree of deterioration of the product. In addition, three periods of 3 months, 6 months, and 12 months are set as storage periods, and the temperature is set as environmental data. The classification label data is set in the order of "G1", "G2", "G3", etc., starting from the one with the lowest degree of deterioration, for each storage period, and the content of each classification label data is as follows: Figure 3 As described.

[0043] In order to estimate the degree of deterioration over time of a new product (design product) without performing a storage test, data (design data) on the design product (target product for deterioration estimation) is input from the input unit 5 to the prediction model 3, and the result of the calculation is output from the output unit 4. The design data may be data of each item included in the teacher data 1 used when generating the prediction model 3. As an example, the design data may be data of each item included in the teacher data 1 used when generating the prediction model 3. Figure 2 Since the design data will not be different from the teacher data as input data for the prediction model 3 even if its specific numerical value or display content is different from the teacher data, the output value calculated and output by the prediction model 3 will be the same or similar to the data indicating the degree of degradation in the teacher data 1. Therefore, for each item indicating the degree of degradation (metal dissolution amount, corrosion depth, corrosion morphology, etc.), the output is as follows: Figure 3 The classification label data (class labels) such as "G1" and "G2" are shown.

[0044] The calculation result output from the output unit 4 is a simple numerical value or label, but the numerical value or label indicates the estimated degree of degradation, and based on the output value, it can be determined whether the designed product has not deteriorated particularly within the pre-assumed period, or even if the degree of degradation is low, there is a possibility of failure, etc. Furthermore, if such a determination cannot be made and the estimated result is still unclear, a determination that further requires a previous storage test can be made.

[0045] Such a determination can be made manually, but if the number of items indicating the degree of degradation is large, there is a possibility that the evaluation or operation of the design product (new product) based on the estimated result of degradation may be inconsistent, or it may be difficult to evaluate. For example, if there are three items indicating the degree of degradation as described above, and four grade labels are assigned to each, the total number of combinations of grade labels for determining the quality of the design product is 64, and there is a possibility that it may be difficult to make a final determination of the quality of the design product, determine the design change, determine whether a storage test is required, etc., or inconsistency may occur. In order to eliminate such an undesirable situation, the evaluation unit 6 may be provided to perform a stable determination.

[0046] An example of the evaluation unit 6 is to determine whether a storage test is required for the combination of the above-mentioned level labels. The table used for the determination is as follows: Figure 4 shown. Figure 4 The example shown is an example in which the level output from the output unit 4 is applied to the three items indicating the degree of degradation, among the four levels described above. Figure 4 The level surrounded by "○" in the table represents the degree of deterioration of each item. Figure 4 In the example shown, all three items of the design product to be evaluated are at the least degraded level (G1), so it is determined that the design product will not degrade to the point of becoming a defective product even if stored for an assumed period under an assumed environment, and therefore it is determined that a storage test is not required. Alternatively, the determination result may be output, for example, as an "○" mark and displayed. Therefore, the design product can be produced without a storage test that requires a long period of time. In addition, the evaluation unit 6 may be configured to receive data from the output unit 4 and perform the evaluation, or may be built into the output unit 4 as a part thereof.

[0047] In contrast, when the level of any of the three items is other than the highest level "G1", it is determined that it is necessary to make an actual product and perform a storage test in which the product is stored under a specified environment for a specified period of time. Alternatively, the determination result may be output as a "△" sign and displayed. Figure 4 Although not shown, when there are two or more "G3"s indicating the progress of degradation or when there is at least one "G4", it may be determined as "redesign" to change the specifications of the metal container and the specifications of the contents. The result of the determination may be outputted and displayed, for example, with an "×" mark.

[0048] If such an evaluation is performed, the estimated degree of degradation can be effectively and stably used in product development or design. In particular, the number of design products that require actual storage tests for a long time (long period of time) can be reduced, so the development cycle of product development can be shortened, and market demands with short-term changes in demand can be promptly responded to. In other words, even if a storage test is actually performed, unnecessary man-hours such as supplying products with "unqualified" results to actual storage tests can be reduced, and such products can be redesigned immediately. Therefore, in this regard, the development cycle of product development can be shortened, and market demands with short-term changes in demand can be promptly responded to.

[0049] Furthermore, the result of the calculation based on the prediction model 3 is the result of estimating the degree of degradation, and it is not 100% guaranteed that the product will not deteriorate. The inventors of the present invention have created the above-mentioned prediction model, and verified the accuracy of the prediction model through the cross validation method, and the accuracy rate reached 70% to 95%. In other words, at least 5% error may occur. In the case where the product is a beverage, food, etc., it is necessary to further ensure the reliability of the estimated result. In such a case, by improving the evaluation criteria of the above-mentioned evaluation unit 6, the reliability of the estimated degradation can be further improved, and the system involved in the present invention can also be used for food, etc. In addition, improving the evaluation criteria can also increase the number of items representing the degree of degradation, or more finely distinguish the grade labels of each item, or further add measured data and update the parameters of the prediction model 3.

[0050] In addition, as described in the above embodiment, the product whose contents are food materials has a short development cycle due to changes in market demand, and its chemical property values ​​are various and different, so the system of the present invention that can estimate the degradation with high accuracy is highly effective as a system for estimating the degradation of such products. However, the present invention is not limited to a system that targets products whose contents are food materials, and can also be applied to the estimation of the degradation of products such as industrial products sealed in metal containers.

Claims

1. A degradation estimation system for a product in which a content is sealed in a metal container, characterized in that: have: A prediction model, which is formed by machine learning using data obtained by performing storage tests on the product as teacher data; an input unit that inputs data used for estimation of degradation; as well as an output unit that outputs the degree of deterioration of the product obtained by using the prediction model based on the data input to the input unit, The teacher data includes: container data about the actual metal container, content data about the content sealed in the actual metal container, environmental data about the environment in which the actual product is stored, and degradation data indicating the degree of degradation caused by the actual product. The data input to the input unit includes: container data about the container of the target product estimated to be deteriorated, content data about the content of the target product, and environmental data about an environment in which the target product is planned to be stored. The container is formed of a metal plate having an inner surface film. The degradation degree output from the output unit and the degradation data representing the degradation degree contained in the teacher data include: the depth of corrosion on the inner surface of the metal plate, the contour shape of the site where the corrosion occurs, the dispersion state of the corrosion, and at least any one of the amount of dissolution of the metal into the content.

2. The degradation estimation system according to claim 1, characterized in that: The deterioration data includes product deterioration data of changes in flavor and taste.

3. The degradation estimation system according to claim 1 or 2, characterized in that: An evaluation unit is further provided for evaluating the degree of degradation output from the output unit and classifying the degree of degradation into a plurality of levels.

4. The degradation estimation system according to claim 1 or 2, characterized in that: The container data includes any one of data indicating the size of each part of the container, data indicating the material of the container, and data on a coating provided on the inner surface of the container. The content data includes any one of the type of the content, the amount of the content, and pH, The environmental data includes the temperature of the environment in which the product is stored.

5. The degradation estimation system according to claim 3, characterized in that: The container data includes any one of data indicating the size of each part of the container, data indicating the material of the container, and data on a coating provided on the inner surface of the container. The content data includes any one of the type of the content, the amount of the content, and pH, The environmental data includes the temperature of the environment in which the product is stored.

6. The degradation estimation system according to claim 1 or 2, characterized in that The output unit is configured to output the degree of degradation as data divided into a plurality of degrees and labeled.

7. The degradation estimation system according to claim 3, characterized in that The output unit is configured to output the degree of degradation as data divided into a plurality of degrees and labeled.

8. The degradation estimation system according to claim 4, characterized in that The output unit is configured to output the degree of degradation as data divided into a plurality of degrees and labeled.

9. The degradation estimation system according to claim 5, characterized in that The output unit is configured to output the degree of degradation as data divided into a plurality of degrees and labeled.

10. The degradation estimation system according to claim 7, characterized in that: The evaluation unit evaluates the degree of degradation based on the data classified into the plurality of degrees and labeled.

Images