Equipment coating deterioration evaluation program and system

Abstract

To provide an apparatus coating deterioration evaluation program and its system that can evaluate coating repair timing with high accuracy by evaluating deterioration over time by considering specifications of a coated apparatus and specifications of coating, in addition to the degree of coating deterioration and coating deterioration promotion elements, while objectively and efficiently evaluating the degree of coating deterioration with high accuracy without the need for visual judgement during checking work.SOLUTION: An apparatus coating deterioration evaluation system 1a includes a trained coating deterioration evaluation model 16 that has learned by inputting coating deterioration image teacher data 11 and coating deterioration degree teacher data 13 obtained by photographing a state of coating in order to evaluate degree of deterioration of the coating of an apparatus whose surface is coated, and has an evaluation unit 6 that inputs coating deterioration image data 22 obtained by photographing the state of the coating to the trained coating deterioration evaluation model 16 and outputs an evaluated degree of coating deterioration as coating deterioration degree evaluation data 25.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a program and system for evaluating deterioration of paint on equipment and devices installed in power plants, substations, etc. [Background technology]

[0002] Conventionally, in power plants and substations, standard items related to equipment painting, such as the basic principles of painting repair and methods for evaluating the deterioration state, have been established for the painting of facilities and equipment that deteriorates over time, and rational painting plans have been formulated based on appropriate deterioration evaluations, thereby effectively maintaining and managing equipment painting. However, these deterioration evaluations are based on human visual inspection, and even if deterioration evaluations are based on established standard items, it is difficult to make an objective evaluation, and it cannot be denied that there is a possibility that variations will occur depending on the level of skill. In addition, even in places other than power plants, the deterioration state of coatings on various structures is commonly diagnosed, inspected, and managed. For example, Patent Document 1 discloses an invention called a "coating diagnosis system" that diagnoses the deterioration state of the coating of a steel structure and the base material under the coating. With this invention, it is possible to register the results of multiple investigations for each diagnosis area (visual measurement of the deterioration area rate, analysis of the deterioration area rate using image processing, measurement of corrosion of the coated metal, chemical analysis of the coating, measurement of the coating thickness, and measurement of the coating adhesion) in a database and make a comprehensive judgment. Furthermore, Patent Document 2 discloses an invention entitled "Paint Deterioration Diagnosis System," which makes it possible to remotely photograph the paint film surface on the surface of a structure and evaluate the state of deterioration based on the peeling condition, rust condition, and color condition of the paint film surface from the image data. Furthermore, Patent Document 3, titled "Painted Surface Inspection Device, Learning Machine, and Program," discloses a technology that uses a neural network or the like to perform machine learning on a set of "training input information" and "training output information" in order to use captured images to identify cracked, peeled, and blistered areas of the paint as areas of deterioration other than rust. Finally, Patent Document 4 discloses a technology called a "paint management system" that manages the paint condition of large structures and plant equipment. The system determines an area in which to observe the paint condition (rust formation and paint peeling), periodically takes digital photographs of that area, and performs image analysis processing to score the progress of paint deterioration or to determine the time for repairs based on the quantitative values ​​obtained from the image analysis. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2009-236609 A [Patent Document 2] JP 2005-283519 A [Patent Document 3] JP 2020-159793 A [Patent Document 4] JP 2004-20498 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, while the inventions disclosed in these patent documents make it possible to objectively inspect, diagnose, or manage the deterioration state of the painted surface of a structure to a certain extent, they do not reflect the effects of equipment specifications, such as the environment in which the equipment is used, what devices are placed inside the equipment and what the internal environment is, or the condition of the equipment's exterior, or the effects of painting specifications, such as what paint is used and how much of a coating is applied based on the equipment specifications, and therefore there is a problem that they may be insufficient in terms of accuracy over time. As a result, although it is possible to diagnose and evaluate the deterioration state of the painted surface at a certain point in time, there is a problem in that there is a lack of accuracy in assessing how much time will pass from that point when repainting or repair is required.

[0005] The present invention has been made in response to such conventional circumstances, and aims to provide an equipment paint deterioration evaluation program and system which can objectively, efficiently, and highly accurately evaluate the degree of paint deterioration without the need for visual judgment during inspection work, and which can evaluate the timing of paint repairs with high accuracy by evaluating deterioration over time taking into account not only the degree of paint deterioration and factors that promote paint deterioration, but also the specifications of the equipment to which the paint is applied and the specifications of the paint. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, the first invention, an equipment paint deterioration evaluation program, is a program executed by a computer to evaluate the degree of deterioration of the paint of equipment having a painted surface, and is characterized in that it has a trained paint deterioration evaluation model that has been trained by inputting paint deterioration image teacher data photographing the state of the paint and paint deterioration degree teacher data, inputs paint deterioration image data photographing the state of the paint into this trained paint deterioration evaluation model to evaluate the deterioration degree of the paint, and outputs the evaluated deterioration degree as paint deterioration degree evaluation data. In the equipment paint deterioration evaluation program having the above configuration, the trained paint deterioration degree evaluation model operates to output paint deterioration degree evaluation data in response to input of paint deterioration image data.

[0007] In addition, the second invention is an equipment paint deterioration evaluation program, which is a program executed by a computer to evaluate the repair timing of the paint of equipment having a painted surface, and is characterized in that it has a trained first paint repair timing evaluation model that has been trained by inputting paint deterioration image teacher data photographed of the state of the paint, equipment specification teacher data, paint specification teacher data, paint deterioration promotion factor teacher data, and paint repair timing teacher data, and inputs paint deterioration image data photographed of the state of the paint, equipment specification data, paint specification data, and paint deterioration promotion factor data into this trained first paint repair timing evaluation model to evaluate the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data. In the equipment paint deterioration evaluation program having the above configuration, the trained first paint repair timing evaluation model operates to output paint repair timing evaluation data for the equipment in response to input of paint deterioration image data, equipment specification data, paint specification data, and paint deterioration promotion factor data.

[0008] The third invention is an equipment paint deterioration evaluation program, which is a program executed by a computer to evaluate the timing of repairing the paint of equipment having a painted surface, and which includes a trained paint deterioration evaluation model trained by inputting paint deterioration image training data obtained by photographing the state of the paint and paint deterioration degree training data, and a first paint deterioration degree evaluation data evaluated using the trained paint deterioration evaluation model as training data, and when the paint deterioration degree evaluation training data, equipment specification training data, paint deterioration promotion factor training data, and the paint deterioration degree evaluation training data are output, the paint deterioration degree evaluation model is evaluated based on the trained paint deterioration degree evaluation model. and a trained second paint repair timing evaluation model which has been trained by inputting paint repair timing teacher data in which the paint repair timing corresponding to the first paint deterioration image data input into the trained paint deterioration degree evaluation model, and inputting the second paint deterioration degree evaluation data output by inputting second paint deterioration image data photographed of the state of the paint into the trained paint deterioration degree evaluation model, equipment specification data, paint specification data, and paint deterioration promotion factor data to evaluate the repair timing of the paint, and outputting the evaluated repair timing as paint repair timing evaluation data. In the equipment paint deterioration evaluation program configured as described above, the trained second paint repair timing evaluation model operates to output paint repair timing evaluation data for the equipment by inputting the paint deterioration degree evaluation data output by the trained paint deterioration evaluation model, equipment specification data, paint specification data, and paint deterioration promotion factor data.

[0009] The fourth invention is an equipment paint deterioration evaluation system, which is characterized in that it has a trained paint deterioration evaluation model trained by inputting paint deterioration image teacher data photographing the state of the paint and paint deterioration degree teacher data in order to evaluate the degree of deterioration of the paint of equipment having a painted surface, and has an evaluation unit that inputs paint deterioration image data photographing the state of the paint into this trained paint deterioration evaluation model to evaluate the degree of deterioration of the paint, and outputs the evaluated degree of deterioration as paint deterioration degree evaluation data. In the equipment paint deterioration evaluation system configured as described above, an evaluation unit equipped with a trained paint deterioration degree evaluation model evaluates the degree of paint deterioration by inputting paint deterioration image data, and outputs it as paint deterioration degree evaluation data.

[0010] The fifth invention is an equipment paint deterioration evaluation system, which is characterized in that it comprises a trained first paint repair timing evaluation model that is trained by inputting paint deterioration image teacher data photographed of the state of the paint, equipment specification teacher data, paint specification teacher data, paint deterioration promotion factor teacher data, and paint repair timing teacher data in order to evaluate the timing of repairing the paint of equipment having a painted surface, and has an evaluation unit that inputs paint deterioration image data photographed of the state of the paint, equipment specification data, paint specification data, and paint deterioration promotion factor data into this trained first paint repair timing evaluation model to evaluate the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data. In the equipment paint deterioration evaluation system configured as described above, an evaluation unit equipped with a trained first paint repair timing evaluation model evaluates the paint repair timing by inputting paint deterioration image data, equipment specification data, paint specification data, and paint deterioration promotion factor data, and outputs this as paint repair timing evaluation data.

[0011] The sixth invention is an equipment paint deterioration evaluation system for evaluating the timing of repairing a paint on an equipment surface, comprising: a trained paint deterioration evaluation model trained by inputting paint deterioration image teacher data photographing the state of the paint and paint deterioration degree teacher data; and a first paint deterioration degree evaluation data evaluated using the trained paint deterioration evaluation model as teacher data, and the first paint deterioration image data input to the paint deterioration evaluation model when the paint deterioration degree evaluation teacher data, equipment specification teacher data, paint specification teacher data, paint deterioration promotion factor teacher data, and the paint deterioration degree evaluation teacher data are output. and a trained second paint repair timing evaluation model trained by inputting paint repair timing teacher data in which paint repair timing corresponding to the paint deterioration data is used as teacher data, and the trained second paint repair timing evaluation model is characterized in having an evaluation unit that inputs second paint deterioration degree evaluation data output by inputting second paint deterioration image data photographed of the state of the paint into the trained paint deterioration degree evaluation model, equipment specification data, paint specification data, and paint deterioration promotion factor data to evaluate the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data. In the equipment paint deterioration evaluation system configured as described above, an evaluation unit equipped with a learned paint deterioration degree evaluation model and a learned second paint repair timing evaluation model evaluates the paint repair timing by inputting the second paint deterioration degree evaluation data, equipment specification data, paint specification data, and paint deterioration promotion factor data, and operates to output this as paint repair timing evaluation data. In the sixth invention, there are two types of paint deterioration evaluation data: first paint deterioration evaluation data and second paint deterioration evaluation data. The first paint deterioration evaluation data is output by inputting the first paint deterioration image data into a trained paint deterioration evaluation model, and serves as training data for generating a second paint repair timing evaluation model. The second paint deterioration evaluation data is output by separately inputting the second paint deterioration image data into the trained paint deterioration evaluation model after the second paint repair timing evaluation model is generated. Effect of the Invention

[0012] The first invention, the equipment paint deterioration evaluation program, enables objective, efficient, and highly accurate evaluation of the degree of paint deterioration from paint deterioration image data that photographs the paint condition, without the need for visual inspection, which requires skilled personnel.

[0013] The second invention, the equipment paint deterioration evaluation program, makes it possible to objectively, efficiently, and highly accurately evaluate the timing of paint repairs based on paint deterioration image data photographed of the paint condition, equipment specification data, paint specification data, and paint deterioration promoting factor data, without the need for visual inspection, which requires skilled personnel.

[0014] Like the second invention, the third invention, which is an equipment paint deterioration evaluation program, is capable of objectively, efficiently, and highly accurately evaluating the timing of paint repairs based on the evaluated paint deterioration degree evaluation data of the first invention, equipment specification data, paint specification data, and paint deterioration promotion factor data, without the need for visual inspection, which requires skilled personnel.

[0015] Like the first invention, the fourth invention, the equipment paint deterioration evaluation system, makes it possible to objectively, efficiently, and highly accurately evaluate the degree of paint deterioration from paint deterioration image data that photographs the condition of the paint, without the need for visual inspection, which requires skilled personnel.

[0016] As with the second invention, the fifth invention, the equipment paint deterioration evaluation system, makes it possible to objectively, efficiently and highly accurately evaluate the timing of paint repairs based on paint deterioration image data photographed of the paint condition, equipment specification data, paint specification data and paint deterioration promoting factor data, without the need for visual inspection, which requires skilled personnel.

[0017] As with the fifth invention, the sixth invention, the equipment paint deterioration evaluation system, makes it possible to objectively, efficiently, and highly accurately evaluate the timing of paint repairs from the second paint deterioration evaluation data evaluated using the trained paint deterioration evaluation model, the equipment specification data, the paint specification data, and the paint deterioration promotion factor data, without the need for visual inspection, which requires skill. [Brief description of the drawings]

[0018] [Figure 1] 1 is a block diagram of an equipment paint deterioration evaluation system according to a first embodiment of the present invention. [Diagram 2] FIG. 1 is a simplified external view of a transformer used to explain an equipment paint deterioration evaluation system according to a first embodiment. [Diagram 3] 1 is a photograph of a transformer used to explain the equipment paint deterioration evaluation system according to the first embodiment. [Figure 4] (a)-(d) are all photographs showing examples of paint deterioration due to rust on the transformer shown in Figure 3, with the deterioration points judged to be 10 points, 20 points, 30 points, and 40 points, respectively. [Diagram 5] This is a conceptual diagram of a list compiled to treat paint deterioration on parts (components) of equipment as paint deterioration of the entire equipment. [Figure 6] FIG. 2 is a flow diagram for explaining generation of a paint deterioration degree evaluation model and a first paint repair timing evaluation model in the equipment paint deterioration evaluation system according to the first embodiment. [Figure 7] FIG. 1 is a conceptual diagram showing the configuration of a paint deterioration evaluation model using a neural network. [Figure 8] FIG. 2 is a conceptual diagram showing the configuration of a first paint repair timing evaluation model using a neural network. [Figure 9] FIG. 11 is a flow chart for explaining the evaluation of the deterioration degree of a paint performed using a paint deterioration degree evaluation model stored in the trained model database of the equipment paint deterioration evaluation system of the first embodiment. [Figure 10] FIG. 11 is a flow chart for explaining the evaluation of paint repair timing performed using a first paint repair timing evaluation model stored in the trained model database of the equipment paint deterioration evaluation system of the first embodiment. [Figure 11] FIG. 5 is a block diagram of an equipment paint deterioration evaluation system according to a second embodiment of the present invention. [Figure 12] FIG. 11 is a flow diagram for explaining generation of a second paint repair timing evaluation model in the equipment paint deterioration evaluation system according to the second embodiment. [Figure 13] FIG. 13 is a conceptual diagram showing the configuration of a second paint repair timing evaluation model using a neural network. [Figure 14] FIG. 11 is a flow chart for explaining the evaluation of paint repair timing performed using a second paint repair timing evaluation model stored in the trained model database of the equipment paint deterioration evaluation system of the second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] (First embodiment) Hereinafter, a system for evaluating deterioration of coating on equipment according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 4. FIG. FIG. 1 is a block diagram of an equipment paint deterioration evaluation system 1a according to the first embodiment. 1, the equipment paint deterioration evaluation system 1a is composed of a calculation device 2, a teacher data database 8, a trained model database 15, an input database 19, and an output database 24. This equipment paint deterioration evaluation system 1a can be assumed to be a system equipped with a calculation circuit and a storage device capable of performing each function within a computer server.

[0020] The calculation device 2 is equipped with an input unit 3 that inputs data from outside the system or from a database within the system, a trained model generation unit 4 that generates a trained model using training data, an evaluation unit 6 that evaluates the degree of deterioration of the equipment paint and the timing of repairs using the trained model and data necessary for evaluation, and an output unit 7 that outputs data outside the system. 1 do not have to be integrated, but the input unit 3 and output unit 7 can be provided as independent, separate units from the arithmetic device 2 to form a system or device that transmits and receives data and information between the trained model generation unit 4 and the evaluation unit 6 via wired or wireless communication. Note that although the word "unit" is used for the input unit 3 etc. in this application, this means a part of the arithmetic device 2, and itself constitutes a device or circuit.

[0021] The teacher data database 8 is a database that stores teacher data used to generate a trained model in the trained model generation unit 4, and readably stores equipment specification teacher data 9, paint specification teacher data 10, paint deterioration image teacher data 11, paint deterioration promotion factor teacher data 12, paint deterioration degree teacher data 13, and paint repair timing teacher data 14. In this embodiment, the teacher data database 8 and the trained model generation unit 4 are included in the equipment paint deterioration evaluation system 1a, but they may be provided separately. That is, the teacher data database 8 may be used in another system to generate the paint deterioration degree evaluation model 16 and the first paint repair timing evaluation model 17 to be stored in the trained model database 15 by the trained model generation unit 4, and only these trained evaluation models may be stored in the trained model database 15, and only the trained model database 15 may be included in the equipment paint deterioration evaluation system 1a. The same applies to the second embodiment.

[0022] The trained model database 15 is a database for storing trained models generated by the trained model generation unit 4 using teacher data, and readably stores a paint deterioration evaluation model 16 and a first paint repair timing evaluation model 17. These evaluation models are read by the evaluation unit 6.

[0023] The input database 19 is a database that stores data required for the evaluation unit 6 to use the learned model to perform evaluation of the degree of paint deterioration and evaluation of the timing of repairs, and stores equipment specification data 20, paint specification data 21, paint deterioration image data 22, and paint deterioration promotion factor data 23. These data are the same as the contents of the respective training data described with reference to FIG. 8, so the description thereof will be omitted here.

[0024] The output database 24 is a database for storing data relating to the degree of deterioration and repair timing of the paint evaluated by the evaluation unit 6 using the input data, and stores paint deterioration degree evaluation data 25 and paint repair timing evaluation data 26.

[0025] The equipment coating deterioration evaluation system 1a will be described in detail below. The input section 3 of the calculation device 2 is used to input the aforementioned data stored in the teacher data database 8 and the input database 19, and also to input evaluation models to be stored in the trained model database 15 if these evaluation models are generated externally. Specifically, the input unit 3 may be a keyboard, a mouse, a pen tablet, an optical reading device, or a receiving terminal for receiving data from other devices or an information communication network. The output unit 7 of the calculation device 2 has a function of outputting to the outside paint deterioration degree evaluation data 25 relating to the deterioration degree of the paint evaluated by the evaluation unit 6 and paint repair timing evaluation data 26 relating to the repair timing of the paint. In addition, it is also possible to output data stored in the teacher data database 8 or the input database 19 or evaluation models stored in the trained model database 15, as necessary. Specifically, the output unit 7 may be an output device such as a display device or a printer device, or further, a transmitting device such as a transmitter for transmitting to an external device, or a transmission terminal, etc. Of course, it may also be something like an interface for output for transmission to an external device or an information communication network.

[0026] Next, the paint deterioration image data 22 used to evaluate the degree of paint deterioration and repair timing in the equipment paint deterioration evaluation system 1a will be explained with reference to Figs. 2 to 4. Fig. 2 is a simplified outline drawing of a transformer used to explain the equipment paint deterioration evaluation system 1a. Fig. 3 is a photograph of the transformer used to explain the equipment paint deterioration evaluation system 1a, and Fig. 4(a)-(d) are photographs showing examples of paint deterioration due to rusting of the transformer in Fig. 3, where the deterioration points are judged to be 10 points, 20 points, 30 points, and 40 points, respectively. Fig. 5 is a conceptual diagram of a list compiled to capture paint deterioration of parts (components) of equipment as paint deterioration of the entire equipment. 2 and 3, a transformer 30 is composed of a main body 31, a conservator 32 (shown only in FIG. 2), a radiator 33, boxes 34, a duct 35, and the like. The main body 31 contains the iron core and windings, immersed in insulating oil for cooling the windings. The transformer repeatedly heats and cools depending on the load, and the insulating oil also expands and contracts repeatedly. During this process, air flows in and out of the transformer (breathing action), so it is necessary to prevent oxidation of the insulating oil. To prevent this breathing action, a conservator 32 is installed on the top of the main body 31. This conservator 32 contains an air bag inside, and absorbs the expansion and contraction of the insulating oil by discharging air from the air bag when the insulating oil expands and drawing in outside air when the insulating oil contracts. The radiator 33 is used to dissipate Joule heat generated by the transformation during operation, and the boxes 34 are used to house panels and instruments used to perform operations required for the operation of the transformer. The duct 35 is used to house the wiring from the low-voltage bushing.

[0027] The transformers 30 in substations vary depending on the degree of step-down, but they are generally large, and therefore the area to be painted is large. In addition, because they contain insulating oil for cooling the windings, a certain degree of temperature is always maintained during operation, which has a large impact on the deterioration of the paint, unlike the paint surface temperature of structures such as bridge piers and steel towers. Moreover, since they are installed outdoors, they are easily affected by the environment in which they are installed, and so electric power companies regularly inspect transformers 30, not only to check their functionality, but also to check for deterioration over time, such as breakdown of insulation inside the iron core, deformation and breakage of the windings, and external damage and thermal deterioration. Furthermore, deterioration of the paint also affects the tanks and casings and can cause leakage of insulating oil and the infiltration of rainwater and outside air into the tanks and casings. Therefore, inspection workers visually check for rust, peeling paint, or cracks in the paint, as shown in Figure 4 (a)-(d). Therefore, in the present invention, first, a score called a deterioration point is assigned when the inspector visually inspects the surface, and based on the accumulated data, deep learning using a neural network is performed to generate an evaluation model. In particular, because rusting is the earliest deterioration phenomenon that causes paint peeling and cracking, the proportion of deterioration points, which will be explained later, is allocated higher than that of paint peeling and cracking. Rules are created for the state of rust, peeling, and cracks, and the inspectors give each state a deterioration score. Based on these rules, the inspectors inspect the paint of various equipment. Specifically, when an inspector visually inspects the rust condition, a deterioration score as shown in Table 1 is assigned.

[0028] [Table 1]

[0029] The state of rust in paint is diagnosed based on the area of ​​rust, with 0% rusted area being no abnormality, and over 5% being considered defective, with each degradation point being rated on a scale from 0 to 50. As specific examples, Figure 4(a) is rated at 10 points, (b) at 20 points, (c) at 30 points, and (d) at 40 points. Next, Table 2 shows the evaluation of deterioration points based on the diagnosis of paint peeling, and Table 3 shows the evaluation of deterioration points based on the diagnosis of paint cracking.

[0030] [Table 2]

[0031] [Table 3]

[0032] Paint peeling is diagnosed based on the size and area of ​​the peeling, and is rated from no abnormality to peeling occurring over most of the surface area, with the deterioration level being rated on a scale from 0 to 10. In addition, paint cracks are diagnosed based on the area of ​​cracks, and are rated on a scale ranging from no abnormalities to cracks occurring over most of the surface area, with the deterioration score being rated on a scale from 0 to 10.

[0033] When evaluating deterioration points based on these three elements (rusting, peeling, and cracking), the occurrence area is the subject of evaluation, so the area that will be used as the standard for the occurrence area is set in advance, and when obtaining the paint deterioration image teacher data 11 and the paint deterioration image data 22, images that correspond to that standard are obtained. In addition, because these deterioration points are evaluated by parts (components) of the equipment, they are summarized in a list as shown in Figure 5. This list is used to determine whether or not painting repairs should be performed on the entire equipment, and the parts items are listed vertically and the material items are listed horizontally. The material items corresponding to the parts items are the parts that are not hatched in Figure 5, but they are different for each part item. For example, the material items "tanks and plates" and "pipes" correspond to the part item "main body," and for radiators No. 1 and No. 2, the material item "tanks and plates" corresponds to the "main body," and the material item "frame" corresponds to the support frame. The material item "tanks and plates" corresponds to the part item "accessories." The total of the deterioration points for the three elements for each part item in each row is a value between 0 and 70, and the average value is entered in the rightmost column. The average value of all of these part items, i.e., the 11 part items related to "Main Body", "Radiator No. 1", "Radiator No. 2" and "Accessories", becomes the deterioration point of the entire target equipment. In addition, if there are many part items used, the method of calculating the average value may be devised appropriately depending on the number of items. For example, if the deterioration points of all 11 part items are evaluated, the average value of the deterioration points of the top 70% (up to the top 8th) is used as the deterioration point of the entire target equipment. In this way, by calculating the average value of the sum of the deterioration points of the three elements for each part item and calculating the deterioration point of the entire target equipment facility, the deterioration level of the entire target equipment is evaluated and the paint repair level is determined as shown in Table 4. In other words, the evaluation of the deterioration points based on these three elements and the evaluation of the deterioration level of the entire equipment correspond to paint deterioration evaluation data 25. The method of calculating and evaluating such deterioration points is not particularly limited, as long as paint rust, paint peeling, and paint cracks are used as elements to evaluate the deterioration points of individual parts and the entire equipment.

[0034] [Table 4]

[0035] Paint repair levels are set according to the deterioration level of the paint on the entire equipment facility, with deterioration level 1 meaning no painting is necessary, and level 2 meaning full painting is not necessary and partial paint repairs will be carried out. Deterioration level 3 is evaluated as a level that requires planning and consideration of full repairs, and partial paint repairs will be carried out. Deterioration levels 4 and 5 cannot be addressed with partial paint repairs and full paint repairs are required. Therefore, when deterioration level 3 is reached, a plan for full-scale repainting repair must be considered, and records must also be kept of any cases in which deterioration level 4 or 5 is actually reached and full-scale repainting repair is carried out. Such evaluations of paint deterioration by experienced inspection workers are accumulated, and images of the paint deterioration of the equipment used in these evaluations are used as input training data, and the evaluation results (deterioration points) are used as output training data to perform deep learning using a neural network, thereby constructing a trained paint deterioration evaluation model 16. In addition, records of full-scale repainting repairs are also accumulated based on the evaluation of the deterioration level of the target equipment as a whole, and an evaluation of changes over time, i.e., how much time must actually pass from the time of evaluation of the deterioration point based on the three elements related to paint deterioration before a full-scale repainting repair is required, is also accumulated. Therefore, in this embodiment, the images of the deteriorated paint of the equipment used to evaluate these deterioration levels are used as input training data, and the time when the full-scale paint repair was actually carried out after the images of the deteriorated paint were taken is used as training data, and deep learning is performed using a neural network to construct a learned first paint repair timing evaluation model 17.

[0036] A specific method for constructing the paint deterioration evaluation model 16 and the first paint repair timing evaluation model 17 will be described with reference to Figs. 6 to 8. Fig. 6 is a flow diagram for explaining the generation of the paint deterioration evaluation model 16 and the first paint repair timing evaluation model 17 of the equipment paint deterioration evaluation system 1a according to this embodiment. This Fig. 6 also shows the process executed by a computer for the equipment paint deterioration evaluation program of the present invention, and explaining the flow of data processing in the equipment paint deterioration evaluation system 1a with reference to this figure is equivalent to explaining an embodiment of the equipment paint deterioration evaluation program. The same applies to Figs. 9, 10, 12, and 14. FIG. 7 is a conceptual diagram showing the configuration of a paint deterioration evaluation model 16 using a neural network, and FIG. 8 is a conceptual diagram showing the configuration of a first paint repair timing evaluation model using a neural network.

[0037] In Figure 6, in steps S1 and S2, the learned model generation unit 4 of the calculation device 2 reads out paint deterioration image teacher data 11 and paint deterioration degree teacher data 13 regarding the evaluation of deterioration points for three elements of paint deterioration, which have been stored in advance in the teacher data database 8 from the input unit 3. Then, in step S3, the trained model generating unit 4 performs deep learning to learn the relationship between the feature quantities related to rusting, peeling off of paint, and cracking of paint contained in the paint deterioration image teacher data 11 corresponding to the deterioration points and the deterioration degree of the paint (deterioration points). At this time, the paint deterioration image teacher data 11 is input and the paint deterioration degree data output is compared with the paint deterioration degree teacher data 13, and the parameters used in the intermediate layer are optimized so that the paint deterioration data, which is the output value, approaches the paint deterioration degree teacher data 13. Examples of these parameters include weights (coupling coefficients) between neurons and coefficients used in activation functions described later. The optimization method of these parameters is performed using, for example, the backpropagation method or the like. In this way, the trained model generation unit 4 generates a paint deterioration evaluation model 16 using a neural network that receives input of paint deterioration image data 22 and outputs paint deterioration evaluation data 25, and stores this in a readable manner in the trained model database 15. In this embodiment, a convolutional neural network (CNN) that is suitable for image processing is used as this neural network, but a network constructed with other learning algorithms such as a support vector machine (SVM) may be used as long as the correct correlation between the input and the output can be maintained. In particular, the second paint repair timing evaluation model 18 described in the second embodiment evaluates the paint repair timing without using paint deterioration image data, so a neural network that does not use a convolution layer or a pooling layer may be configured.

[0038] The paint deterioration evaluation model 16 using CNN generated in this way will be described with reference to Fig. 7. In Fig. 7, the input layer of the paint deterioration evaluation model 16 includes a plurality of neurons to which the pixel values ​​of each pixel included in the image information related to paint rust, paint peeling, and paint cracks included in the paint deterioration image data 22 are input. These pixel values ​​are then sent to the intermediate layer, where a function for the trained paint deterioration evaluation is formed, and the intermediate layer includes a plurality of neurons that extract feature amounts included in the image information according to the function. Since the paint deterioration assessment model 16 employs CNN, in the intermediate layer, convolution layers that perform a convolution process for convolving the pixel values ​​of each pixel and pooling layers that perform a pooling process for downsampling the convolved pixel values ​​are alternately arranged to compress the pixel information of the image while extracting image features, while appropriately combining and arranging fully connected layers using affine functions that perform linear transformation, and using an activation function that performs nonlinear transformation such as tanh with them. Furthermore, when determining the deterioration point for each part of the equipment within a desired numerical range, the final function of the intermediate layer may be a softmax function or the like that outputs the input value as a probability and selects the deterioration point with the highest probability. The learned function for evaluating the degree of deterioration of the paint is configured by combining these functions. The output layer has neurons that output evaluation values ​​of deterioration points related to the three elements of paint rusting, paint peeling, and paint cracking for each equipment part, evaluates the deterioration degree of the equipment paint, and outputs it as paint deterioration degree evaluation data 25. The evaluation unit 6 stores this paint deterioration degree evaluation data 25 in the output database 24 so that it can be read out. As described above, the deterioration level of the equipment as a whole is calculated by finding the average value of the sum of the deterioration points of the three elements for each part item. The evaluation unit 6 calculates the average value of the sum of the deterioration points of the three elements for each part item output by the paint deterioration degree evaluation model 16, evaluates the deterioration level of the equipment as a whole in accordance with the numerical conversion shown in Table 4, outputs it as paint deterioration degree evaluation data 25, and stores it in the output database 24 so as to be readable.

[0039] Next, returning to Fig. 6, a description will be given of the generation of the first paint repair timing evaluation model 17. Step S4 is a process in which the trained model generation unit 4 reads out the paint deterioration image teacher data 11 stored in advance in the teacher data database 8 from the input unit 3, and step S5 is a process in which the trained model generation unit 4 similarly reads out the equipment specification teacher data 9. The equipment specification teacher data 9 is teacher data on the material and plate thickness of the painted part, the inclination angle of the painted part, and the temperature of the painted part, as shown in Fig. 8. Painted steel is often used as the material of the painted part, but zinc, aluminum, magnesium alloys, etc. may also be used to improve corrosion resistance. In addition, the plate thickness of the painted part is also used as teacher data because it affects deterioration such as rusting of the paint. The inclination angle of the painted area affects the wetness time after rainfall, and is used as training data because a long wet time with rainwater from rainfall can lead to deterioration such as rusting. The larger the inclination angle, the shorter the wet time, but this inclination angle of the painted area is optional training data and may or may not be included in the equipment specification training data 9. In addition, with regard to the temperature of painted parts, equipment used in power generation and transmission businesses, such as transformers, often requires cooling to prevent heat generation due to Joule heat inside, and the temperature of the casing surface can lead to deterioration of the paint, such as rust, so the temperature is used as training data.

[0040] Next, in step S6, the trained model generation unit 4 reads out the painting specification teacher data 10 that has been pre-stored in the teacher data database 8 from the input unit 3, and in step S7, the trained model generation unit 4 similarly reads out the painting deterioration promotion factor teacher data 12. The paint specification teacher data 10 is teacher data of the paint film material and paint film thickness, which are also shown in FIG. 8. These are teacher data on the elements of the paint film itself that affect the deterioration of the paint. On the other hand, the paint deterioration promotion element teacher data 12 is teacher data of data on the elements that promote the deterioration of the paint. Specifically, it is teacher data of data on elements related to the environment in which the transformer is installed, such as the amount of rainfall, temperature, amount of ultraviolet rays, and amount of salt, which are shown in FIG. 8, that can promote the deterioration of the paint. The more the amount of rainfall, the higher the temperature, the higher the amount of ultraviolet rays, and the higher the amount of salt, the more the deterioration of the paint will be promoted. In addition, it is preferable to set the data on these elements using average values ​​or integrated values ​​within a certain period of time. Step S8 is a process in which the trained model generation unit 4 reads out paint repair timing teacher data 14 that has been stored in advance in the teacher data database 8 from the input unit 3. As described above, the paint repair timing teacher data 14 is teacher data that represents the time when a full-surface paint repair was actually carried out after the image of the paint deterioration was captured.

[0041] Then, in step S9, the learned model generation unit 4 performs deep learning to learn the paint repair timing contained in the paint repair timing teacher data 14 from feature amounts related to rust, paint peeling, and paint cracking contained in the paint deterioration image teacher data 11 corresponding to the deterioration points, the equipment specifications contained in the equipment specification teacher data 9, the paint specifications contained in the paint specification teacher data 10, and feature amounts related to paint deterioration promotion elements contained in the paint deterioration promotion element teacher data 12. In this case, similarly to step S3, paint deterioration image teacher data 11, equipment specification teacher data 9, paint specification teacher data 10, and paint deterioration promotion element teacher data 12 are input, and the output paint repair time data is compared with the paint repair time teacher data 14, and the parameters used in the intermediate layer are optimized so that the paint repair time data, which is the output value, approaches the paint repair time teacher data 14. As described above, these parameters include, for example, the weights (coupling coefficients) between neurons and coefficients used in the activation functions described below, and the parameter optimization method is performed using, for example, the error backpropagation method. Then, the trained model generation unit 4 generates a first paint repair timing evaluation model 17 using CNN that receives paint deterioration image data 22, equipment specification data 20, paint specification data 21, and paint deterioration promotion factor data 23 as input and outputs paint repair timing evaluation data 26, and stores this in the trained model database 15 in a readable manner.

[0042] The first paint repair timing evaluation model 17 using the CNN generated in this manner will be described with reference to FIG. 8, the input layer of the first paint repair timing evaluation model 17 receives image information on paint rusting, paint peeling, and paint cracking contained in the paint deterioration image data 22, the material of the painted part, the inclination angle of the painted part, and the temperature of the painted part contained in the equipment specification data 20, the material of the paint film, the thickness of the paint film contained in the paint specification data 21, and the amount of rainfall, temperature, amount of ultraviolet rays, and amount of salt at the location where the equipment is installed contained in the paint deterioration promotion element data 23. Therefore, it has a plurality of neurons corresponding to these inputs. Note that the inclination angle of the painted part in the equipment specification data 20 is a selective element, similar to the inclination angle of the painted part contained in the equipment specification teacher data 9, and is included if it is included in the equipment specification teacher data 9, and is not included if it is not included. These four types of data input values ​​are then sent to the middle layer, where a learned function for the first paint repair timing evaluation is formed and which has multiple neurons that extract features contained in the input information according to the function. The first paint repair timing evaluation model 17 also employs a CNN like the paint deterioration evaluation model 16, and the idea regarding the configuration of the function in the middle layer is the same as that of the paint deterioration evaluation model 16. The output layer includes neurons that output evaluation values ​​relating to the timing of paint repair, evaluates the timing of equipment paint repair, and outputs it as paint repair timing evaluation data 26. For example, neurons corresponding to the number of years expected in the future are provided for this evaluation value relating to the timing of paint repair, and a softmax function or the like is used as the last function in the middle layer to output the input value of the immediately preceding neuron as a probability, thereby selecting the number of years for the paint repair timing with the highest probability. In addition, the intermediate layer of the first paint repair timing evaluation model 17 is configured to calculate the deterioration level of the entire equipment calculated by the evaluation unit 6 from the evaluation values ​​of the deterioration points related to paint rust, paint peeling, and paint cracking outputted from the paint deterioration evaluation model 16, and finally, paint repair timing evaluation data 26 evaluating the painting timing is outputted in the output layer. Of course, the paint deterioration degree evaluation model 16 may be configured so that the deterioration level of the entire facility is calculated in the intermediate layer as well.

[0043] Next, a paint deterioration degree evaluation by the equipment paint deterioration evaluation system 1a according to the first embodiment will be described with reference to Fig. 9. Fig. 9 is a flow diagram for explaining the evaluation of the deterioration degree of paint executed using the paint deterioration degree evaluation model 16 stored in the trained model database 15 of the equipment paint deterioration evaluation system 1a. In FIG. 9, in step S10, the evaluation unit 6 of the calculation device 2 selects and reads out the paint deterioration degree evaluation model 16 from the trained model database 15. Also, in step S11 , the evaluation unit 6 reads out the paint deterioration image data 22 from the input database 19 , and inputs it to the input layer of the paint deterioration degree evaluation model 16 . Then, in step S12, the evaluation unit 6 executes an evaluation of the deterioration points related to the three elements related to the deterioration of each component of the equipment, that is, paint rusting, paint peeling, and paint cracking, using the middle layer of the paint deterioration degree evaluation model 16, and outputs the evaluation result as paint deterioration degree evaluation data 25 using the output layer. The evaluation unit 6 also calculates the deterioration level of the entire facility as described above, and outputs this as paint deterioration degree evaluation data 25. These are then stored in the output database 24 so that they can be read out. Of course, the evaluation unit 6 may use the output unit 7 to display or output the paint deterioration degree evaluation data 25 to the outside.

[0044] Next, a paint repair timing evaluation by the equipment paint deterioration evaluation system 1a according to the first embodiment will be described with reference to Fig. 10. Fig. 10 is a flow diagram for explaining a paint repair timing evaluation executed using a first paint repair timing evaluation model 17 stored in the trained model database 15 of the equipment paint deterioration evaluation system 1a. In FIG. 10, in step S13, the evaluation unit 6 of the calculation device 2 selects and reads out the first paint repair timing evaluation model 17 from the trained model database 15. In addition, in step S14, the evaluation unit 6 reads out paint deterioration image data 22 from the input database 19, in step S15, the evaluation unit 6 reads out equipment specification data 20 from the input database 19, in step S16, the evaluation unit 6 reads out paint specification data 21 from the input database 19, and further, in step S17, the evaluation unit 6 reads out paint deterioration promotion factor data 23 from the input database 19 and inputs it into the input layer of the first paint repair timing evaluation model 17. Then, in step S18, the evaluation unit 6 executes an evaluation regarding the paint repair timing using the intermediate layer of the first paint repair timing evaluation model 17, and outputs the evaluation result as paint repair timing evaluation data 26 by the output layer. The output paint repair timing evaluation data 26 is stored in an output database 24 by the evaluation unit 6 so as to be readable. The evaluation unit 6 may use the output unit 7 to display or output the paint repair timing evaluation data 26 to the outside.

[0045] In the equipment paint deterioration evaluation system 1a according to the first embodiment described above, since visual inspection, which requires skill, is not required, inspection can be performed by unskilled workers, which makes it easy to secure human resources and enables efficient inspection work, and also makes it possible to objectively obtain paint deterioration image data 22. Furthermore, based on paint deterioration degree teacher data 13 based on paint deterioration image teacher data 11 by a skilled inspection worker and paint repair timing teacher data 14 related to the time when paint repair was actually performed, a paint deterioration degree evaluation model 16 and a first paint repair timing evaluation model 17 are obtained by deep learning, and paint deterioration degree evaluation data 25 and paint repair timing evaluation data 26 are obtained using these, making it possible to perform objective, efficient, and highly accurate evaluation of the deterioration degree and repair timing.

[0046] (Second embodiment) Hereinafter, a system for evaluating deterioration of coating on equipment according to a second embodiment of the present invention will be described with reference to FIGS. FIG. 11 is a block diagram of an equipment paint deterioration evaluation system 1b according to the second embodiment. 11, the only difference between the configuration of equipment paint deterioration evaluation system 1b and the equipment paint deterioration evaluation system 1a according to the first embodiment is that paint deterioration degree evaluation teacher data 27 is added to the teacher data database 8, and a second paint repair timing evaluation model 18 is provided instead of the first paint repair timing evaluation model 17. The other identical components are given the same reference numerals, and a description of those components will be omitted. In the equipment paint deterioration evaluation system 1a according to the first embodiment, as described with reference to Fig. 6, in order to evaluate the paint repair timing, the paint deterioration image teacher data 11, the equipment specification teacher data 9, the paint specification teacher data 10, the paint deterioration promotion factor teacher data 12, and the paint repair timing teacher data 14 are inputted and deep learning using CNN is performed to construct the first paint repair timing evaluation model 17. However, in the present embodiment, instead of the paint deterioration image teacher data 11, the first paint deterioration degree evaluation data 25 outputted as a result of evaluation using the paint deterioration degree evaluation model 16 is inputted as the paint deterioration degree evaluation teacher data 27 for generating the second paint repair timing evaluation model 18. In other words, since the paint deterioration image is not included as the teacher data, although it depends on the accuracy of the input and output, the neural network may be constructed without necessarily using the convolution layer or the pooling layer. This difference will be explained further with reference to FIG.

[0047] FIG. 12 is a flow diagram for explaining generation of the second paint repair timing evaluation model 18 in the equipment paint deterioration evaluation system 1b according to the second embodiment. 12, in step S19, the learned model generation unit 4 of the calculation device 2 reads out, as teacher data, paint deterioration degree evaluation teacher data 27 prepared in advance for deep learning in order to generate the second paint repair timing evaluation model 18 from the teacher data database 8. This paint deterioration degree evaluation teacher data 27 is obtained by first generating the paint deterioration degree evaluation model 16 prior to the generation of the second paint repair timing evaluation model 18, and then storing the first paint deterioration degree evaluation data 25 output using the paint deterioration degree evaluation model 16 in the teacher data database 8 as teacher data for generating the second paint repair timing evaluation model 18. Furthermore, step S20 is a process in which the trained model generation unit 4 reads out equipment specification teacher data 9 that has been previously stored in the teacher data database 8 from the input unit 3. The contents of the equipment specification teacher data 9 have already been explained with reference to Fig. 8, so a detailed description is omitted. Next, step S21 is a process in which the trained model generation unit 4 reads out paint specification teacher data 10 that has been previously stored in the teacher data database 8 from the input unit 3, step S22 is a process in which the trained model generation unit 4 similarly reads out paint deterioration promotion element teacher data 12, and step S23 is a process in which the trained model generation unit 4 reads out paint repair timing teacher data 14 that has been previously stored in the teacher data database 8 from the input unit 3. The painting specification teacher data 10, the painting deterioration promoting factor teacher data 12, and the painting repair timing teacher data 14 have also been described with reference to Fig. 8, so a description thereof will be omitted. However, for the painting repair timing teacher data 14, the painting repair timing corresponding to the first painting deterioration image data 22 inputted into the painting deterioration degree evaluation model 16 when the painting deterioration degree evaluation teacher data 27 was outputted is used as the teacher data.

[0048] Then, in step S24, the learned model generation unit 4 performs deep learning to learn the paint repair timing included in the paint repair timing teacher data 14 from the evaluated feature amounts related to rust, paint peeling, and paint cracking included in the paint deterioration degree evaluation teacher data 27 corresponding to the deterioration point, the equipment specifications included in the equipment specification teacher data 9, the paint specifications included in the paint specification teacher data 10, and the feature amounts related to paint deterioration promotion elements included in the paint deterioration promotion element teacher data 12. The parameter optimization method used in this case is the same as that in the first paint repair timing evaluation model 17 in the equipment paint deterioration evaluation system 1a according to the first embodiment. In addition, unlike the paint deterioration evaluation teacher data 27 prepared for deep learning for generating the above-mentioned second paint repair timing evaluation model 18, after generating the second paint repair timing evaluation model 18, the trained model generation unit 4 generates a second paint repair timing evaluation model 18 to be used in a neural network that receives second paint deterioration evaluation data 25, which evaluates the degree of paint deterioration by inputting second paint deterioration image data 22 into the paint deterioration evaluation model 16, equipment specification data 20, paint specification data 21 and paint deterioration promotion factor data 23, and outputs paint repair timing evaluation data 26, and stores this in the trained model database 15 so as to be readable.

[0049] The second paint repair timing evaluation model 18 using the CNN generated in this manner will be described with reference to FIG. In Fig. 13, the input layer of the second paint repair timing evaluation model 18 receives feature information on paint rusting, paint peeling, and paint cracking contained in the paint deterioration evaluation data 25, the material of the painted part, the inclination angle of the painted part, and the temperature of the painted part contained in the equipment specification data 20, the material of the paint film, the thickness of the paint film contained in the paint specification data 21, and the amount of rainfall, temperature, amount of ultraviolet rays, and amount of salt at the point where the equipment is installed contained in the paint deterioration promotion element data 23. Therefore, it has a plurality of neurons corresponding to these inputs. Note that the inclination angle of the painted part in the equipment specification data 20 is a selective element as in the embodiment of the first paint repair timing evaluation model 17, and is included if it is included in the equipment specification teacher data 9, and is not included if it is not included. These four types of data input values ​​are then sent to the middle layer, where a learned function for evaluating the timing of the second paint repair is formed and which has multiple neurons that extract features contained in the input information according to the function. The second paint repair timing evaluation model 18 also adopts a CNN like the first paint repair timing evaluation model 17, and the idea regarding the configuration of the function in the middle layer is the same as that of the paint deterioration evaluation model 16 and the first paint repair timing evaluation model 17. The output layer includes neurons that output evaluation values ​​related to the paint repair timing, and evaluates the repair timing of the equipment paint and outputs it as paint repair timing evaluation data 26. In addition, the intermediate layer of the second paint repair timing evaluation model 18 is also configured to calculate the deterioration level of the entire equipment calculated by the evaluation unit 6 from the evaluation values ​​of the deterioration points related to paint rust, paint peeling, and paint cracking outputted from the paint deterioration degree evaluation model 16, and finally, paint repair timing evaluation data 26 evaluating the painting timing is outputted in the output layer, or, although not shown in Figure 13, the paint deterioration degree evaluation data 25 also includes the deterioration level of the entire equipment calculated by the evaluation unit 6 and the paint repair timing evaluation data 26 is outputted in the output layer.

[0050] Next, a coating repair timing evaluation by the equipment coating deterioration evaluation system 1b according to the second embodiment will be described with reference to Fig. 14. Note that the coating deterioration evaluation by the equipment coating deterioration evaluation system 1b is the same as that by the equipment coating deterioration evaluation system 1a according to the first embodiment, and therefore the description thereof will be omitted. FIG. 14 is a flow diagram for explaining the evaluation of paint repair timing performed using the second paint repair timing evaluation model 18 stored in the trained model database 15 of the equipment paint deterioration evaluation system 1b. In FIG. 14, in step S25, the evaluation unit 6 of the calculation device 2 selects and reads out the second paint repair timing evaluation model 18 from the trained model database 15. Furthermore, in step S26, the evaluation unit 6 reads out the second paint deterioration degree evaluation data 25 from the output database 24, in step S27, the evaluation unit 6 reads out the equipment specification data 20 from the input database 19, in step S28, the evaluation unit 6 reads out the paint specification data 21 from the input database 19, and further, in step S29, the evaluation unit 6 reads out the paint deterioration promotion factor data 23 from the input database 19 and inputs them into the input layer of the second paint repair timing evaluation model 18. In order to input the second paint deterioration degree evaluation data 25, the evaluation unit 6 must already input the second paint deterioration degree image data 22 using the paint deterioration degree evaluation model 16, perform an evaluation regarding the paint deterioration degree, and store the second paint deterioration degree evaluation data 25 in the output database 24. Then, in step S30, the evaluation unit 6 executes an evaluation regarding the paint repair timing using the intermediate layer of the second paint repair timing evaluation model 18, and outputs the evaluation result as paint repair timing evaluation data 26 using the output layer. The output paint repair timing evaluation data 26 is stored in an output database 24 by the evaluation unit 6 so as to be readable. The evaluation unit 6 may use the output unit 7 to display or output the paint repair timing evaluation data 26 to the outside. The effects of the equipment paint deterioration evaluation system 1b according to the second embodiment described above are similar to those of the equipment paint deterioration evaluation system 1a according to the first embodiment. [Industrial Applicability]

[0051] As described above, the inventions described in claims 1 to 6 can be used in equipment paint deterioration evaluation systems and programs that can objectively, efficiently, and highly accurately evaluate the deterioration of paint applied to industrial equipment and the timing of paint repairs without visual inspection by an inspection worker. [Explanation of symbols]

[0052] 1a, 1b... Equipment paint deterioration evaluation system 2... Calculation device 3... Input unit 4... Trained model generation unit 6... Evaluation unit 7... Output unit 8... Teacher data database 9... Equipment specification teacher data 10... Paint specification teacher data 11... Paint deterioration image teacher data 12... Paint deterioration promotion factor teacher data 13... Paint deterioration degree teacher data 14... Paint repair timing teacher data 15... Trained model database 16... Paint deterioration degree evaluation model 17... First paint repair timing evaluation model 18... Second paint repair timing evaluation model 19... Input database 20... Equipment specification data 21... Paint specification data 22... Paint deterioration image data 23... Paint deterioration promotion factor data 24... Output database 25... Paint deterioration degree evaluation data 26... Paint repair timing evaluation data 27... Paint deterioration degree evaluation teacher data 30... Transformer 31... Main body 32... Conservator 33... Radiator 34... Boxes 35…Duct

Claims

1. A program executed by a computer to evaluate the repair timing of the paint of equipment having a painted surface, the program comprising: a trained first paint repair timing evaluation model trained by inputting paint deterioration image teacher data photographed to show the state of the paint, equipment specification teacher data including the material of the painted area, the inclination angle of the painted area, and the temperature of the painted area, paint specification teacher data, paint deterioration promotion factor teacher data, and paint repair timing teacher data; the equipment paint repair timing evaluation model is trained by inputting paint deterioration image data photographed to show the state of the paint, equipment specification data including the material of the painted area, the inclination angle of the painted area, and the temperature of the painted area, paint specification data, and paint deterioration promotion factor data into the trained first paint repair timing evaluation model to evaluate the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data.

2. 2. The equipment paint deterioration evaluation program according to claim 1 , wherein the paint deterioration evaluation model is a trained model trained by inputting paint deterioration image teacher data photographed to show the state of the paint and paint deterioration degree teacher data, and the trained second paint repair timing evaluation model is trained by inputting paint deterioration degree evaluation teacher data having first paint deterioration degree evaluation data evaluated using the trained paint deterioration degree evaluation model as teacher data, equipment specification teacher data, paint specification teacher data, paint deterioration promotion element teacher data, and paint repair timing teacher data having paint repair timing corresponding to the first paint deterioration image data input to the paint deterioration degree evaluation model when the paint deterioration degree evaluation teacher data is output.

3. The equipment paint deterioration evaluation program according to claim 1 , wherein the equipment paint deterioration evaluation model is trained by inputting second paint deterioration image data photographed to show the state of the paint, the output second paint deterioration degree evaluation data, the equipment specification data, the paint specification data, and the paint deterioration promotion element data, and the trained second paint repair timing evaluation model is inputted to the trained paint deterioration evaluation model to evaluate the repair timing of the paint, and the evaluated repair timing is output as paint repair timing evaluation data.

3. An equipment paint deterioration evaluation system comprising: a trained first paint repair timing evaluation model which is trained by inputting paint deterioration image teacher data photographed to show the state of the paint; equipment specification teacher data including the material of the painted area, the inclination angle of the painted area, and the temperature of the painted area; paint specification teacher data; paint deterioration promotion factor teacher data; and paint repair timing teacher data, in order to evaluate the timing of repairing the paint of equipment having a painted surface; and an evaluation unit which inputs paint deterioration image data photographed to show the state of the paint, equipment specification data including the material of the painted area, the inclination angle of the painted area, and the temperature of the painted area, paint specification data, and paint deterioration promotion factor data into this trained first paint repair timing evaluation model to evaluate the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data.

4. a trained paint deterioration evaluation model which has been trained by inputting paint deterioration image teacher data photographed of the state of the paint and paint deterioration degree teacher data in order to evaluate the timing of repairing the paint of equipment having a painted surface; and a trained second paint repair timing evaluation model which has been trained by inputting paint deterioration degree evaluation teacher data, which has first paint deterioration degree evaluation data evaluated using the trained paint deterioration degree evaluation model as teacher data, equipment specification teacher data, paint specification teacher data, paint deterioration promotion element teacher data, and paint repair timing teacher data, which has as teacher data a paint repair timing corresponding to the first paint deterioration image data inputted to the paint deterioration degree evaluation model when the paint deterioration degree evaluation teacher data is outputted, wherein the equipment paint deterioration evaluation system is characterized in that it has an evaluation unit which inputs second paint deterioration image data photographed of the state of the paint into the trained second paint deterioration evaluation model, inputs the output second paint deterioration degree evaluation data, equipment specification data, paint specification data, and paint deterioration promotion element data into the trained paint deterioration evaluation model, evaluates the repair timing of the paint, and outputs the evaluated repair timing as paint repair timing evaluation data.

Images