Information processing device, information processing method, and program

Abstract

To provide an information processor that can easily determine which change the soundness of a structure was determined on when the information processor automatically determined the soundness.SOLUTION: In the information processor, an evaluation region managing unit 201 selects a relation reference for setting change information used to determine the soundness of an evaluation region, selects an evaluation region to be a soundness determination target, and relates the change information according to the selected evaluation reference. A soundness determination unit 202 determines the soundness of the evaluation region by using the related change information and specification information stored in a specification information storing unit 207. The determination reference related to the determination is extracted from a plurality of determination references and is output as a determination basis.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, a control method of the information processing apparatus, and a program, and particularly relates to a technique for determining the soundness of an object.

Background Art

[0002] In the inspection of structures such as bridges and tunnels, the soundness is determined using deformation information such as cracks. The soundness is an index representing the degree of soundness of the structure. Note that the soundness can also be said to be an index representing the degree of deterioration and damage of the structure. In the conventional soundness determination, an inspection engineer determines the soundness of the structure based on information such as the number and width of cracks and the presence or absence of water leakage after grasping predetermined determination criteria and determination rules.

[0003] On the other hand, in order to reduce the burden on inspection engineers, a technique for automatically determining the soundness by an information processing apparatus has been disclosed. Patent Document 1 describes a method for determining the soundness by multiple regression analysis using inspection records such as the size of deformation as an input.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Non-Patent Documents

[0005]

Non-Patent Document 1

Non-Patent Document 2

[0006] However, in the prior art, consideration has not been given to facilitating the confirmation by an inspection technician of the automatically determined soundness. It is conceivable that an inspection technician performs an operation of confirming the automatically determined soundness determination result, such as confirming the agreement between the soundness determination result by the inspection technician and the automatically determined soundness determination result. At this time, in the conventional method, a user such as an inspection technician could not grasp the basis, such as based on which abnormality the information processing device determined the soundness. Therefore, in the prior art, a user such as an inspection technician could not grasp the basis of the automatically determined soundness determination result, and it was not easy to confirm whether the automatic determination result was correct or not.

[0007] The present invention has been made in view of the above problems, and an object thereof is to provide information on abnormalities related to the determination of soundness. [Means for Solving the Problems]

[0008] To solve the above problems , loveThe information processing device includes a first determination means that determines a first degree of soundness indicating the soundness of each of a plurality of evaluation regions in a structure, based on the deformation contained in the structure, A second determination means for determining a second soundness indicating the overall soundness of the structure based on the first soundness in each of the plurality of evaluation areas, Determination by the second determination means An output means that outputs information regarding the deformation of the structure that formed the basis for the second degree of soundness, Yes, The output means outputs information such that images extracted from each of the multiple deformations that formed the basis for determining the second degree of health are displayed in an order corresponding to the degree of contribution of each of the multiple deformations to the determination of the second degree of health. . [Effects of the Invention]

[0009] According to the present invention, it is possible to provide information on deformations related to the determination of soundness. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows the hardware configuration of the information processing device according to this embodiment. [Figure 2] This is a block diagram showing the functional configuration of an information processing device. [Figure 3] This is a processing flowchart of this embodiment. [Figure 4] This diagram illustrates the data handled by the information processing device according to this embodiment. [Figure 5] This is a diagram illustrating the learning process for determining health status. [Figure 6] This is a diagram explaining the method for determining health status. [Figure 7] This diagram explains the basis for the judgment output by the health assessment unit. [Figure 8] This is a diagram illustrating an example of display information according to Embodiment 1. [Figure 9] This figure illustrates another example of the display information according to Embodiment 1. [Figure 10] This diagram illustrates an example of changing the association criteria in Embodiment 1. [Figure 11] This figure illustrates another example of modifying the association criteria in Embodiment 1. [Figure 12] This figure illustrates another example of modifying the association criteria in Embodiment 1. [Figure 13] This is a diagram illustrating the determination rule for Embodiment 2. [Figure 14] This diagram illustrates the learning process for the health determination method of Embodiment 3. [Figure 15] This diagram illustrates the change in the association criteria in Embodiment 4. [Figure 16] This diagram illustrates the change in the association criteria in Embodiment 5. [Figure 17] This is a diagram illustrating the configuration of the information processing device in Embodiment 6. [Modes for carrying out the invention]

[0011] <Embodiment 1> Embodiments of the present invention will be described below with reference to the drawings. In this embodiment, the processing operation related to the information processing method by the information processing device 100 for determining the soundness of a structure will be described.

[0012] First, the data used in the structural integrity determination by the information processing device 100 of this embodiment will be explained using Figure 4. Figure 4 shows data that represents a portion of the unfolded view of the tunnel to be inspected. Although Figure 4 shows an example of a tunnel as the object to be inspected, the object to be inspected in this embodiment is not limited to tunnels; it may also be other structures such as bridges, dams, or buildings.

[0013] The tunnel layout diagram in Figure 4 is data created by superimposing images of the tunnel wall onto a drawing of the tunnel as it is laid out. Furthermore, deformation information indicating tunnel deformation is recorded in association with the tunnel layout diagram in Figure 4. For example, Figure 4 shows how deformations such as cracks 401, 402, and 403, water leakage 411, and efflorescence 412 and 413 are recorded. Crack 402 is represented by a thick line, indicating that it is a wider crack than crack 401. In this embodiment, information indicating the crack width, such as 0.2 mm width and 0.5 mm width, is recorded for each crack. For each crack, the maximum crack width value is recorded. The crack width stored as deformation information is not limited to the maximum value, but may also be an average value, a minimum value, or multiple values. The types of deformation are not limited to these examples and may include exposed rebar, honeycombing, rust stains, etc.

[0014] This deformation information is created by the information processing device 100 based on user input, using deformations observed by the user on-site at the structure under inspection. Alternatively, as shown in Figure 4, the user may visually inspect an image of the structure's wall surface superimposed on a drawing and input the location and extent of each deformation. Furthermore, the information processing device 100 may also use the results of automatic deformation detection based on the image as deformation information. Automatic deformation detection can be achieved, for example, by applying the target to a detection model that has been pre-trained using existing deformation images.

[0015] In this embodiment, the information processing device 100 stores the location and extent of deformation information indicating cracks as linear vector data. Furthermore, deformation information represented by areas such as water leakage is stored as polygon data. Hereinafter, these vector data and polygon data will be collectively referred to as vector data. This vector data is assumed to represent the coordinates of each point in a coordinate system based on the drawing.

[0016] Table 420 in Figure 4 shows an example of deformation information. For example, the deformation information indicating a crack with ID C401 includes vector data in the drawing coordinate system in the "Location" field, and further includes information on the crack width in the "Width" field. In addition, information on the crack length is included in the "Size" field. Note that for deformations in areas other than cracks, the "Size" field indicates the area of ​​the deformation.

[0017] Furthermore, the deformation information may also include information indicating deformations composed of combinations of multiple deformations. For example, the dotted frame 422 in Figure 4 indicates a closed crack composed of multiple cracks. H422 in Table 420 of the deformation information indicates this closed crack, and the location item shows the coordinate information indicating the dotted frame 422. In addition to closed cracks, deformation information composed of multiple cracks can also include tortoise-shell cracks and grid cracks.

[0018] Furthermore, information on the degree of damage to specific parts of a structure, such as structural members, may also be included in the deformation information. The degree of damage is an index that indicates the extent of the damage to the target. For example, 431 in Figure 4 shows a rock bolt in a tunnel, and the results of evaluating the degree of damage to this rock bolt may also be included in the deformation information. L431 in Table 420 is deformation information indicating damage to a rock bolt, and the location item records the coordinates of the range 432 that indicates the rock bolt 431. The degree of damage item records the degree of damage to the rock bolt 431 as C. The degree of damage may be expressed in three stages, for example, from A to C, with A being the most severely damaged.

[0019] The degree of damage can be assessed by a human inspecting the object on-site and determining the degree of damage, or by a human visually inspecting an image of the object and making a determination. Alternatively, the information processing device 100 can automatically determine the degree of damage from an image of the object. In this case, a determination model for automatically determining the degree of damage from an image can be generated using machine learning, and the information processing device 100 can use this determination model to determine the degree of damage. The damage determination model can be constructed by preparing a large amount of data in advance using existing data, consisting of pairs of images of objects to be damaged and training labels, and training the model with this data. An image to be damaged is, for example, an image like the area 432 showing the rock bolt in Figure 4. The results of a human's determination of the degree of damage to such an image are used as training labels. Training can be carried out using any machine learning algorithm such as Deep Learning.

[0020] As described above, deformation information includes information such as crack width, deformation size, and degree of damage. Below, this information, excluding location information, will be collectively referred to as attribute information.

[0021] As described above, the deformation information in this embodiment is as shown in Table 420 of Figure 4. As will be described later, this deformation information is stored in the information storage unit 206.

[0022] Next, the structural integrity determined in this embodiment will be explained. The structural integrity is determined using the deformation information described above to assess the structural integrity. As will be discussed later, the structural integrity may also be determined using the structural specifications information. Furthermore, the structural integrity in this embodiment is determined for each predetermined range of the structure. For example, in the tunnel example in Figure 4, the structural integrity is determined for each construction span of the tunnel (hereinafter referred to as span). Specifically, the structural integrity is determined for each of spans A, B, and C. Hereafter, such areas subject to structural integrity determination will be called evaluation areas. Note that evaluation areas are not limited to construction spans; for example, predetermined unit areas such as every 10m may be used as evaluation areas. Also, the structural integrity is determined on a scale of 1 to 5. A structural integrity of 5 is the most structurally sound, and a structural integrity of 1 is the most severely damaged. Note that the structural integrity is not limited to 5 levels.

[0023] In the following section, the configuration and processing of the information processing device 100 according to this embodiment will be described using the case of performing a health determination using the deformation information shown in Figure 4 as an example.

[0024] The configuration of the information processing device 100 in this embodiment will be explained using Figures 1 and 2.

[0025] (Hardware configuration) Figure 1 is a hardware configuration diagram of the information processing device 100 according to this embodiment. As shown in Figure 1, the information processing device 100 includes a CPU 101, a ROM 102, a RAM 103, an HDD 104, a display unit 105, an operation unit 106, and a communication unit 107. The CPU 101 is a central processing unit that performs calculations and logical decisions for various processes and controls each component connected to the system bus 108. The ROM (Read-Only Memory) 102 is a program memory that stores programs for control by the CPU 101, including various processing procedures described later. The RAM (Random Access Memory) 103 is used as the main memory, work area, and other temporary storage area of ​​the CPU 101. Program memory can also be realized by loading a program into the RAM 103 from an external storage device connected to the information processing device 100.

[0026] HDD104 is a hard disk for storing electronic data and programs according to this embodiment. An external storage device may be used to perform a similar role. Here, the external storage device can be implemented, for example, by a media (recording medium) and an external storage drive for accessing the media. Examples of such media include flexible disks (FD), CD-ROMs, DVDs, USB memory, MOs, flash memory, etc. The external storage device may also be a server device connected via a network.

[0027] The display unit 105 is, for example, a CRT display or a liquid crystal display, and is a device that outputs an image to the display screen. The display unit 105 outputs video according to the display control by the CPU 101. The display unit 105 may also be an external device connected to the information processing device 100 by wire or wireless. The operation unit 106 includes a keyboard or mouse and accepts various operations from the user. The communication unit 107 performs wired or wireless bidirectional communication with other information processing devices, communication equipment, external storage devices, etc., using known communication technologies.

[0028] Although the information processing device 100 is described as a single device, it is not limited to a single device and may consist of multiple devices. Furthermore, the information processing device 100 may be a logical device virtualized by multiple devices.

[0029] (Functional Configuration) Figure 2 is an example of a block diagram showing the functional configuration of the information processing device 100 according to this embodiment. The information processing device 100 includes an evaluation area management unit 201, a health determination unit 202, a deformation information analysis unit 203, a display information processing unit 204, a setting unit 205, a deformation information storage unit 206, a specifications information storage unit 207, and a learning unit 208. Each of these functional units is realized by the CPU 101 loading the program stored in the ROM 102 into the RAM 103 and executing processing according to the flowcharts described later. The results of each processing are then stored in the RAM 103 or HDD 104. Furthermore, for example, if hardware is configured as an alternative to software processing using the CPU 101, calculation units and circuits corresponding to the processing of each functional unit described here can be configured. Each functional unit may be configured as, for example, an ASIC or an FPGA.

[0030] Next, an overview of each functional unit in Figure 2 will be described. The deformation information storage unit 206 stores the deformation information described in Figure 4. The specification information storage unit 207 stores the specification information of the structure to be inspected. The specification information includes various types of information about the structure, such as the number of years since the start of service (service life), construction method, concrete type, geographical conditions such as coastal or cold region, and usage environment conditions such as traffic volume. The evaluation area management unit 201 associates the deformation information with the soundness evaluation area based on predetermined association criteria (association rules). The soundness determination unit 202 uses the deformation information and specification information associated by the evaluation area management unit 201 to determine the soundness of the evaluation area. The soundness determination unit 202 also outputs the basis for the soundness determination. The deformation information analysis unit 203 analyzes the deformations included in the basis for the soundness determination. The display information processing unit 204 displays information about deformations related to the health determination on the display unit 105 based on the analysis results of the deformation information analysis unit 203. The setting unit 205 receives user input via the operation unit 106 and controls the criteria for associating deformations with the evaluation area. The learning unit 208 learns the health determination method of the health determination unit 202. Note that not all of these functional units need to be included in the information processing device 100; at least some of their functions may be implemented by external devices that can be connected to the information processing device 100. Furthermore, each functional configuration may be realized by distributed processing by multiple devices.

[0031] (process) Next, the processing according to this embodiment will be described. The processing according to this embodiment will be described below using the flowchart shown in Figure 3. Note that the flowchart shown in Figure 3 is started when the information processing device 100 is instructed to perform a health status determination.

[0032] (Processing related to the Evaluation Area Management Department) In Figure 3, steps S301 to S303 are processes executed by the evaluation area management unit 201. Step S301 is the step of setting the association criteria. The association criteria are information that shows how to set the deformation information to be used to determine the health of the evaluation area. Specifically, the association criteria are the criteria for deciding whether or not to use deformations existing at the boundary of the evaluation area for determining the health. The association criteria can be changed by the setting unit 205, but the details of this will be explained later. Here, in step S301, the initial setting of the association criteria is used as the association criteria for subsequent association processing. In this embodiment, the initial setting of the association criteria is "Use deformation information within the evaluation area range of the evaluation target for determining the health."

[0033] In this embodiment, the structural integrity is determined sequentially for each evaluation area of ​​the structure under inspection. Step S302 is the step of selecting the evaluation area to be determined for structural integrity at the moment. In the following explanation, we will assume that span A in Figure 4 was selected as the evaluation area for structural integrity determination in step S302.

[0034] In the next step, S303, the evaluation area management unit 201 associates the abnormality information with the evaluation area according to the association criteria. That is, in step S303, the information processing device 100 selects the abnormality information to be used when determining the health of the evaluation area according to the association criteria. This process will be explained using Figure 4.

[0035] As mentioned above, we will explain assuming that the initial setting, "Use deformation information within the evaluation area range of the evaluation target for soundness determination," was set in step S301. The evaluation area management unit 201 selects the deformation information included in the coordinate range of span A from the deformation information storage unit 206 according to the set association criterion. For example, the range of span A in Figure 4 includes deformations such as cracks 401 and 402, water leakage 411, and efflorescence 412. In step S303, the deformation information of these deformations is used as the deformation information for determining the soundness of span A.

[0036] Here, crack 403 in Figure 4 is a crack that spans from span A to span C. For this crack 403, according to the current association criteria, the portion 421 of crack 403 that is included in span A will be used as deformation information for determining the soundness of span A. Therefore, the evaluation area management unit 201 changes the deformation information indicating crack 403 to deformation information indicating the crack up to the boundary of span A.

[0037] Specifically, the evaluation area management unit 201 obtains deformation information for crack 403 from the deformation information storage unit 206 and changes the vector data to include only the range 421 contained in span A. Furthermore, the size of the attribute information (crack length) is also changed to the length of the range 421. The deformation information, thus modified according to the association criteria for the evaluation area, is added to the deformation information for determining the soundness of span A. The evaluation area management unit 201 may also convert the deformation information indicating crack 403 into deformation information indicating the three cracks present in spans A, B, and C, respectively.

[0038] Through the processing by the evaluation area management unit 201 described above, deformation information for determining the health of span A is set.

[0039] (Processing related to the learning section) In the next step S304 in Figure 3, the information processing device 100 determines the health of the evaluation area using the deformation information associated with the evaluation area in the evaluation area management unit 201 and the specification information stored in the specification information storage unit 207.

[0040] First, the method for determining the health status will be explained. In this embodiment, the health status is determined using a pre-trained model that has been trained through machine learning. This training is performed in advance by the learning unit 208.

[0041] Here, in order to show the user the basis for the health assessment and to show information about the abnormalities related to the health assessment, it is preferable that the health assessment method be composed of judgment criteria that are understandable to humans, for example, binary judgment is preferred as the judgment criteria. Among the machine learning algorithms, a decision tree, which is a set of binary judgments, is a suitable algorithm for this embodiment. Below, an embodiment in which a decision tree is used as the health assessment method will be described. Note that the health assessment method is not limited to a decision tree, and health assessment may be performed using a trained model using other machine learning algorithms.

[0042] First, let's explain the training data for training the health assessment model. The training data for health assessment is created from existing inspection data and other deformation information, as well as data with known health levels. One training data set D i This can be expressed as follows:

[0043]

number

[0044] D i This is data relating to a specific evaluation area of ​​a certain structure. i This consists of deformation information associated with a certain evaluation area and specification information for a certain structure. i is X i Using this information, the health status data determined by a human is shown in this embodiment as a health status of 1 to 5. In the training data, this health status y i This will indicate the teacher class label. Such training data D i Collect a large amount of data to create a training dataset D={D i Create}.

[0045] Next, a method for learning a decision tree for soundness determination will be described using FIG. 5(A). The decision tree is composed of branch nodes 501 and 502 and a terminal node (leaf node) 503. At the branch nodes 501 and 502, the branching direction is determined based on a predetermined determination criterion. In FIG. 5(A), it is assumed that if the determination is True, the process proceeds to the node on the right side, and if the determination is False, the process proceeds to the node on the left side.

[0046] Next, the determination criteria for the branch nodes will be described. The determination criteria perform a binary determination using deformation information and specification information. Simply put, using the deformation information, the presence or absence of deformation such as "Is there a crack in the evaluation area?", "Is there a water leak in the evaluation area?", "Is there a closed crack?" can be the determination criteria. Also, a criterion combining the deformation type and attribute information may be used as the determination criteria. For example, in the determination criteria using the crack width of the attribute information, determination criteria such as "Is there a crack with a crack width of 0.5 mm or more?", "Is there a crack with a crack width of 1.0 mm or more?" can be created. That is, the presence or absence of deformation, the size or width of the deformation can be the determination criteria for soundness. At this time, by setting threshold values such as a width of 0.5 mm or a width of 1.0 mm for the attribute information, a plurality of different determination criteria can be created. Also, in the determination criteria using the deformation size of the attribute information, determination criteria such as "Is there a crack with a length of 5 m or more?", "Is there a water leak of 1 m 2 or more?" can be created. That is, the length or area of the deformation can be the determination criteria for soundness. Furthermore, the determination criteria using the damage degree of the attribute information can create determination criteria such as "Is there a bolt damage degree A?", "Is there a bolt damage degree B?". That is, the damage degree can be the determination criteria for soundness. Also, the number of deformations may be used as the determination criteria for soundness.

[0047] In this way, judgment criteria using deformation information can be created based on the presence or absence of deformation type, the type of deformation, and attribute information (deformation size, etc.). Judgment criteria can also be created by combining these. For example, one could create a judgment criterion such as, "Are there cracks with a width of 0.5 mm or more, AND is efflorescence present?"

[0048] Criteria using specification information can be created in the same way as criteria using deformation information. For example, criteria can be created such as "Has it been in service for 30 years or more?", "Has it been in service for 50 years or more?", "Is it located in a coastal area?", "Is it located in a cold region?", or "Is it PC concrete?". Alternatively, criteria can be created by combining deformation information and specification information. For example, criteria such as "Has it been in service for 30 years or more, and does it have cracks with a width of 0.5 mm or more?" can be created.

[0049] As described above, various patterns of decision criteria can be created. In decision tree training, effective decision criteria for separating training data are selected from these various criteria and set at each branch node. Below, as an example of a decision tree training method, we will explain a training method in which decision criteria for each branch node are set from candidate criteria prepared in advance by humans.

[0050] Table 500 in Figure 5(A) shows n candidate criteria C1 to Cn that have been prepared in advance by humans. First, we will explain the process of learning the criteria for the first branch node 501. In the first step of learning branch node 501, the training dataset D is subjected to binary classification using the n candidate criteria C1 to Cn to classify the data. In the second step, from the data classification results of the n candidate criteria, the candidate criterion that best classifies the training data is selected and used as the criterion for node 501. Here, "best classifying the training data" means the state in which the health of the split training dataset can be best classified. A concrete example of this will be explained using Figure 5(B).

[0051] In Figure 5(B), B1 shows the branch node 501 with the judgment criterion candidate C1 "Are there cracks with a width of 0.5 mm or more?" set, while B2 shows the branch node 501 with the judgment criterion candidate C4 "Water leakage 1m 2 The state is set to "Are there any of the above?". Histogram 510 is a histogram of the health levels included in training dataset D. In this example, it is assumed that the same number of training data for health levels 1 to 5 are prepared. In B1, training data judged as False by criterion candidate C1 are classified into the left node, and training data judged as True are classified into the right node. Furthermore, histograms 511 and 512 of the health levels of the datasets classified into each node are shown. Similarly, B2 shows the state in which the training dataset has been classified by criterion candidate C4, and histograms 513 and 514 of the health levels of each node are shown. Here, histograms 513 and 514 of each node in B2 show that the health levels have not been sufficiently classified.

[0052] On the other hand, in histograms 511 and 512 of B1, the health level is classified with good accuracy. Specifically, histogram 511 shows that the training data classified in the left node contains only data with health levels 4 and 5, while the training data classified in the right node contains only data with health levels 1, 2, and 3. In such a case, the candidate criterion C1 is a better criterion for classifying the training data than C4. By performing such comparisons on the training data classification results using each candidate criterion, the criterion that performs the best classification is set as the criterion for the node.

[0053] The basis for this comparison will be information gain. Information gain IG is expressed by the following formula.

[0054]

number

[0055] Here, D is the dataset, the subscript p is the parent node, and the subscripts left and right indicate the left and right child nodes, respectively. G is the Gini impurity, and N is the number of data points in the dataset. Since a larger information gain IG indicates better classification of the data, the information gain IG is calculated for the classification result of each candidate criterion, and the candidate criterion with the highest information gain IG is selected as the node's criterion.

[0056] Furthermore, since the information gain IG indicates the decision performance of each node, it can be used to calculate the importance of the decision criteria, which will be discussed later. Therefore, the information gain IG of each node calculated during the training of the decision tree is recorded as information for each node, along with the decision criteria.

[0057] In this way, the decision criteria to be set for node 501 can be determined. In the training of the next node, the decision criteria that best classifies the data are selected for the training dataset classified by the higher-level parent node, in the same manner as above. By repeating this process, the decision tree can be trained.

[0058] One way to terminate the learning of a decision tree is to repeat the classification until only one type of health score is contained in each node. However, decision trees learned in this way generally tend to overfit, so it is preferable to continue learning the nodes up to a predetermined depth (number of levels). The example in Figure 5(A) shows a situation where learning is terminated at a depth of 3. As a result, a terminal node (leaf node) is created, for example, node 503 in Figure 5(A). Figure 5(A) shows the health score histogram 505 of the training dataset contained in terminal node 503. Since histogram 505 contains multiple health scores, here we select the health score that appears most frequently and use it as the health score indicated by the terminal node. In Figure 5(A), the health score indicated by terminal node 503 is health score 5.

[0059] (Processing related to the health assessment unit) In step S304 of Figure 3, the health status of the target data is determined using the health status determination model learned as described above. Figure 6 is a diagram illustrating the health status determination process. First, deformation information and specification information of the evaluation area are input. Then, branching to child nodes is performed using the determination criteria of each node. As a result, in Figure 6, it is assumed that the terminal node 601 has been reached. Also, it is assumed that the health status indicated by terminal node 601 is health status 3. As a result, the health status of the input data is determined to be health status 3.

[0060] In the explanation above, we described a method for learning decision trees in which candidate criteria are prepared in advance and the criteria for each branch node are set from these candidate criteria. However, the method for learning the criteria for nodes is not limited to this method. For example, when learning each node, multiple candidate criteria may be randomly generated, and the criterion with the best classification performance may be selected from among these candidates. As a concrete example of a method for randomly creating candidate criteria, first, randomly select the type of deformation or the type of parameter information. Also, randomly select the threshold for attribute information, and create candidate criteria based on the combination of the type of deformation or parameter information and the attribute information threshold. By repeating this random selection process to create candidate criteria multiple times, multiple candidate criteria can be created. Alternatively, these candidate criteria may be combined to create candidate criteria consisting of multiple combinations of deformation information and parameter information, as shown in C7 and C8 in Figure 5.

[0061] Furthermore, the health determination unit 202 extracts the judgment criteria related to the health determination from the judgment criteria of the health determination method and outputs them as the basis for the determination. This process corresponds to step S305 in Figure 3. In this embodiment, the judgment criteria of the node in the decision tree where the determination is True are used as the basis for the determination. For example, in Figure 6, the determination is True at node 502. Let's assume that the judgment criterion for node 502 was "Total crack length 10m or more?". In this case, "Total crack length 10m or more?" becomes the basis for the determination. In the simple example in Figure 6, there is only one basis for the determination, but if multiple nodes in the decision tree are determined to be True, the health determination unit 202 outputs multiple basis for the determination. Note that although an example of using the judgment criterion that was set to True at a node in the decision tree as the basis for the determination has been shown, it is not limited to this. For example, the judgment criterion that was set to False at a node in the decision tree may also be used as the basis for the determination.

[0062] (Processing related to the Abnormal Information Analysis Department) Next, the process of step S306 in Figure 3 will be explained. Step S306 is a process executed by the deformation information analysis unit 203, which analyzes the deformations related to the soundness determination based on the judgment basis output by the soundness determination unit 202. One specific process of analyzing the deformations is to identify the type of deformation included in the judgment basis. In the example in Figure 6, the soundness determination unit 202 outputs "Total crack length 10m or more?" as the judgment basis. In this case, the deformation information analysis unit 203 identifies the deformation that was used as the basis for the soundness determination of the target evaluation area.

[0063] Figure 7 shows the information output by the structural integrity determination unit 202 as the basis for determining the structural integrity of tunnel span A in Figure 4. Using this example, an embodiment for identifying deformations related to structural integrity determination will be further explained, as well as the calculation of importance.

[0064] Figure 7 shows the results of extracting three criteria as the basis for the judgment when span A is judged to have a soundness level of 3. Figure 7 shows the case in which, when the deformation information and the basis for the judgment of span A are input into the decision tree, the input data is judged as True at nodes N502, N511, and N533 of the decision tree. In other words, span A is judged to have a soundness level of 3 based on the judgment criteria of nodes N502, N511, and N533. The soundness judgment unit 202 uses these judgment criteria as the basis for its judgment. Figure 7 shows the basis for the judgment obtained in this way and the information related to the basis for the judgment. The deformation information analysis unit 203 considers the deformation types included in the basis for the judgment to be deformations related to the soundness judgment. From the basis for the judgment in Figure 7, as shown in the item related to the type of deformation, cracks and water leakage are identified as deformations related to the soundness judgment.

[0065] Next, we will explain the process for calculating the importance of deformations related to health assessment. In the example in Figure 7, as mentioned earlier, each node of the decision tree has its information gain recorded during training. Figure 7 shows the information gains of decision tree nodes N502, N511, and N533. By aggregating these information gains for each type of deformation related to health assessment, we can calculate the importance of each type of deformation. The importance is an index that indicates the degree to which the target deformation contributes to the health assessment result. In the example in Figure 7, the node related to cracks is N502, so the importance of cracks is 0.32. Also, the nodes related to water leakage are N511 and N533, and the sum of their information gains gives the importance of water leakage to 0.26.

[0066] The above describes an embodiment for calculating the importance of each type of deformation based on the information gain recorded in the nodes of the decision tree. The method for calculating the importance of each type of deformation is not limited to this, and other methods may be used. For example, the number of times a type of deformation appears in the judgment criteria may be used as the importance. If the number of times a type of deformation appears in the judgment criteria is used as the importance, in the example in Figure 7, cracks appear only in the judgment criteria of node N502, so the importance of cracks is 1. Leaks appear in nodes N511 and N533, so their importance is 2.

[0067] Furthermore, the deformation information analysis unit 203 acquires location information for individual deformations related to the soundness determination. Specifically, it identifies the deformation used as the basis for the determination from the deformations in the evaluation area and obtains its location from the deformation information storage unit 206. For example, in Figure 7, the determination basis for node N502 is "Are there cracks with a width of 0.5 mm or more?", so cracks with a width of 0.5 mm or more are extracted from the deformations in the evaluation area. Then, location information (vector data) of cracks with a width of 0.5 mm or more is obtained from the deformation information storage unit 206. Also, the determination basis for node N511 is "Is there water leakage?", so water leakage is extracted from the deformations in the evaluation area and location information of the water leakage is obtained from the deformation information storage unit 206. Note that the determination basis for N533 is "Elapsed time is 30 years or more, and water leakage is 1 m 2 Since the above conditions are met, and the location information of the water leak has already been obtained, the process of obtaining location information for this determination is omitted.

[0068] As described above, in step S306, the deformation information analysis unit 203 identifies deformations related to the basis for the judgment, acquires their location information, and calculates the importance of each deformation. The deformation information analysis unit 203 also identifies deformations that satisfy the judgment criteria that formed the basis for the judgment. In the following steps, the deformation information analysis unit 203 uses this information to perform a process that visualizes the deformations related to the health judgment.

[0069] In the next step, S307, shown in Figure 3, if the soundness determination process for all evaluation areas of the structure under inspection has been completed, the process proceeds to step S308. If the soundness determination process for all evaluation areas has not been completed, the process returns to step S302, another evaluation area is selected, and the processes from steps S303 to S306 described above are executed.

[0070] (Processing related to the display information processing unit) Step S308 in Figure 3 is a process executed by the display information processing unit 204. The display information processing unit 204 creates and outputs display information showing the health assessment result and information on the deformations related to the health assessment, and displays this display information on the display unit 105. In other words, the display information processing unit 204 generates and outputs information on deformations that satisfy the assessment criteria that formed the basis of the assessment. Figure 8 is an example of information created by the display information processing unit 204 and displayed on the display unit 105, and the display information processing unit 204 will be explained below using this image.

[0071] The display window 800 in Figure 8 shows the GUI application window. The display window 800 also includes an image display area 801 and an importance display area 802. First, the image display area 801 displays the information described in Figure 4. Specifically, it displays a structural drawing, an image of the structure superimposed on the drawing, and deformation information. The deformation information is superimposed on the image as polylines and polygons based on vector data of positional information. In addition, the soundness judgment results are displayed for each evaluation area; for example, 811 displays the soundness judgment result for span A.

[0072] In the image display area 801, deformations related to the soundness determination are highlighted based on the information created by the deformation information analysis unit 203. For example, cracks 822 and 823 are cracks with a width of 0.5 mm or more, and since these are deformations related to the soundness determination, they are highlighted. On the other hand, crack 821 is a crack with a width of less than 0.5 mm and is not selected as a basis for the soundness determination, so it is not highlighted. In Figure 8, as an example of highlighting, cracks 822 and 823 are displayed with thicker polyline lines compared to crack 821, which was not selected as a basis for the soundness determination. The method of highlighting is not limited to this, and other display methods are also possible. For example, deformations may be displayed in a conspicuous color such as red, or they may be displayed in a flashing manner. As another example of highlighting, deformations that formed the basis for the soundness determination may be circled, or they may be enclosed in a predetermined object.

[0073] Similarly, the water leak 824 is also displayed with a thicker outline as it is a deformation related to the health assessment. On the other hand, the efflorescence 825 is not highlighted as it was not selected as a basis for the health assessment. Alternatively, deformations related to the health assessment may be extracted and further displayed in a separate display area that makes it possible to identify that they were the basis for the assessment.

[0074] Furthermore, if the deformation related to the soundness assessment is a combination of multiple deformations or the degree of damage to a specific part of the structural member of the structure, that area may be highlighted. For example, suppose span B in Figure 8 is judged to have a soundness of 2, and the basis for this judgment is a closed crack. In Figure 8, the area 826 surrounding the closed crack is highlighted with a thick line. Also, although not shown in Figure 8, if, for example, the degree of damage to a rock bolt was related to the soundness assessment, the rock bolt area may be highlighted.

[0075] Furthermore, information on the basis for the determination may be displayed for the deformations related to the health determination in the image display area 801. For example, the callout 830 indicates that the crack 822 is related to the health determination because it satisfies the determination criterion "Are there cracks with a width of 0.5 mm or more?". The display position of the callout 830 can be determined based on the deformation position information acquired by the deformation information analysis unit 203.

[0076] Furthermore, Span C has a health rating of 5, and a health rating of 5 does not include any unhealthy deformations, meaning there are no deformations related to the health rating. Therefore, the evaluation area of ​​Span C does not highlight any deformations related to the health rating.

[0077] Next, the display content of the importance display area 802 will be explained. In this embodiment, the importance of each type of deformation is displayed for each evaluation area. The importance display area 802 in Figure 8 displays the importance of each type of deformation for the soundness determination of span A. The user can switch between evaluation areas for which importance is viewed by operating the pull-down menu 804 or by other means. As mentioned above, the importance is calculated by the deformation information analysis unit 203, and in Figure 8, this is shown as an importance bar graph 803. This display shows that the types of deformation involved in the soundness determination of span A are cracks and water leakage, and that cracks, in particular, are strongly related to the determination.

[0078] Furthermore, the importance information may be used to control the degree of emphasis in the highlighting of deformations in the image display area 801. For example, the thickness of the display lines may be increased according to the importance of the deformation, so that more important deformations are highlighted. Specifically, in the example in Figure 8, since cracks are more important than water leaks, the thickness of the lines for cracks 822 and 823 is increased compared to the outline of water leak 824 to give them relative emphasis.

[0079] As shown in Figure 8, displaying the health assessment result and information on the defects related to the health assessment makes it easier for the user to understand the defects related to the health assessment. This makes it easier for the user to confirm the health assessment automatically determined by the information processing device 100. In particular, by not only showing the basis (criteria) for the assessment, but also showing the location and importance of the defects related to the health assessment, it becomes easier to understand the relationship between the assessment result and the defects.

[0080] Next, another example of how to display deformations related to soundness assessment will be explained using Figure 9. In Figure 9, the deformations that formed the basis for the soundness assessment, based on the information created by the deformation information analysis unit 203, are displayed in the form of an inspection report. Inspection reports generally record images of important deformations and their information. Deformities related to soundness assessment are important deformations from an inspection perspective, and by automatically organizing them in the form of an inspection report, the user's document creation work can be simplified.

[0081] In the display window 900 of Figure 9, the soundness of span A is 3, and information about the deformations related to this soundness determination is displayed. Figure 9 displays images of two deformations, damage numbers 1 and 2, and information about the deformations. In addition, Figure 9 has a page-turning function 901, and multiple deformations are organized in the form of inspection reports. Here, when arranging deformations in order, such as damage numbers 1 and 2, the display order may be determined based on importance. For example, deformations that are more important to the soundness determination may be displayed first, and the deformations may be arranged in order of importance.

[0082] Images of deformation regions, such as Image 902, which serve as the basis for determining structural integrity, are created by extracting them from an image superimposed on the drawing, based on the deformation location information acquired by the deformation information analysis unit 203. Specifically, the rectangular coordinates of the area surrounding the deformation are determined from the deformation location information in the drawing coordinate system. Since the image is displayed superimposed on the drawing, it has a coordinate system that is common to or convertible to the drawing. Therefore, by extracting the image range indicated by these rectangular coordinates, an image like Image 902 can be created. Specifically, Image 902 is an image extracted from the surrounding image of crack 823 based on the coordinates of crack 823 in Figure 8. Similarly, Image 903 is an image extracted from the surrounding image of crack 822.

[0083] Furthermore, the deformation information display area 904 displays the coordinate information of the deformation and the name of the type of deformation. In addition, the basis for the judgment may be converted into general sentences and displayed as observation information. For example, the deformation information display area 904 displays the sentence "A crack with a width of 0.5 mm or more exists" as observation information created from the information on the basis for the judgment. Moreover, the deformation information display area 904 may also be equipped with a comment field as a function for creating inspection reports.

[0084] As described above, Figure 9 illustrates an embodiment in which defects related to the assessment of soundness are displayed in the form of an inspection report. In the example shown in Figure 9, a function may also be provided to output the inspection report data in any document file format.

[0085] The above explains how to display the deformations related to the determination of structural integrity. Users can review these results and adjust the structural integrity determination if the automatic determination differs from their own judgment. One way to adjust the results is to modify the deformation information itself. Specifically, for example, if it is determined that the record of a crack is incorrect, the crack information can be deleted or, conversely, added. By correcting the deformation record in this way and then re-evaluating the structural integrity, the result of the structural integrity determination can be adjusted.

[0086] Another method for adjusting the health assessment result is to change the criteria for associating deformations with the evaluation area. For example, in the above embodiment, the health assessment was determined by associating deformations with the evaluation area using an association criterion that uses only deformations within the evaluation area range. If the association criterion for the evaluation area is changed, for example, "deformations located at the boundary of the evaluation area are assessed by associating the entire deformation with the evaluation area," the health assessment result may change.

[0087] For example, regarding the crack 403 in Figure 4, the above embodiment described an embodiment in which only the portion of the crack 421, which is within the range of span A, is used for determining the soundness of span A. On the other hand, if the entire crack 403 is used for determining the soundness of span A, the total length of the crack used for determining the soundness of span A becomes longer, which may result in a judgment of worse soundness compared to the soundness determination result in the above embodiment. Thus, the result of the soundness determination changes depending on how the deformation is related to the evaluation area of ​​the soundness determination, so the method of relating the deformation to the evaluation area is one method of adjusting the soundness determination result.

[0088] The following section will explain in detail how to adjust the health level by changing the association criteria of the evaluation area management unit 201 via the setting unit 205.

[0089] (Processing related to the setting section) The setting unit 205 sets the association criteria by displaying information on the display unit 105 (described below) and by receiving user input from the operation unit 106.

[0090] The following describes an embodiment in which, after determining the health level using the process described above, the criteria for associating the deformation with the evaluation area are changed to adjust the health level determination. Figure 10 shows the display unit 105 when adjusting the health level by changing the association criteria. Figure 10 shows the display window 1000 in the state where the association criteria for span A are being changed, and the evaluation area for which the association criteria are to be changed can be changed, for example, by the pull-down menu 1001.

[0091] In the image display area 1002 of the display window 1000, similar to the image display area 801 in Figure 8, an image of the tunnel wall surface superimposed on the tunnel drawing and deformation information are displayed. In addition, the soundness display area 1004 displays the processing result of the soundness determination according to the embodiment described above as "Soundness 3". Furthermore, the image display area 1002 shows the association of deformations with evaluation area span A. In the above embodiment, deformation information was associated with the evaluation area according to the initial association criterion. This initial association criterion was "Use deformations within the evaluation area for soundness determination". Therefore, in the image display area 801, deformations that exist within span A, such as cracks 1020 and water leaks 1021, are highlighted as being associated with span A. Specifically, in Figure 10, the lines and contours of deformations that exist within span A are displayed as solid and thick lines. On the other hand, deformations that are not associated with span A (deformations in spans B and C) are displayed as dotted lines.

[0092] Furthermore, although crack 1010 is a deformation that spans from span A to C, according to this association criterion, the crack portion 1011 within the range of span A is associated with span A. Therefore, only portion 1011 of crack 1010 is highlighted. In this way, the image display area 1002 in Figure 10 highlights the deformation associated with a predetermined evaluation area. In this embodiment, the deformation associated according to the set association criterion is displayed in a way that is identifiable to the user. This makes it easier for the user to understand which deformations are associated with an evaluation area in order to determine the soundness of a certain evaluation area.

[0093] In the display window 1000 of Figure 10, the association of deformations with the evaluation area can be confirmed by pressing the setting button 1006. Figure 11(A) shows the association criterion setting screen that appears when the setting button 1006 is pressed. On this screen, the method of associating deformations existing at the evaluation area boundary with the evaluation area is displayed for each type of deformation, such as cracks and efflorescence. In this embodiment, three types of association criteria, association criteria 1 to 3, are prepared for deformations existing at the evaluation area boundary, and one of the association criteria is set for each type of deformation. Association criterion 1 (pattern 1) is "For deformations that straddle the boundary of the evaluation area, the entire deformation is used for determining the soundness." Association criterion 2 (pattern 2) is "For deformations that straddle the boundary of the evaluation area, only the deformation within the range of the evaluation area is used for determining the soundness." Note that pattern 2 is the association criterion that is set as the initial setting in the above explanation. Association criterion 3 (pattern 3) is "For deformations that straddle the boundary of the evaluation area, they are excluded from the deformations used for determining the soundness of the evaluation area." Figure 11(A) shows the settings for the association patterns of the deformations at the evaluation area boundaries for each type of deformation. In Figure 11(A), the initial association criterion is set as association pattern 2 for all types of deformations.

[0094] Next, the image display area 1003 and other elements located at the bottom of the display window 1000 in Figure 10 will be explained. These are displays for adjusting the setting of the association criteria for deformations to the evaluation area. By using these displays and the operations described below, the user can set different association criteria from the association criteria at the top of the display window 1000 (hereinafter referred to as setting 1), re-evaluate the health level, and adjust the health level evaluation result.

[0095] The following describes the procedure for adjusting the health assessment by changing the association criteria. First, when the user presses the setting button 1007, the association criteria setting screen shown in Figure 11(B) is displayed. This screen is the same as the one shown in Figure 11(A). The user can select any association criteria for each type of deformation by clicking the radio buttons. Here, we show an example where only cracks are changed to association pattern 1 from setting 1 in Figure 11(A). Hereafter, this setting will be referred to as setting 2. The image display area 1003 in Figure 10 shows the deformation associated with span A by this setting 2. Unlike the image display area 1002 of setting 1, in the image display area 1003, the entire crack 1012 that exists at the boundary of the evaluation area (span A) is associated with the evaluation area, and the entire crack 1012 is highlighted.

[0096] In step S301 of Figure 3, the association criteria can be set based on user instructions. For example, once the association criteria for setting 2 are set, the user presses the judgment button 1008 to determine the health level based on these criteria. This executes the processes from steps S302 to S306 in Figure 3, and the health level of span A is determined (in step S302, span A is selected as the evaluation area). This determination result is displayed in the health level display unit 1005. In Figure 10, the health level determination results before and after changing the association criteria are displayed in the health level display units 1004 and 1005, respectively. By displaying the health level before and after the change on the same screen, the user can compare the results and decide which health level determination result to adopt. In the above, an embodiment was described in which the health level determination process is executed when the user presses the judgment button 1008, but the execution of the health level determination process may be set to be performed automatically when the association criteria are set.

[0097] The method described above allows you to adjust the health assessment result by changing the criteria for associating deformations with the evaluation area. While the above explanation described an example where only the association criteria for span A were changed, it is also possible to change the association criteria for all evaluation areas at once and re-execute the health assessment process.

[0098] In the above-described embodiment, the evaluation area was defined as the tunnel span, but the evaluation area is not limited to this. Any range that is a predetermined range for determining the soundness of the tunnel may be used as the evaluation area.

[0099] Figure 12 illustrates another example of an evaluation area, specifically a case where a portion of the bridge deck is used as the evaluation area. Similar to Figure 10, Figure 12 shows the display window 1200 for changing the association criteria. Image display areas 1201, 1202, and 1203 show wall images superimposed on drawings of the bridge deck 1210 and piers 1211, along with deformation information. The frames 1221, 1222, and 1223 shown on the deck indicate the evaluation areas for structural integrity. This example illustrates how, in determining the structural integrity of the deck, the evaluation area is set to a region of the deck near the piers, rather than the entire deck.

[0100] In Figure 12, image display areas 1201 and 1202 show the association status of deformations with the evaluation area when the association criteria are changed for each type of deformation. Specifically, image display area 1201 shows the state in which only deformations within the evaluation area are associated, while image display area 1202 shows the state in which deformations that exist across the boundary of the evaluation area are associated with the entire deformation area. Similar to Figure 10, these are highlighted so that the range of deformations associated with the evaluation area can be easily understood.

[0101] Furthermore, as a variation of the method for setting the criteria for associating deformation with the evaluation area, the range of the evaluation area may be changed. The evaluation area 1223 of the image evaluation area 1203 has a wider range set as the evaluation area compared to evaluation areas 1221 and 1222. The setting unit 205 may thus allow the evaluation range of the bridge piers of the deck to be arbitrarily set based on user instructions.

[0102] The following describes a modified version of Embodiment 1.

[0103] First, Embodiment 1 describes an embodiment in which a health assessment method is learned using a machine learning algorithm, and a decision tree is described as an example. The machine learning algorithm used in Embodiment 1 is not limited to a decision tree; other methods may be used. In particular, an embodiment using a Randomized Tree, which is composed of a tree structure with binary judgments similar to a decision tree, is a suitable modification.

[0104] Furthermore, as shown in Figure 8, when highlighting the health assessment results for multiple evaluation areas and the deformations associated with those assessments, the method of displaying the deformations may be changed depending on the health assessment value. For example, first, a predetermined color may be set for each health assessment level from 1 to 5. Then, the color of the deformations related to the health assessment to be highlighted in each evaluation area may be displayed using the color corresponding to the health assessment level of that evaluation area. This makes it easier to simultaneously check the health assessment results for multiple evaluation areas and the deformations associated with each health assessment. Also, for a similar purpose, the degree of emphasis on the deformations associated with the health assessment may be changed depending on the health assessment level. For example, the deformations associated with the health assessment of an evaluation area judged to have a health assessment level of 1 may be highlighted more than the deformations associated with the health assessment of an evaluation area judged to have a health assessment level of 4. To achieve this, the lines of the deformations and area contours associated with the health assessment level of 1 may be displayed thicker or in a more prominent color.

[0105] Furthermore, while the above embodiment describes an embodiment for determining the soundness of each evaluation area, there are also cases where the representative soundness of the entire structure is viewed. In such cases, the worst soundness (the soundness closest to 1) among the soundness of each evaluation area of ​​the structure is used as the representative soundness of the entire structure. Therefore, when displaying the representative soundness of the entire structure, the deformations related to the soundness determination result should be displayed for the evaluation area for which the soundness of the representative soundness has been determined, as described above. Note that the overall soundness of the structure may be configured to use the average value of the soundness of multiple evaluation areas. In this case, the deformations related to the soundness determination result should be displayed for the evaluation area for which a soundness close to the average value has been determined among the soundness of each evaluation area of ​​the structure.

[0106] <Embodiment 2> Embodiment 1 describes an embodiment in which a judgment model trained using machine learning is used as the method for determining health. The method for determining health in this embodiment is not limited to this, and other methods may be used. Embodiment 2 describes a method in which a predetermined rule is used as the judgment criterion for determining health. Note that the hardware configuration and functional configuration of Embodiment 2 are the same as those of Embodiment 1, so the description will be omitted.

[0107] In Embodiment 2, a human designer pre-designs and sets the rules for determining each health level. The information processing device 100 (specifically, the health level determination unit 202) automatically determines the health level by using these determination rules. Each of the determination rules corresponds to the determination criteria in Embodiment 1. The health level determination process of the health level determination unit 202 in Embodiment 2 will be described below.

[0108] Figure 13 shows examples of judgment rules for soundness levels 2 and 3, and examples of soundness judgment results for a given evaluation area. First, each judgment rule performs a binary judgment based on the deformation information associated with the evaluation area and the structural specifications. For example, rule R2-1 for judging soundness level 2 in Figure 13 is "Are there cracks with a width of 1.0 mm or more?". According to this rule, if there are cracks with a width of 1.0 mm or more in the evaluation area, the soundness level of that evaluation area will be 2. In this way, if either the deformation and specifications of the evaluation area satisfy the judgment rule for soundness level 2, the soundness level will be 2. Furthermore, in the soundness judgment method using judgment rules, the lowest soundness level for which the judgment rule is true becomes the soundness judgment result. For example, in Figure 13, the judgment result based on the deformation information and specifications of a given evaluation area is shown in judgment item 1301. In this example, it is assumed that all the judgment rules for soundness level 1 were false.

[0109] As shown in Figure 13, for the health level 2 determination rule, all deformation information and specification information of the evaluation area are False. On the other hand, for the health level 3 determination rule, two of the rules are True. As a result, regardless of the results of the health level 4 and 5 determination rules, the health level of the evaluation area in question is determined to be 3.

[0110] The health determination unit 202 determines the health level as described above and, similar to Embodiment 1, outputs the basis for determining the health level. The basis for determination in Embodiment 2 is the determination rule (determination criterion) that was True in the health level determination. In the example in Figure 13, the basis for determining a health level of 3 is the determination rules R3-1 and R3-3.

[0111] The deformation information analysis unit 203 identifies deformation information related to the soundness determination based on the judgment basis output by the soundness determination unit 202. For example, based on the judgment rule R3-1 in the example in Figure 13, it is assumed that cracks with a width of 0.5 mm or more in the evaluation area are related to the soundness determination. Then, the position information of cracks with a width of 0.5 mm or more in the evaluation area is obtained from the deformation information storage unit 206. Similarly, based on the judgment rule R3-3, the position information of grid-like cracks (grid spacing 50 cm) is obtained from the deformation information storage unit 206. The display information processing unit 204 uses this deformation position information to highlight deformations related to the soundness determination and crop images of the deformation range, as described in Embodiment 1.

[0112] Furthermore, in a health determination method that uses judgment rules, the importance of the type of deformation related to the health determination may also be calculated. In Embodiment 1, importance was calculated based on the information gain obtained during the learning of the decision tree, but since there is no information gain in the judgment rules, the importance will be calculated by a different method. As a method for calculating importance in Embodiment 2, for example, the number of times the judgment rule becomes True for each type of deformation can be counted. For example, in the example in Figure 13, assuming that the judgment rule related to cracks became True for both judgment rules R3-1 and R3-3, the counter for cracks is set to 2. As a result, the importance of cracks for the health determination result becomes 2. This process is performed for each type of deformation to obtain the importance for each type of deformation. In this way, the display information processing unit 204 can also display the importance for each type of deformation, similar to Embodiment 1. Furthermore, the judgment basis output by the health determination unit 202 may be displayed in the same way as in Embodiment 1.

[0113] As described above, even when a judgment rule is used as the method for determining the health level, it becomes possible to display information about the abnormalities related to the health level determination, and the user can understand the basis for the health level determination of the information processing device 100.

[0114] <Embodiment 3> Embodiment 2 describes an embodiment in which the degree of structural integrity is determined based on judgment rules created by humans. Creating judgment rules by humans has the advantage of making the criteria for determining structural integrity easier to understand and allowing the criteria to be controlled based on human knowledge. On the other hand, when it is necessary to prepare complex judgment criteria in order to correctly determine the degree of structural integrity, there is a problem in that it is difficult for humans to manually design all the judgment criteria. For example, when setting judgment criteria that include not only the presence and size of deformations such as cracks, but also information such as the number of years the structure has been in service and its geographical location, the number of combinations of conditions becomes enormous, making it difficult for humans to determine the judgment criteria while considering all conditions. Therefore, it is sometimes preferable to statistically learn the judgment criteria for determining structural integrity through machine learning. Learning the judgment criteria using a decision tree in Embodiment 1 is one example of a method to solve this problem.

[0115] This embodiment describes an embodiment in which a judgment rule is learned by frequent pattern mining. Learning a judgment rule by frequent pattern mining is one form of learning a binary judgment criterion that can be understood by humans. This is one embodiment in which a judgment criterion is acquired by learning, and is a modified embodiment of Embodiment 1. Note that the hardware configuration and functional configuration of Embodiment 3 are the same as those of Embodiment 1, so the explanation will be omitted.

[0116] Frequent pattern mining extracts items or combinations of items that frequently occur under predetermined conditions. Below, an embodiment of applying this technique to learning judgment rules for health assessment will be explained using Figure 14. Figure 14 is a diagram showing the overall picture of judgment rule learning using frequent pattern mining.

[0117] First, in order to explain the learning of the judgment rules in this embodiment within the context of frequent pattern mining, we will explain items and sets of items. In this embodiment, an item is a single judgment criterion for determining the degree of soundness. Table 1401 in Figure 14 shows an example of an item definition. For example, item I1 is a judgment criterion, "Are there cracks in the evaluation area?". Thus, in this embodiment, a single judgment criterion becomes a single item. Note that items of judgment criteria may include thresholds for attribute information such as the size of the deformation. For example, item I2 in Figure 14 is "Are there cracks with a width of 0.5 mm or more?", and is a judgment criterion composed of deformation information and attribute information thresholds. Similarly, judgment criteria related to specification information are also items. For example, item IM is defined as an item with the judgment criterion, "Has the elapsed time since the start of service been 50 years or more?". Hereinafter, the entire set of items is I all = {I1, I2, ...IM}, I all A subset of an item is called an item set. An item set can consist of a single item or a combination of items. In frequent pattern mining, the entire set of items is called the item set. all The objective is to extract a set of items that can be used for health assessment.

[0118] In the item definitions above, we have shown an example of defining items using thresholds for deformation size, but items may also be defined according to discretized numerical values. Specifically, for example, the classification of crack width may be discretized, and items such as "Are there cracks with a width of 0.2 mm to 0.5 mm?" and "Are there cracks with a width of 0.5 mm to 1.0 mm?" may be defined.

[0119] Next, in order to perform frequent pattern mining, transaction data is created from the training data D. The training data D is defined as follows, D = {D i} and D i This is data relating to a specific evaluation area of ​​a certain structure. Also, D i It is defined as follows:

[0120]

number

[0121] X i This consists of deformation information associated with a certain evaluation area and specification information for a certain structure. i is X i Based on this, the correct data for health status determined by a human is shown, and in this embodiment, the health status is represented by a range of 1 to 5.

[0122] This single training data D i From one transaction T i Create. Transaction T will be created. i For example, T i ={I1, I2, I9, I20;y i It will look like this. In this case, data D i X i This indicates that (the deformation information and specification information of the evaluation area) matches the conditions of item definitions I1, I2, I9, and I20. Also, as another specific example, X i If a crack with a width of 0.5 mm is present, items I1 and I2 meet the condition. The condition for the other items is X i If it does not match, T i ={I1, I2;y i}

[0123] Thus, each D in the training data D i Regarding transaction data T, i Create transaction data T. i The set of items included is I i Therefore, transaction data T i is, T i ={I i ;y i It is represented as}. Also, let T be the dataset of all transaction data. Here, T = {T i} In the following, we will perform frequent pattern mining using this transaction dataset T. In frequent pattern mining, we extract sets of items that frequently appear in the transaction dataset T and use these as judgment rules. Furthermore, the frequent pattern extraction process is performed for each health level, and judgment rules for each health level are learned. In the following, we will explain the frequent pattern extraction process using the learning of judgment rules for health level 3 as an example.

[0124] In this embodiment, Apriori (Non-Patent Literature 1) is used for extracting frequently occurring patterns. In Apriori, the frequency of occurrence of a certain set of items in a transaction dataset is called support. Support is expressed by the following formula.

[0125]

number

[0126] N is the number of data points in the transaction dataset T, σ T (I) is transaction data T containing item set I. i This is the number of items. In Apriori, a threshold called minimum support is given in advance, and sets of items that show support equal to or greater than this minimum support are extracted as frequent patterns. Since Apriori is a well-known algorithm, a detailed explanation will not be given, but by specifying the minimum support in advance when performing a search for frequent item sets, it is possible to efficiently find sets of items that show support equal to or greater than this minimum support.

[0127] Furthermore, in learning the rules for determining health level 3, we focus only on transaction data with health level 3. Transaction data with health level 3, i.e., y i A transaction dataset consisting only of transaction data with a value of 3 is T (3) Therefore, this support is,

[0128]

number

[0129] This is the result. N (3) is, T (3) This is the number of data points. By running Apriori with this support, it is possible to extract a set of events that frequently occur in transaction datasets with a health level of 3. This set of events is I (3) This is represented as {Ia, Ib..., In}, where Ia and Ib are event sets such as Ia={I1, I4, I10} and Ib={I8}.

[0130] Next, we define the set of events specific to transactions with a health level of 3. (3) Select from: Event Set I (3) This represents a set of events that frequently occur in training data with a health level of 3, but it may also include event sets that frequently occur in training data with other health levels. In other words, it may include items or combinations of items that are always likely to appear, regardless of the health level. Therefore, we will use training datasets other than health level 3 to select event sets that are specific to the training dataset with a health level of 3. Specifically, I (3) Transaction dataset T for each event set (3) We calculate the ratio of support (frequency of occurrence) in this dataset to the support in transaction datasets other than health level 3, and identify event sets with a large ratio as event sets specific to health level 3. This support ratio is called the Growth Rate and is expressed by the following formula.

[0131]

number

[0132] Here, a transaction dataset created from training datasets other than those with a health score of 3 is used. (not3) Let I be the set of events for which the Growth Rate is calculated. (3)This is the set of events included in the dataset. Hereafter, Growth Rate will be referred to as GR in this text. Patterns that appear predominantly in one dataset like this are called Emerging Patterns (Non-Patent Literature 2).

[0133] In the example above, a higher GR indicates that the item set is specific to a health level of 3. Therefore, an item set with a GR above a certain threshold is called Item Set I. (3) By selecting from these options, you can create a set of items specific to health level 3. This set of items specific to health level 3 is called R. (3) The values ​​{R3-1, R3-2, R-3N} can be used as judgment rules to determine a health level of 3. Specifically, as shown in Table 1402 in Figure 14, for example, R3-1 = {I2}, and item I2, "Are there cracks with a width of 0.5 mm or more?", is learned as one of the judgment rules. In addition, judgment rules for combinations of multiple items are also learned, for example, R3-2 = {I5, I10}. Note that the selection of event patterns based on GR may be done by selecting the top n GR event patterns instead of thresholding.

[0134] As described above, a judgment rule for health level 3 can be learned through frequent pattern mining. The same process is performed for each health level to learn judgment rules for health levels 1 to 5. The process of determining the health level of the target evaluation area using the learned judgment rules, and the subsequent process of displaying information on deformations related to the health level determination, can be performed in the same manner as in Embodiment 2. That is, a judgment rule is determined in which the input deformation information and specification information become True, and this judgment rule is used as the basis for the judgment. Then, based on the deformation information included in the basis for the judgment, displays such as highlighting deformations are performed.

[0135] Alternatively, instead of learning the rule for determining a health level of 5 (the healthiest), the system may automatically determine a health level of 5 if it is not determined to be between 1 and 4.

[0136] Furthermore, GR may be used to calculate the importance of each type of abnormality. In this case, during training, the GR of each decision rule, for example R3-1, R3-2, etc., is saved together with the decision rule. A higher GR indicates a greater degree to which the set of events is specific to that level of health. Therefore, by using GR in the same way as the information gain of each node in the decision tree in Embodiment 1, it becomes possible to calculate the importance of each type of abnormality to the health determination result.

[0137] Furthermore, the learned judgment rules may be displayed to the user, allowing the user to adjust the judgment rules. In this case, first, the judgment rules learned using the method described above are displayed in a way that the user can understand the content of the item. Specifically, the items of the item set in Table 1402 in Figure 14 are displayed as the content of the specific item. For example, the content of judgment rule R3-1 is displayed as the judgment criterion "Are there cracks with a width of 0.5 mm or more?". This allows the user to see what judgment rules have been learned through frequent pattern mining. Based on this display, the user can adjust the judgment rules. Specifically, they can remove judgment rules that are judged to be inappropriate, or add judgment rules that have not been learned but are judged to be worth adding. As a result, it becomes possible to create judgment rules based on human knowledge and experience, in addition to the judgment rules acquired mechanically.

[0138] Furthermore, generally, preparing training data requires time for data accumulation. Therefore, in an operation where health assessment is performed by an information processing device, initially, as described in Embodiment 2, health assessment is performed based on judgment rules created by a human. The input data at this time (deformation information and specification information of the evaluation area) is accumulated, and once a predetermined amount of data has been accumulated, this data may be used as training data to learn the judgment rules. The labels for health assessments 1 to 5 to be learned are data confirmed by a human regarding the health assessment results of the information processing device, or the results of a human correcting the health assessment results of the information processing device. This makes it possible to operate automatic health assessment by the information processing device from the initial stage when no training data exists. In addition, the learning of judgment rules accompanying data accumulation may be performed not only once, but also in stages multiple times depending on the accumulation status of training data. Note that the embodiment in which training data is accumulated and training is performed when a predetermined amount of training data has been accumulated can also be implemented in the training of the health assessment method using the machine learning algorithm of Embodiment 1.

[0139] The above description explains the two-step process of learning judgment rules: frequent pattern extraction using Apriori and selection of item sets specific to each health level by calculating GR. However, the judgment rule learning method in this embodiment is not limited to this. In other words, any method that extracts a specific item set using training data may be used. For example, in addition to Apriori, a method called FP-Growth is also well known as a method for extracting frequent patterns. Alternatively, frequent pattern extraction may be performed using other computationally efficient algorithms.

[0140] Furthermore, the above describes how the learned judgment rules are used in the same manner as in Embodiment 2. That is, if the input data matches any of the health judgment rules, the health level is obtained as the health judgment result. However, the judgment rules obtained through the above learning process have different confidence levels, as indicated by GR. Therefore, it is also possible to calculate a score from the judgment rules using the support and GR, and determine the health level based on the score. As explained above, the judgment rules are a set of items that appear in a transaction dataset of a specific health level, and these are called Emerging Patterns. An ensemble identification method using these Emerging Patterns is described in Non-Patent Literature 3. In the following embodiment, an embodiment in which the health level is determined from the learned judgment rules using this method will be described. In order to implement this method, the support and GR of each judgment rule are recorded when the judgment rules are learned.

[0141] First, there is a health level H with a judgment score S. H Based on Non-Patent Document 3, this can be expressed as follows:

[0142]

number

[0143] Here, IN is an item included in the deformation information and specification information of the evaluation area subject to health assessment, I (H) This shows the rules for determining health level H. In other words, in the rules for determining health level H, the sum of the values ​​calculated based on GT and support in the input data of the deformation information and specification information of the evaluation area is the score S for determining health level H. H This is the result. According to this formula, the higher the GR and support values, the more the score S H A high value can be obtained. This score S H This is calculated for each health level, and the highest score is S H The resulting health score is defined as the health score H of the evaluation domain.

[0144]

number

[0145] In Embodiment 3, a method for learning health determination rules using frequent pattern mining was described. By learning determination rules using frequent pattern mining, it is possible to create determination rules that are more complex and have higher determination performance than those set by humans. Furthermore, since the learned result is a determination rule that can perform binary determination, it is easy for humans to understand the learned result and it is also possible to adjust the determination rule. In addition, by using the learned determination rule for health determination in the same way as in Embodiment 2, information on abnormalities related to the health determination can be displayed, making it possible to display the basis of the health determination result in an easy-to-understand manner.

[0146] <Embodiment 4> In the embodiments described above, the setting of association criteria for deformations to the evaluation domain was described in an embodiment in which the criteria were set for each type of deformation. The setting of association criteria may be done not only for each type of deformation, but also for each individual deformation. Embodiment 4 describes an embodiment in which association criteria for each individual deformation are set. Note that the hardware configuration and functional configuration of Embodiment 4 are the same as those of Embodiment 1, so their description is omitted.

[0147] Figure 15 shows the display window 1500 displayed on the display unit 105 by the setting unit 205 for setting associations to evaluation areas for each individual deformation. In Figure 15, a soundness evaluation area is set for each span of the tunnel, and the setting criteria for associating deformations with the evaluation area of ​​span E are shown. Four cracks, with crack IDs 001 to 004, are displayed on the image information display unit 1501. Here, cracks ID 003 and ID 004 are located in positions unrelated to span E, and are therefore not associated with span E and are displayed with dotted lines.

[0148] The user selects a deformation to adjust settings for in order to set the association criteria for individual deformations. For example, the mouse cursor 1502 is used to select the deformation whose settings you want to adjust. Figure 15 shows the state where crack ID 002 is selected. The dotted line 1504 indicates that crack ID 002 is in focus. When crack ID 002 is selected, information for selecting the association criteria for crack ID 002 to span E is displayed in the criteria setting area 1503.

[0149] In Figure 15, as with Embodiment 1, three types of association criteria, association patterns 1 to 3, are prepared for deformations located at the boundary of the evaluation area, as criteria for associating deformations with the evaluation area. Association pattern 1 is a pattern in which the entire deformation located at the boundary of the evaluation area is used for determining the health of the deformation. Association pattern 2 is a pattern in which only the part of the deformation within the evaluation area is used for determining the health of the deformation located at the boundary of the evaluation area. Association pattern 3 is a pattern in which deformations located at the boundary of the evaluation area are excluded from the deformations used for determining the health of the evaluation area.

[0150] Figure 15 shows a situation where the user has set association pattern 1 as the association criterion for crack ID 002. With this setting, the entire crack ID 002 is displayed with a thick line on the image information display unit 1501, indicating that it is associated with span E. On the other hand, although the setting situation for crack ID 001 is not shown, it is assumed that association pattern 2 is set as the association criterion. As a result, crack ID 001 is displayed with a thick line to indicate that the range of span E is associated with span E, and the part outside the range of span E is displayed with a dotted line to indicate that it is not associated with span E. Thus, in Embodiment 4, different association criteria can be set for crack ID 001 and ID 002, which are of the same deformation type.

[0151] Furthermore, the screen for setting association criteria for individual deformations may also display how the soundness changes depending on the association settings. In the criterion setting area 1503 of Figure 15, the soundness of span E is determined and the results are displayed when each association pattern is set for crack ID 002. In this way, it becomes possible to set the association criteria for deformations in the evaluation area while checking how the soundness changes when the association criteria for individual deformations are changed.

[0152] As described above, Embodiment 4 describes an embodiment in which association criteria for each individual deformation are set to the evaluation area. This makes it possible to set the association of deformations to the evaluation area in detail, and the user can adjust the health determination result in detail.

[0153] <Embodiment 5> In the embodiments described above, we explained embodiments in which deformations located within or at the boundary of the evaluation area are associated with the evaluation area to determine the health of the evaluation area. In this embodiment, the association of deformations with the evaluation area may target deformations that do not directly overlap with the evaluation area. Embodiment 5 describes an embodiment in which deformations not directly related to the evaluation area are associated as deformations used to determine the health of the evaluation area. That is, in Embodiment 5, an association rule is set to associate deformations located in a related area that is a different area from the evaluation area but is relevant in determining the health of the evaluation area, with the deformations used to determine the health of the evaluation area. Note that the hardware configuration and functional configuration of Embodiment 5 are the same as those of Embodiment 1, so the explanation is omitted.

[0154] Figure 16 shows a drawing and image of a road bridge that is the subject of structural integrity assessment. The road bridge in Figure 16 shows the deck slab 1602 in the span between the bridge piers 1603. The deck slab 1602 is the part viewed from the ground, and Figure 16 also shows a drawing and image of the road surface 1601. Here, in order to perform structural integrity assessment of the deck slab 1602, a process is performed to associate the deformations used for structural integrity assessment of the deck slab 1602. The frame line 1604 surrounding the deck slab 1602 indicates that the entire deck slab 1602 is the evaluation area to be processed.

[0155] The example shown in Figure 16 illustrates an embodiment in which deformations on the road surface 1601 on the back side of the slab are included in the deformations used to determine the soundness of the slab 1602. Specifically, potholes 1611 are associated with the evaluation area 1604 and used to determine the soundness of the evaluation area 1604 of the slab 1602. To achieve this, the association criterion for deformations in the evaluation area 1604 in Figure 16 is set to "associate potholes on the road surface on the back side with the evaluation area." Since areas where the road surface is severely damaged, such as potholes, can allow water such as rain to affect the slab, it is preferable to include these in determining the soundness of the slab. A pothole is a round hole or depression that forms when the surface layer of asphalt road peels off. In this way, deformations that are not directly included in the evaluation area may also be associated with the evaluation area to determine their soundness. This association criterion can be changed for each type of deformation or for each individual deformation, similar to the methods described in Embodiments 1 and 4. For example, in Figure 16, the cracks 1612 in the road surface 1601 are not associated with the evaluation area 1604 of the deck slab 1602, and are not used in determining the soundness of the evaluation area 1604. The user can configure whether or not this association is present via the operation unit 105 as needed.

[0156] Furthermore, when associating deformations that are not directly related to the evaluation area, the target area should be predetermined to be structurally or inspectionally related to the evaluation area, as shown in Figure 16 for the floor slab and road surface. The association criteria for whether or not to associate deformations in these related areas with the evaluation area should then be adjustable. Alternatively, it may be possible to set up the system to individually associate arbitrary deformations at arbitrary locations with the evaluation area.

[0157] As described above in Embodiment 5, deformations located in positions not directly related to the evaluation area can also be used for determining the structural integrity. This makes it possible to perform structural integrity determination that takes into account deformations outside the evaluation area that affect the structural integrity of the evaluation area. Furthermore, by adjusting the association criteria, the results of the structural integrity determination can be adjusted.

[0158] <Embodiment 6> In the embodiments described above, embodiments of performing various processes using the information processing device 100 in Figure 1 were explained. Embodiment 6 describes a processing configuration in which services to users are performed using a server-client system configuration.

[0159] Figure 17(A) is a diagram illustrating the configuration of the information processing device 100 (information processing system) according to this embodiment. It is a diagram illustrating one form of processing in a server-client configuration. In Figure 17(A), the client terminal 1711 and the service provider system 1731 are information processing devices having the same hardware configuration as the information processing device 100 in Figure 1. In Figure 17(A), the client terminals 1701 to 1703, operated by users 1701 to 1703, are connected to the service provider system 1731 via the internet 1720.

[0160] In Figure 17(A), only the processing related to the display unit 105 and the operation unit 106 is executed on the client terminal, while other processing (for example, the processing of the health determination unit 202) is executed on the service provider system 1731. With this configuration, only light processing is executed on the client terminal, and computationally intensive processing such as health determination is executed on the high-performance service provider system 1731. As a result, users 1701 to 1703 can receive the service by simply preparing an inexpensive client terminal with low processing power.

[0161] Note that the distribution of processing is not limited to the configuration shown in Figure 17(A); other configurations are also possible. For example, Figure 17(B) shows a configuration that includes an on-premise server 1742, which is a different processing configuration from Figure 17(A). In Figure 17(B), the processing of this embodiment is distributed and executed by the client terminal 1741, the on-premise server 1742, and the service provider system 1743. For example, the service provider system 1743 performs learning of the health assessment method, creates a health assessment model or health assessment rule from the learning results, and sends it to the on-premise server 1742. The on-premise server performs processing such as health assessment processing, and the client terminal 1741 displays the health assessment results. In this way, even if the inspection images and health assessment results are confidential, the health assessment processing is performed on the on-premise server 1742 within the company, making it a secure configuration from an information security standpoint. On the other hand, it is possible to receive the latest services, such as the latest health assessment model, from the service provider system 1743.

[0162] Furthermore, the configuration of this embodiment allows for the provision of appropriate services to each user. For example, the method for determining structural integrity may differ depending on the user. For instance, users in this embodiment are assumed to be inspection companies, structural managers, and infrastructure management departments of local governments. These users may have their own inspection standards, and the method and standards for determining structural integrity may differ depending on the structure being managed. Therefore, by providing a structural integrity determination method (structural integrity determination model or determination rules) that is suitable for the user's needs and circumstances, it becomes possible to perform structural integrity determination appropriate for each user.

[0163] Similarly, the criteria for associating evaluation domains may also be tailored to provide appropriate criteria for each user. For example, the service provider system 1731 can store the previously set criteria of a user along with the user information. Then, when the user performs a health assessment for the first time, the previously set criteria are provided as the initial setting for the association criteria.

[0164] Furthermore, the association criteria may be designed to learn the user's association setting criteria. For example, as described in Embodiment 4, association criteria can be set for each individual deformation. In learning the association criteria, the user's setting tendencies are learned. Specifically, the deformation size, its position relative to the evaluation area, and the degree of overlap between the evaluation area and deformations at the evaluation area boundary are used as input variables, and the presence or absence of association with the evaluation area is used as a training label, and learning is performed using an arbitrary machine learning algorithm. The resulting learned model is then used to associate deformations with the evaluation area.

[0165] In this way, it is possible to provide services that are tailored to the health assessment tendencies of each user.

[0166] [Other embodiments] The present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., an ASIC) that implements one or more functions. [Explanation of symbols]

[0167] 201 Evaluation Area Management Department 202 Soundness judgment section 203 Deformation information analysis department 204 Display Information Processing Unit 205 Settings Section 206 Abnormal Information Storage Unit 207 Specifications Information Storage Unit 208 Learning Department

Claims

1. A first determination means for determining a first degree of soundness indicating the soundness of each of several evaluation regions in a structure, based on the deformation contained in the structure, A second determination means for determining a second soundness, which indicates the overall soundness of the structure, based on the first soundness in each of the plurality of evaluation areas, The system includes an output means that outputs information regarding the deformation of the structure that formed the basis for the determination of the second determination means, and information indicating the second degree of soundness. The output means outputs information such that images extracted from each of the multiple deformations that formed the basis for determining the second degree of health are displayed in an order corresponding to the degree of contribution of each of the multiple deformations to determining the second degree of health. An information processing device characterized by the following:

2. The first determination means determines the first degree of soundness based on determination criteria relating to the presence or absence of deformation, the type of deformation, or the size of the deformation. The information processing apparatus according to claim 1, characterized in that the output means outputs information relating to a deformation that satisfies the judgment criteria.

3. The information processing apparatus according to claim 2, characterized in that the output means outputs information that highlights deformations that satisfy the judgment criteria in an image of the structure, and information that indicates the judgment criteria.

4. The structure has a calculation means for calculating the importance of each deformation, which represents the degree to which it contributed to the determination by the first determination means. The information processing apparatus according to any one of claims 1 to 3, wherein the output means outputs information in which images extracted from each of the plurality of deformations are arranged in order of increasing importance of each of the plurality of deformations.

5. The information processing device according to claim 2 or 3, characterized in that the judgment criteria include criteria relating to the specifications of the structure.

6. The information processing device according to claim 5, characterized in that the specifications information includes at least one of the number of years since the start of service of the structure, geographical conditions, and usage environment.

7. The information processing apparatus according to any one of claims 1 to 6, characterized in that the first determination means determines the first health level using a trained model learned by machine learning.

8. The information processing device according to any one of claims 1 to 7, characterized in that the deformation of the structure includes at least one of cracks, water leakage, potholes, exposed rebar, honeycombing, rust stains, and efflorescence.

9. The information processing apparatus according to any one of claims 1 to 8, characterized in that the second health level is the health level that is the most unhealthy among the first health levels of the plurality of evaluation areas.

10. The information processing apparatus according to any one of claims 1 to 9, characterized in that the image extracted from each of the plurality of deformations is an image generated by extracting a rectangular region encompassing the deformation from an image of a structure taken based on the positional information of each deformation.

11. An information processing method performed by one or more processors of a computer, A first determination step in which, for each of the multiple evaluation areas in a structure, a first soundness level indicating the soundness of the evaluation area is determined based on the deformation contained in the structure, A second determination step in which a second soundness indicating the overall soundness of the structure is determined based on the first soundness in each of the plurality of evaluation areas, The system includes an output step that outputs information regarding the deformation of the structure that formed the basis for the determination in the second determination step, and information indicating the second degree of soundness. In the output step, the images extracted from each of the multiple deformations that formed the basis for determining the second degree of health are output in an order corresponding to the degree of contribution of each of the multiple deformations to the determination of the second degree of health. An information processing method characterized by doing so.

12. A program for causing a computer to function as an information processing device according to any one of claims 1 to 10.

Images