Evaluation device used for residual proof stress evaluation in structure, evaluation method, generating method and learned model

Abstract

To provide a technique for evaluating residual proof stress in a structure such as a pier.SOLUTION: A machine learning section 12 performs machine learning that uses teacher data using a deterioration degree of a floor slab, a deterioration degree of a beam and an external force condition to the floor slab as explanation variables and a damage state of the floor slab as an objective variable to generate a learned model. A recording section 13 records the learned model generated by the machine learning section 12. An input section 14 inputs the deterioration degree of the floor slab that becomes an evaluation target, the deterioration degree of the beam that supports the floor slab and the external force condition to the floor slab to the learned model recorded in the recording section 13 as input items. A determination section 15 determines use propriety of the floor slab when external force in accordance with the external force condition is applied, on the basis of the damage state of the floor slab obtained in accordance with the input to the learned model by the input section 14. An output section 16 outputs damage state data corresponding to the damage state of the floor slab obtained in accordance with the input to the learned model by the input section 14.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a technique for evaluating the remaining strength of a structure.

Background Art

[0002] Among various port structures, particularly piers are placed in a harsh environment against salt damage and must be properly maintained. However, there is very little achievement regarding the evaluation of the remaining strength of deteriorated piers.

[0003] In response to such problems, at present, it is impossible to evaluate the remaining strength without conducting a detailed investigation (detailed inspection diagnosis) and then performing a structural analysis based on the results. The timing of the detailed inspection diagnosis is, in the case of a normal inspection diagnosis facility, at least once at an appropriate time during the service period and also at the timing of extending the service period (see Non-Patent Document 1).

[0004] Even if a detailed inspection diagnosis is carried out, in order to evaluate the remaining strength of a pier, it is necessary to perform a structural analysis of the entire pier system, but often the remaining strength is not evaluated. In this case, since the strength that the pier currently has cannot be grasped, when repairs or reinforcements are carried out, a quantitative effect cannot be confirmed, and it is not reasonable to restore it to the original holding strength on the premise of restoring the original shape.

[0005] Regardless of the timing of the regular diagnosis, it is possible to grasp the remaining strength by performing a detailed inspection diagnosis. However, the cost is high and the inspection period is also long. Therefore, at present, it is not actively carried out particularly for privately-owned piers. In addition, since the period from inspection to evaluation of the remaining strength becomes long, the possibility of a large external force (such as seismic force) attacking during that period also increases. Furthermore, if the remaining strength is not evaluated even after a detailed inspection diagnosis is carried out, as described above, it will be restored to the original holding strength on the premise of restoring the original shape, so a quantitative and effective repair and reinforcement will not be carried out, and as a result, the cost will increase, which is not reasonable.

[0006] From the above, there is a need for a technique that can perform remaining strength evaluation more simply, quickly, and inexpensively. Therefore, in Non-Patent Document 2, a method for relatively easily performing remaining strength evaluation of a pier using a general-purpose structural analysis tool has been proposed. However, this proposes a remaining strength evaluation method targeting beam deterioration and does not target the floor slab.

[0007] Regarding the evaluation of the remaining performance of a floor slab with corroded reinforcement, the loading test results are shown in Non-Patent Document 3. Here, when the structural performance index of the superstructure of the pier is set to 100 at the time of construction, the state where it has decreased to 60 (65 in the example shown in the document) is defined as the design limit point, and this state is defined as the deterioration degree "A". A floor slab with this deterioration degree "A" was subjected to a loading test, and it was confirmed that the member strength was about 60%.

Prior Art Documents

Non-Patent Documents

[0008]

Non-Patent Document 1

Non-Patent Document 2

Non-Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0009] Incidentally, on a pier, vehicles such as trucks pass or wait on the upper surfaces of girders and floor slabs, generating loading weights thereby. The loaded goods include not only vehicles but also goods loaded and unloaded from vehicles, and transported goods such as containers unloaded from ships by cranes. When the floor slab is deteriorated, it may not be able to hold the loading weight assumed at the design time. In that case, damage such as holes or deformation occurs in the floor slab, and damages such as vehicles and loaded goods falling or being damaged occur.

[0010] In a pier composed of deteriorated floor slabs, it is considered possible to evaluate the remaining strength of the floor slab alone based on the deterioration degree determination result etc. However, it is not easy to grasp which floor slab is damaged to what extent by external forces in the entire pier. The following two main factors can be cited for this. (1) Since the number of floor slabs constituting the pier is large, like the girders, it takes a great deal of time and cost to grasp all of them. (2) Since the floor slab is supported by (or connected to) the girder, depending on the deterioration status of the girder, even if the floor slab is sound or has deterioration that is not severe enough to cause damage to the floor slab, it may not be able to resist external forces such as superimposed loads. In particular, it is necessary to judge the usability of the floor slab in consideration of the deterioration and remaining strength evaluation of the girder.

[0011] The present invention has been made in view of the above-described background, and an object thereof is to provide a technique for evaluating the remaining strength in structures such as piers.

Means for Solving the Problem

[0012] The evaluation device according to the present invention is an evaluation device used for evaluating the remaining load-bearing capacity of a structure. It uses the degradation degree of the floor slab of the structure and the external force conditions acting on the floor slab as explanatory variables, and uses the damage state of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab as the target variable. It includes a recording unit that records a learned model generated by machine learning using teacher data, an input unit that inputs the degradation degree of the floor slab to be evaluated and the external force conditions acting on the floor slab as input items into the learned model, and an output unit that outputs damage state data corresponding to the damage state of the floor slab obtained according to the input to the learned model by the input unit.

[0013] It may be provided with a determination unit that determines whether the floor slab can be used when an external force corresponding to the external force conditions is applied, based on the damage state of the floor slab and the arrangement information of the floor slab obtained according to the input to the learned model by the input unit.

[0014] Moreover, the evaluation device according to the present invention is an evaluation device used for evaluating the remaining load-bearing capacity of a structure. It uses the degradation degree, arrangement information of the floor slab of the structure, and the external force conditions acting on the floor slab as explanatory variables, and uses whether the floor slab can be used as the target variable when an external force corresponding to the external force conditions is applied to the floor slab. It includes a recording unit that records a learned model generated by machine learning using teacher data, an input unit that inputs the degradation degree, arrangement information of the floor slab to be evaluated, and the external force conditions acting on the floor slab as input items into the learned model, and an output unit that outputs availability data corresponding to whether the floor slab can be used obtained according to the input by the input unit.

[0015] As explanatory variables in the learned model, it further includes the degradation degree of a plurality of support members that support the floor slab and the arrangement information. The input unit may further input, as the input items, the degradation degree of a plurality of support members that support the floor slab to be evaluated, the arrangement information, and the external force conditions acting on the plurality of support members into the learned model.

[0016] The explanatory variables may include the dimensions of the floor slab.

[0017] The explanatory variable may include the dimensions of the support member.

[0018] The explanatory variable may include the number of the support members that support the floor slab.

[0019] The external force condition may be an upper load on the floor slab.

[0020] The damage state data may be data representing at least any one of the degree of damage, the damaged area, or the damaged area ratio.

[0021] As the damage state data, the output unit maps at least the damage state of each floor slab to a structure diagram showing a structure including a plurality of floor slabs and a plurality of support members that support each of the floor slabs, and maps the usability of each floor slab to the structure diagram and outputs it.

[0022] A determination unit that determines the arrangement position and repair priority of the floor slab to be repaired based on the damage state of each floor slab and the usability of each floor slab, and the output unit may output the determined arrangement position and repair priority of the floor slab to be repaired.

[0023] A prediction unit that predicts the degree of deterioration at an arbitrary future time from the current degree of deterioration using a probability model for future prediction, and the explanatory variable may include the predicted degree of deterioration.

[0024] Further, the present invention may be an evaluation method implemented using the above evaluation device.

[0025] Further, the present invention may be a generation method for generating a learned model by machine learning using teacher data in which the degree of deterioration of the floor slab of the structure and the external force condition for the floor slab are explanatory variables, and the damage state of the floor slab when an external force corresponding to the external force condition is applied to the floor slab is the objective variable.

[0026] The explanatory variables in the learned model further include the degrees of deterioration of a plurality of support members that support the floor slab, and the objective variable in the learned model may include the damage states of the plurality of support members that support the floor slab when an external force corresponding to the external force condition is applied to the floor slab and the plurality of support members that support the floor slab.

[0027] Further, the present invention may be a learned model generated by the above-described generation method.

Advantages of the Invention

[0028] According to the present invention, the remaining strength in a structure can be evaluated.

Brief Description of the Drawings

[0029]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Best Mode for Carrying Out the Invention

[0030] (Example 1) (Configuration) FIG. 1 is a diagram showing the hardware configuration of an evaluation device 10 according to Example 1 which is an example of an embodiment of the present invention. The evaluation device 10 is an evaluation device used for evaluating the remaining load-carrying capacity in a structure (here, a trestle). Physically, it is configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, and a bus connecting these. Each of these devices operates by electric power supplied from a power source (not shown).

[0031] Each function in the evaluation device 10 is realized by causing a processor 1001 to perform operations, control communication by a communication device 1004, acquire data transmitted from other devices, or control at least one of reading and writing data in a memory 1002 and a storage 1003 by loading a predetermined software (program) onto hardware such as the processor 1001 and a memory 1002.

[0032] The processor 1001 controls the entire computer by operating an operating system, for example. The processor 1001 may be configured by a central processing unit (CPU: Central Processing Unit) including an interface with peripheral devices, a control device, an arithmetic device, registers, and the like.

[0033] The processor 1001 reads a program (program code), a software module, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002, and executes various processes according to these. As the program, a program for causing a computer to execute at least a part of the operations described later is used.

[0034] The memory 1002 is a computer-readable recording medium and may be configured by at least one of, for example, a ROM (Read Only Memory), an EPROM (Erasable Programmable ROM), an EEPROM (Electrically Erasable Programmable ROM), a RAM (Random Access Memory), and the like. The memory 1002 may be referred to as a register, a cache, a main memory (main recording device), or the like. The memory 1002 can store a program (program code), a software module, etc. executable for implementing the method according to the present embodiment.

[0035] Storage 1003 is a computer-readable recording medium, such as a hard disk drive or a flash memory.

[0036] The communication device 1004 is hardware (a transceiver device) for communication between computers.

[0037] The input device 1005 is an input device that receives external input (e.g., a keyboard, a mouse, a microphone, a switch, a button, etc.). The output device 1006 is an output device that performs output to the outside (e.g., a display, a speaker, an LED lamp, etc.).

[0038] FIG. 2 is a diagram showing the functional configuration of the evaluation device 10. The evaluation device 10 realizes functions of an acquisition unit 11, a machine learning unit 12, a recording unit 13, an input unit 14, a determination unit 15, and an output unit 16.

[0039] The acquisition unit 11 acquires teacher data for the machine learning unit 12 to perform supervised learning. The explanatory variables included in this teacher data are the degradation degrees of each floor slab constituting piers of various sizes and shapes, and the external force conditions for each floor slab. Also, the objective variable included in this teacher data is the damage state of the floor slab when an external force corresponding to the above external force condition is applied to each floor slab. Various values are prepared in advance as analysis conditions for the degradation degree and external force condition of each floor slab. The damage state of the floor slab is obtained by performing a structural analysis according to these analysis conditions. The damage state of each floor slab is at least one of the degree of damage of each floor slab, the area of damage of each floor slab, or the area ratio of damage of each floor slab.

[0040] FIG. 3 is a plan view illustrating the degradation degree of the floor slab. In FIG. 3, each rectangular figure arranged in a grid represents each floor slab f. In the first embodiment, the degradation degree is expressed in four levels of a, b, c, and d according to the degree of degradation, but the number of levels may be three levels or five levels, and the number of levels is not particularly limited.

[0041] The external force condition is information representing the stress generated on each floor slab to be evaluated. Once the external force condition is determined, the value of the external force is also determined based on the external force condition. In this embodiment, the external force condition represents the superimposed load on each floor slab. When there are multiple types of superimposed loads as the external force condition, the external force condition is set as exemplified in FIG. 4. In FIG. 4, External Force No. 1, External Force No. 2, …, External Force No. 5 are, for example, the superimposed load by a vehicle, the superimposed load by the goods loaded and unloaded from the vehicle, the superimposed load by transport items such as containers unloaded from a ship by a crane, etc. By combining these External Force No. 1, External Force No. 2, …, External Force No. 5, Combined External Force No. 1, Combined External Force No. 2, …, Combined External Force No. 10 are set.

[0042] As described above, one set of teacher data consists of the deterioration degree of a certain floor slab and the external force condition (explanatory variable) for that floor slab, and the damage state (objective variable) of that floor slab when an external force corresponding to the external force condition is applied to that floor slab. The entire teacher data is composed of these multiple sets of explanatory variables and objective variables. Furthermore, the explanatory variable may include the dimensions of each floor slab (for example, the thickness in the z direction and / or the lengths in the x and y directions). Here, FIG. 5 is a diagram exemplifying the numbering of each floor slab, and FIG. 6 is a diagram exemplifying the dimensions of each floor slab. In FIG. 5, a rule for numbering the floor slab numbers for each floor slab is determined in advance, and the dimensions of each floor slab (for example, the thickness in the z direction and / or the lengths in the x and y directions) are set in association with the numbered floor slab numbers. When including the dimensions of the floor slab in the explanatory variable, in addition to the method of using the real values of the dimensions, it is also possible to use the maximum value ratio (the ratio to the dimension of the largest floor slab in the pier). Also, a method using so-called two-class classification, three-class classification, etc. may be used. For example, in the case of two-class classification, a method can be considered where if the target floor slab is equal to or greater than the average value of the dimensions of the floor slabs in the pier, it is expressed as 1, and if it is less than the average value, it is expressed as 0.

[0043] In addition, in order to clarify which floor slab among a plurality of floor slabs is the evaluation target when evaluating the damage degree of the floor slab, by using the floor slab number, it is possible to obtain together the arrangement information indicating the position of the floor slab to be evaluated in the structure (here, the trestle) to be evaluated. As a result, even if a plurality of floor slabs have the same degree of damage, the influence of the damaged floor slab on the surrounding members may be different depending on the arrangement position on the trestle, and thus different judgment results may be obtained when determining whether it can be put into use as described later. For example, even if the damage degree is the same and the external force conditions are the same, for a floor slab located near the background, a floor slab on the movement path of loads or containers, a floor slab located at the end of the trestle, etc., the degree of influence on the surrounding members is different even with the same stress.

[0044] Returning to FIG. 2, the machine learning unit 12 performs machine learning using, as explanatory variables, the deterioration degree of the floor slab and the external force conditions for the floor slab acquired by the acquisition unit 11, and, as the objective variable, the damage state of the floor slab acquired by the acquisition unit 11, thereby generating a learned model (hereinafter referred to as a damage state learned model).

[0045] The recording unit 13 records the damage state learned model generated by the machine learning unit 12.

[0046] The input unit 14 inputs, as input items, the deterioration degree of the floor slab to be evaluated and the external force conditions for that floor slab into the damage state learned model recorded in the recording unit 13. As described above, the dimensions of each floor slab (for example, the thickness in the z direction and / or the lengths in the x and y directions) may be included as input items.

[0047] The determination unit 15 is a separate function provided in the recording unit 13, and based on the damage state of the floor slab, which is the target variable obtained in response to the input to the learned damage state model recorded in the recording unit 13 by the input unit 14, and the arrangement information of the floor slab, it determines whether the floor slab can be used when an external force corresponding to the external force condition is applied. More specifically, the determination unit 15 calculates in advance the correspondence relationship between the damage state and arrangement information of the floor slab and the usability when an external force corresponding to the external force condition is applied to the floor slab, and obtains the usability result corresponding to the damage state of the floor slab obtained in response to the input to the learned damage state model recorded in the recording unit 13.

[0048] Further, the determination unit 15 may use a learned model obtained by machine learning the relationship between the damage state of the floor slab, the arrangement information regarding the position where the floor slab is arranged, and the usability when an external force corresponding to the external force condition is applied to the floor slab. In this case, the machine learning unit 12 performs supervised learning using teacher data composed of the damage state and arrangement information of the floor slab as the explanatory variables prepared in advance and the usability when an external force corresponding to the external force condition is applied to the floor slab as the target variable, and records the learned model (referred to as a learned usability model) obtained as a result in the determination unit 15. The determination unit 15 inputs the damage state of the floor slab output as the target variable by the learned damage state model recorded in the recording unit 13 as described above together with its arrangement information into the learned usability model recorded in the determination unit 15 to obtain the usability of the floor slab.

[0049] The output unit 16 outputs damage state data corresponding to the damage state of the floor slab obtained in response to the input to the learned damage state model by the input unit 14, and usability data corresponding to the usability of each floor slab constituting the pier based on the damage state data. More specifically, the output unit 16 can output by mapping the damage state of each floor slab to, for example, a structural diagram composed of a plurality of floor slabs as the damage state data. Further, the output unit 16 can output by mapping the usability of each floor slab to, for example, the structural diagram as the usability data of the floor slab.

[0050] (Operation) Next, the operation of machine learning by the evaluation device 10 will be described. In FIG. 7, the acquisition unit 11 acquires teacher data for the supervised learning by the machine learning unit 12 (step S1). This teacher data includes, as explanatory variables, the degree of deterioration of each floor slab constituting the pier of various sizes and shapes and the external force conditions for each floor slab, etc., and as the objective variable, the damage state of the floor slab when an external force corresponding to the above external force conditions is applied to each floor slab.

[0051] Next, the machine learning unit 12 generates a learned damage state model by performing machine learning using the teacher data composed of the explanatory variables and the objective variable acquired by the acquisition unit 11 (step S2).

[0052] The recording unit 13 records the learned damage state model generated by the machine learning unit 12 (step S3).

[0053] Next, the operation when the evaluation device 10 performs the remaining strength evaluation will be described. In FIG. 8, the input unit 14 inputs, as input items, the degree of deterioration of the floor slab to be evaluated and the external force conditions for the floor slab into the learned damage state model recorded in the recording unit 13 (step S11). Note that the dimensions of each floor slab (for example, the thickness in the z direction and / or the lengths in the x and y directions) may be included as input items.

[0054] The determination unit 15 acquires the damage state of the floor slab and the layout information of the floor slab in response to the input to the learned damage state model recorded in the recording unit 13 by the input unit 14 (step S12). Then, the determination unit 15 determines the usability of the floor slab when an external force corresponding to the external force conditions is applied based on the acquired damage state and layout information (step S13).

[0055] The output unit 16 outputs damage state data corresponding to the damage state of the floor slab acquired by the determination unit 15 in response to the input to the learned damage state model recorded in the recording unit 13 by the input unit 14, and usability data corresponding to the usability of each floor slab obtained in response to the input to the learned usability model recorded in the determination unit 15 (step S14).

[0056] More specifically, as shown in FIG. 9, the output unit 16 maps and outputs by classifying, for example, with colors, different gradations, or symbols, the damage states of the respective floor slabs with respect to the structure diagram of the trestle composed of a plurality of floor slabs. Further, as shown in FIG. 10, the output unit 16 maps and outputs by classifying, for example, with colors, different gradations, or symbols, the availability of the respective floor slabs with respect to the structure diagram of the trestle composed of a plurality of floor slabs. Also in FIGS. 9 and 10, as in FIG. 3, each rectangular figure arranged in a grid pattern means each floor slab f. Note that the output unit 16 may superpose and output simultaneously the mapping example shown in FIG. 9 and the mapping example shown in FIG. 10. Also in FIG. 9, the damage state is expressed in two classifications of large and small, but the classification of the damage state is not limited to this example. For example, it may be classified by damage up to the yield load, damage up to the ultimate load, or the like.

[0057] According to the first embodiment described above, it is possible to evaluate the remaining strength of the floor slab under external force conditions of various superimposed loads (such as crane working load, vehicle load, cargo handling load, etc.) without performing structural analysis from the deterioration degree of the floor slab. The number of floor slabs in the trestle is very large, and a great deal of time and cost are required when performing structural analysis. However, since structural analysis is not required as described above, it is possible to immediately grasp the remaining strength. Further, even when attempting to evaluate the remaining strength of the floor slabs of a plurality of trestles, according to the first embodiment, the labor and time can be significantly reduced. Therefore, for example, after evaluating all the remaining strengths of the floor slabs of a plurality of trestles, it is also possible to use it as a primary screening function, such as performing a more detailed structural analysis only on those with significant damage to grasp the specific damage locations.

[0058] (Second Embodiment) (Configuration) In Example 2 described below, in addition to those described in Example 1, the degree of deterioration of the beam, which is a support member supporting the floor slab, and the external force conditions for the beam are used as explanatory variables for machine learning. In addition to those described in Example 1, the damage state of the beam, which is a support member supporting the floor slab, is used as the objective variable for machine learning. In this example, a beam is used as the support member for supporting the floor slab, but the support member is not limited to this, and any member having a role of supporting the floor slab may be used. The hardware configuration and functional configuration of the evaluation device 10 in Example 2 are the same as those in FIGS. 1 and 2 described in Example 1, and the operations of each configuration are different.

[0059] In Example 2, the acquisition unit 11 (FIG. 2) acquires teacher data for the machine learning unit 12 (FIG. 2) to perform supervised learning. The explanatory variables included in this teacher data are the degree of deterioration of each floor slab constituting the trestle of various sizes and shapes, the degree of deterioration of the beams supporting each floor slab, and the external force conditions for each floor slab and a plurality of beams supporting the floor slab. Further, the objective variables included in this teacher data are the damage state of the floor slab when an external force corresponding to the external force condition is applied to each floor slab and a plurality of beams supporting the floor slab, and the damage state of each beam supporting each floor slab. Various values are prepared in advance as analysis conditions for the degree of deterioration of each floor slab, the degree of deterioration of each beam, and the external force conditions. The damage states of each floor slab and each beam are obtained by performing structural analysis according to these analysis conditions. The damage state of each floor slab is at least any one of the degree of damage of each floor slab, the area of damage of each floor slab, or the area ratio of damage of each floor slab. Further, the damage state of each beam is at least any one of the degree of damage of each beam, the area of damage of each beam, or the area ratio of damage of each beam. Here, the area mainly refers to the area on the upper surface of the beam, but is not limited to this, and may be the area of the surface of the beam, or the volume of the beam may be used instead of the area of the beam.

[0060] FIG. 11 is a plan view illustrating the degradation degrees of the floor slab and the beams. It shows the addition of each beam supporting each floor slab to the plan view illustrating the degradation degree of the floor slab in FIG. 3. In FIG. 11, each rectangular figure in a substantially square shape arranged in a grid pattern means each floor slab f, and each rectangular figure in a rectangular shape located around each floor slab f means the beam b supporting that floor slab f. Note that, although the shapes of the floor slab and the beam are shown using simple shapes for the sake of explanation, they may have different shapes in the actual plan view. In the second embodiment, the degradation degrees of the floor slab and the beam are expressed in four levels of a, b, c, and d according to the degree of degradation. However, the number of divisions may be three divisions or five divisions, and the number of divisions is not particularly limited.

[0061] The external force condition is the superimposed load on each floor slab. When there are multiple types of superimposed loads as the external force condition, the external force condition is set in the same manner as FIG. 4 described in the first embodiment.

[0062] As described above, in the second embodiment, one set of training data consists of the degradation degree of a certain floor slab, the degradation degrees of a plurality of beams supporting that floor slab, and the external force condition (explanatory variable) for that floor slab and the plurality of beams supporting that floor slab, and the damage state of that floor slab and the damage state of the beams supporting that floor slab (objective variable) when an external force corresponding to the above external force condition is applied to that floor slab and the plurality of beams supporting that floor slab. The entire training data is constituted by these multiple sets of explanatory variables and objective variables.

[0063] Furthermore, the explanatory variables may include the dimensions of each floor slab, the dimensions of the beams supporting each floor slab, or the number of beams supporting each floor slab. Here, FIG. 12 is a diagram illustrating the numbering of each beam, and FIG. 13 is a diagram illustrating the dimensions of each beam. In FIG. 12, a rule for numbering beam numbers for each beam is determined in advance, and the dimensions of each beam (for example, the thickness in the z direction and / or the lengths in the x and y directions) are set in association with the numbered beam numbers. When including the dimensions of the beams in the explanatory variables, in addition to the method of using the real values of the dimensions as in the floor slabs in Example 1, methods using the maximum value ratio, so-called two-class classification, three-class classification, etc. may also be used. Also, in order to clarify which beam among a plurality of beams is the beam to be evaluated when evaluating the damage degree of the beam, by using the beam number, it is possible to obtain together the arrangement information indicating the position where the beam to be evaluated is arranged in the trestle to be evaluated. As a result, even if a plurality of beams have the same degree of damage, the influence of the damaged beam on the surrounding members may differ depending on the arrangement position in the trestle, and thus different judgment results may be obtained when determining whether it can be put into use, which will be described later.

[0064] The machine learning unit 12 (FIG. 2) performs machine learning using, as explanatory variables, the degradation degree of the floor slab, the degradation degree of the beams, and the external force conditions for the floor slab and the plurality of beams supporting the floor slab acquired by the acquisition unit 11, and using, as objective variables, the damage states of the floor slab and the beams acquired by the acquisition unit 11, to generate a learned damage state model.

[0065] The recording unit 13 (FIG. 2) records the learned damage state model generated by the machine learning unit 12.

[0066] The input unit 14 (FIG. 2) inputs, as input items, the degradation degree of the floor slab to be evaluated, the degradation degrees of the plurality of beams supporting the floor slab, and the external force conditions for the floor slab and the plurality of beams supporting the floor slab, into the learned damage state model recorded in the recording unit 13. Note that the dimensions of each beam of the floor slab and the plurality of beams supporting the floor slab (for example, the thickness in the z direction and / or the lengths in the x and y directions) may be included as input items.

[0067] The determination unit 15 (Fig. 2) determines the usability of the floor slab when an external force corresponding to the external force condition is applied, based on the damage state of the floor slab and the damage state of the beam obtained in response to the input to the damage state learned model recorded in the recording unit 13 by the input unit 14, and the arrangement information of the floor slab and the beam. More specifically, the determination unit 15 obtains in advance by calculation the correspondence relationship between the damage state of the floor slab, the damage state of the beam supporting the floor slab, and the usability when an external force corresponding to the external force condition is applied to the floor slab and the beam, and obtains the usability result corresponding to the damage state of the floor slab and the damage state of the beam obtained in response to the input to the damage state learned model. Further, the determination unit 15 may use a learned model obtained by machine learning the relationship between the damage state of the floor slab and the usability when an external force corresponding to the external force condition is applied to the floor slab and the beam. In this case, the machine learning unit 12 performs supervised learning using teacher data including the damage state of the floor slab and the damage state of the beam as explanatory variables prepared in advance, the arrangement information of the floor slab and the beam, and the usability when an external force corresponding to the external force condition is applied to the floor slab and the beam as the objective variable, and records the learned model for usability, which is the learned model obtained as a result, in the determination unit 15. The determination unit 15 inputs the damage states of the floor slab and the beam output as the objective variable by the damage state learned model recorded in the recording unit 13 as described above and the arrangement information of the floor slab and the beam into the learned model for usability recorded in the determination unit 15 to obtain the usability of the floor slab. Here, for example, when the floor slab is usable and all of the plurality of beams supporting the floor slab are usable, the floor slab is determined to be usable, while when the floor slab is usable and at least one of the plurality of beams supporting the floor slab is unusable, the floor slab is determined to be unusable.

[0068] The output unit 16 outputs damage state data corresponding to the damage state of the floor slab obtained according to the input to the learned damage state model recorded in the recording unit 13 by the input unit 14 and the damage state of the beam supporting the floor slab. More specifically, the output unit 16 can output, as the damage state data, by mapping at least the damage state of the floor slab among each floor slab and the beam supporting each floor slab to a structural diagram showing a structure composed of a plurality of floor slabs and a plurality of beams supporting each of the floor slabs. Further, the output unit 16 can output, as the usability data of the trestle having the floor slab and the beam, by mapping the usability data of each floor slab obtained according to the input to the learned usability model recorded in the determination unit 15 to the structural diagram.

[0069] (Operation) Next, the operation of the machine learning by the evaluation device 10 will be described. In FIG. 7 mentioned above, the acquisition unit 11 acquires teacher data for the machine learning unit 12 to perform supervised learning (step S1). This teacher data includes, as explanatory variables, the degradation degree of each floor slab constituting trestles of various sizes and shapes, the degradation degree of each beam supporting each floor slab, the external force conditions for each floor slab, etc., and as objective variables, the damage state of the floor slab and the damage state of the beam supporting the floor slab when an external force corresponding to the above external force conditions is applied to each floor slab and the beam supporting the floor slab.

[0070] Next, the machine learning unit 12 performs machine learning using the teacher data composed of the explanatory variables and the objective variables acquired by the acquisition unit 11 to generate a learned damage state model (step S2).

[0071] The recording unit 13 records the learned damage state model generated by the machine learning unit 12 (step S3).

[0072] Next, the operation when the remaining strength evaluation is performed by the evaluation device 10 will be described. In FIG. 8 mentioned above, the input unit 14 inputs, as input items, the deterioration degree of the floor slab to be evaluated, the deterioration degrees of a plurality of beams supporting the floor slab, and the external force conditions for the floor slab and the plurality of beams supporting the floor slab, into the learned damage state model recorded in the recording unit 13 (step S11).

[0073] The determination unit 15 acquires the damage state of the floor slab, the damage state of the beams, and the layout information of the floor slab and the plurality of beams supporting the floor slab in response to the input to the learned damage state model by the input unit 14 (step S12). Then, the determination unit 15 determines the usability of the floor slab when an external force corresponding to the external force condition is applied, based on the acquired damage state of the floor slab, the damage state of the beams, and the layout information of the floor slab and the beams (step S13).

[0074] The output unit 16 outputs damage state data corresponding to the damage state of the floor slab obtained in response to the input to the learned damage state model recorded in the recording unit 13 by the input unit 14, and usability data corresponding to the usability of each floor slab constituting the trestle (step S14).

[0075] More specifically, as shown in FIG. 14, the output unit 16 maps and outputs, for a structure diagram showing the structure of a trestle composed of a plurality of floor slabs and a plurality of beams, the damage states of the respective floor slabs and the damage states of the respective beams, for example, by distinguishing them with colors, different gradations, symbols, or the like. Further, as shown in FIG. 15, the output unit 16 maps and outputs, for a structure diagram showing the structure of a trestle composed of a plurality of floor slabs and a plurality of beams, the usability of the respective floor slabs, for example, by distinguishing them with colors, different gradations, symbols, or the like. In FIGS. 14 and 15, each rectangular figure that is substantially square and arranged in a grid pattern means each floor slab f, and each rectangular figure that is rectangular and located around each floor slab f means a beam b that supports that floor slab f. The shapes of the floor slabs and beams are shown using simple shapes for the sake of explanation, but they may have different shapes in an actual plan view. Note that the output unit 16 may output by superimposing the mapping example shown in FIG. 14 and the mapping example shown in FIG. 15 at the same time. Also, in FIG. 14, the damage state is expressed in two categories of large and small, but the classification of the damage state is not limited to this example. For example, it may be classified by damage up to the yield load, damage up to the ultimate load, or the like.

[0076] According to the second embodiment described above, it is possible to evaluate the remaining strength of the floor slab under external force conditions of various superimposed loads (such as crane working load, vehicle load, and cargo handling load) without performing structural analysis based on the degradation degrees of the floor slab and the beam. There are a very large number of floor slabs in a trestle, and a great deal of time and cost are required when performing structural analysis. However, since structural analysis is not required as described above, it is possible to immediately grasp the remaining strength. Further, even when attempting to evaluate the remaining strengths of the floor slabs of a plurality of trestles, according to the second embodiment, the labor and time can be significantly reduced. Therefore, for example, after evaluating all the remaining strengths of the floor slabs of a plurality of trestles, it is also possible to utilize it as a primary screening function, such as performing a more detailed structural analysis only on those with significant damage to grasp the specific damaged locations.

[0077] (Modification example) The present invention is not limited to the above-described embodiments (Examples 1 and 2). The above-described embodiments may be modified as follows. Also, two or more of the following modification examples may be combined and implemented.

[0078] (Modification Example 1) In the above-described embodiment, a pier was used as an object of remaining strength evaluation. However, the present invention is applicable to structures other than piers as long as they are floor slabs or structures having support members for supporting the floor slabs.

[0079] For example, the present invention may be applied to a road bridge which is a structure composed of a floor slab and a plurality of bridge girders for supporting the floor slab. In the case of a road bridge, by performing a soundness diagnosis in accordance with the inspection requirements described in the "Road Bridge Regular Inspection Guide" (Ministry of Land, Infrastructure, Transport and Tourism, Road Bureau, p. 3, 2019.2), a soundness determination distribution as shown in FIG. 17 can be obtained in accordance with the soundness determination classification shown in FIG. 16. As the external force conditions assumed in this case, the live load etc. described in the "Road Bridge Design Manual and Explanation I Common Volume" (Japan Road Association, p. 43, 2017.11) may be targeted.

[0080] Further, the present invention may be applied to a bridge having a double-deck structure (or a multi-deck structure with more than two decks). For example, it is a structure having a double-deck structure on the uphill and downhill lanes of a road bridge (or a railway bridge). Such a structure has a plurality of road surfaces (superstructure) in the vertical direction. Therefore, there will be two or more girder structures (beam structures) composed of a floor slab and support members, one above the other. In this case, it may be possible to create input data (explanatory variables) for each girder structure. However, if the characteristic quantities of each girder structure are calculated individually, the relationship between the topmost girder structure (hereinafter referred to as the first surface) and the girder structures located below it (the second surface, the third surface, ···, the nth surface) will be lost. Although structures located higher are more likely to sway during an earthquake, they will be learned or predicted without distinction. Therefore, by assigning a girder number indicating which surface the girder is on and performing machine learning based on the deterioration degree and damage state including the girder number, it becomes possible to consider the characteristics unique to each surface as a result. As a method of numbering the girder numbers in this case, a method as illustrated in FIG. 18 can be considered.

[0081] (Modification 2) Steps S11 to S14 in FIG. 8 in the above-described Example 1 may be realized by a single learned model. The learned model in this case is generated by machine learning using teacher data having the deterioration degree of the floor slab of the structure, the external force conditions for the floor slab, and the layout information of the floor slab as explanatory variables, and the usability of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab as the target variable. That is, the evaluation device according to the present invention includes a recording unit that records a learned model generated by machine learning using teacher data having the deterioration degree of the floor slab of the structure, the external force conditions for the floor slab, and the layout information of the floor slab as explanatory variables, and the usability of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab as the target variable, an input unit that inputs the deterioration degree of the floor slab to be evaluated, the external force conditions for the floor slab, and the layout information of the floor slab as input items to the learned model, and an output unit that outputs usability data corresponding to the usability obtained in response to the input by the input unit. In this modification, it is not necessary to provide the determination unit 15 in the recording unit 13 shown in FIG. 2.

[0082] Similarly, steps S11 to S14 in FIG. 8 in the above-described Example 2 may be realized by a single learned model. In this case, the learned model uses, as explanatory variables, the degree of deterioration of the floor slab of the structure, the degree of deterioration of the support members, the external force conditions for the floor slab and a plurality of support members supporting the floor slab, and the layout information of the floor slab and a plurality of beams supporting the floor slab, and is generated by machine learning using teacher data having, as the objective variable, the usability of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab and a plurality of support members supporting the floor slab. That is, the evaluation device according to the present invention uses, as explanatory variables, the degree of deterioration of the floor slab of the structure, the degree of deterioration of the support members, the external force conditions for the floor slab and a plurality of support members supporting the floor slab, and the layout information of the floor slab and a plurality of beams supporting the floor slab, and may include a recording unit that records a learned model generated by machine learning using teacher data having, as the objective variable, the usability of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab and a plurality of support members supporting the floor slab, an input unit that inputs, as input items, the degree of deterioration of the floor slab to be evaluated, the degree of deterioration of the support members, the external force conditions for the floor slab and a plurality of support members supporting the floor slab, and the layout information of the floor slab and a plurality of beams supporting the floor slab into the learned model, and an output unit that outputs usability data corresponding to the usability obtained according to the input by the input unit. In this modification, it is not necessary to provide the determination unit 15 in the recording unit 13 shown in FIG. 2.

[0083] (Modification 3) In the present invention, the determination unit 15 for determining the usability described in Examples 1 and 2 is not essential, and it may only output the damage state. Further, in the present invention, the configuration for performing machine learning and the configuration for performing the remaining strength evaluation do not necessarily have to be provided in a single device, and may be provided in different devices respectively.

[0084] (Modification 4) When repairing the floorboards, it is important to determine which floorboards to repair and in what order. Therefore, as shown in FIG. 19, a determination unit 17 may be provided between the recording unit 13 and the output unit 16 shown in FIG. 2, and this determination unit 17 determines the position of the floorboards to be repaired and the priority of repair based on the damage state of each floorboard and the judgment of whether each floorboard can be used. Note that when there is no determination unit 15, the determination unit 17 determines the priority of repair based on the damage state and the arrangement position of each floorboard. Specifically, an example is conceivable in which a group of floorboards that cannot be used is extracted, and among the floorboard group, the ones closer to the end of the trestle are repaired in order, and then the ones that can be used but have a large damage state are repaired in order. That is, the evaluation device according to the present invention includes a determination unit that determines the arrangement position of the floorboards to be repaired and the priority of repair based on the damage state of each floorboard and the usability of each floorboard, and the output unit may output the determined arrangement position of the floorboards to be repaired and the priority of repair.

[0085] (Modification Example 5) Regarding each floorboard or support member (beam), it is also conceivable to predict the degree of deterioration at an arbitrary future time point from the current degree of deterioration. As an algorithm for predicting the degree of deterioration at an arbitrary future time point from the current degree of deterioration, those generally used as probability models for future prediction (for example, Markov chain models, etc.) can be utilized. That is, the evaluation device according to the present invention includes a prediction unit that predicts the degree of deterioration at an arbitrary future time point from the current degree of deterioration using a probability model for future prediction, and the explanatory variable may include the predicted degree of deterioration.

[0086] (Modification Example 6) Note that the present invention may be implemented as an evaluation method implemented using the above-described evaluation device.

[0087] (Modification Example 7) Further, the present invention may be implemented as a generation method for generating a learned model (damage state learned model) by machine learning using teacher data in which the deterioration degree of the pier floor slab and the external force conditions applied to the floor slab are explanatory variables, and the damage state of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab is the objective variable. In this case, the explanatory variables in the learned model (damage state learned model) may further include the deterioration degree of a plurality of support members that support the floor slab, and the objective variable in the learned model (damage state learned model) may include the damage states of the plurality of support members that support the floor slab when an external force corresponding to the external force conditions is applied to the floor slab. Further, the present invention may be implemented as the learned model (damage state learned model) generated by the above generation method. The program for operating the evaluation device described above may be provided in a state recorded on a computer-readable recording medium such as a CD-ROM (Compact Disc Read only memory), or may be downloaded via a communication network such as the Internet.

Explanation of Signs

[0088] 10: Evaluation device, 11: Acquisition unit, 12: Machine learning unit, 13: Recording unit, 14: Input unit, 15: Determination unit, 16: Output unit, 17: Decision unit, 1001: Processor, 1002: Memory, 1003: Storage, 1004: Communication device, 1005: Input device, 1006: Output device, f... Floor slab, b... Beam.

Claims

1. An evaluation device used for evaluating the remaining load-bearing capacity of a structure, wherein the degree of deterioration of the floor slab of the structure and the external force conditions acting on the floor slab are used as explanatory variables, and a recording unit that records a learned model generated by machine learning using teacher data with the damage state of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab as the target variable; an input unit that inputs the degree of deterioration of the floor slab to be evaluated and the external force conditions acting on the floor slab as input items into the learned model; and an output unit that outputs damage state data corresponding to the damage state of the floor slab obtained according to the input to the learned model by the input unit. The evaluation device comprising the above components.

2. The evaluation device according to claim 1, further comprising a determination unit that determines whether the floor slab can be used when an external force corresponding to the external force conditions is applied based on the damage state and arrangement information of the floor slab obtained according to the input to the learned model by the input unit. The evaluation device according to claim 1.

3. An evaluation device used for evaluating the remaining load-bearing capacity of a structure, wherein the degree of deterioration of the floor slab of the structure, the arrangement information, and the external force conditions acting on the floor slab are used as explanatory variables, and teacher data with whether the floor slab can be used as the target variable when an external force corresponding to the external force conditions is applied to the floor slab is used for machine learning to generate a learned model, and a recording unit that records the learned model; an input unit that inputs the degree of deterioration of the floor slab to be evaluated, the arrangement information, and the external force conditions acting on the floor slab as input items into the learned model; and an output unit that outputs availability data corresponding to whether the floor slab can be used obtained according to the input by the input unit. The evaluation device comprising the above components.

4. As explanatory variables in the learned model, further including the degree of deterioration of a plurality of support members supporting the floor slab and the arrangement information, The evaluation device according to claim 3, wherein the input unit further inputs, as the input items, the degree of deterioration of a plurality of support members supporting the floor slab to be evaluated, the arrangement information, and the external force conditions acting on the plurality of support members into the learned model.

5. The explanatory variables include the dimensions of the floor slab. The evaluation device according to any one of claims 1 to 4.

6. The explanatory variables include the dimensions of the support members. The evaluation device according to claim 4.

7. The explanatory variables include the number of support members supporting the floor slab. The evaluation device according to claim 4.

8. The external force conditions are the superimposed load acting on the floor slab. The evaluation device according to any one of claims 1 to 4.

9. The damage state data is data representing at least any one of the degree of damage, the area of damage, or the ratio of the area of damage. The evaluation device according to claim 1 or 2.

10. The output unit As the damage state data, with respect to a structural diagram showing a structure composed of a plurality of floor slabs and a plurality of support members supporting each of the floor slabs, the damage state of at least the floor slab among each of the floor slabs and the support members is mapped. With respect to the structural diagram, the availability of each of the floor slabs is mapped and output. The evaluation device according to claim 1 or 2.

11. A determination unit is provided that determines the arrangement position and repair priority of the floor slab to be repaired based on the damage state of each of the floor slabs and the availability of each of the floor slabs. The output unit outputs the determined arrangement position and repair priority of the floor slab to be repaired. The evaluation device according to any one of claims 2 to 4.

12. A prediction unit is provided that predicts the degree of deterioration at an arbitrary future time point from the current degree of deterioration using a probability model for future prediction. The explanatory variable includes the predicted degree of deterioration. The evaluation device according to any one of claims 1 to 4.

13. An evaluation method implemented using the evaluation device according to any one of claims 1 to 4.

14. A generation method for generating a learned model by machine learning using teacher data with the degree of deterioration of the floor slab of the structure and the external force conditions on the floor slab as explanatory variables, and the damage state of the floor slab when an external force corresponding to the external force conditions is applied to the floor slab as the target variable.

15. The explanatory variable in the learned model further includes the degree of deterioration of a plurality of support members supporting the floor slab. The target variable in the learned model includes the damage state of a plurality of support members supporting the floor slab when an external force corresponding to the external force conditions is applied to the floor slab and a plurality of support members supporting the floor slab. The generation method according to claim 14.

16. A learned model generated by the generation method according to claim 14 or 15.

Images