Grout filling learning data generation system, grout filling learning data generation program, concrete cavity evaluation learning data generation system, and concrete cavity evaluation learning data generation program

Abstract

To provide a grout filling learning data generation system, a grout filling learning data generation program, a concrete cavity evaluation learning data generation system, and a concrete cavity evaluation learning data generation program capable of generating learning data for evaluating cavities of concrete, such as grout filling status.SOLUTION: A grout filling learning data generation system comprises: first acquisition means for acquiring generation waveform data obtained by non-destructively inspecting a cable sheath of a PC structure, and condition data related to a grout filling condition; and generation means for generating learning waveform data based on the condition, on the basis of the generation waveform data and the condition data acquired by the first acquisition means.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a grout filling learning data generation system and a grout filling learning data generation program for evaluating the grout filling status in the cable sheath of a PC structure by non-destructive inspection.

Background Art

[0002] Conventionally, when injecting grout into the cable sheath of a PC structure such as a post-tensioned bridge, viaduct, or building, a pressure pump is connected to the grout injection side of the cable sheath. Then, the pressure pump is operated to fill the cable sheath with grout.

[0003] By the way, in recent years, corrosion and breakage accidents of tendons due to poor grout filling in the cable sheath of PC structures have been occasionally seen. In such cases, as a method of grasping the grout filling status in the cable sheath, a method of drilling holes and directly observing the inside is common, but it is desired to evaluate by non-destructive inspection in consideration of minimizing the influence on the structure.

[0004] For example, in Patent Document 1, in a grout filling status evaluation system for evaluating the grout filling status in the cable sheath of a PC structure, acquisition means for acquiring waveform data obtained by performing non-destructive inspection on the cable sheath for evaluating the grout filling status, and a restoration model that is generated by machine learning using the waveform data as learning data, limits the frequency having characteristics in the input waveform data, and outputs restoration data obtained by restoring the evaluation waveform data based on the limited characteristics, and output means for outputting restoration data based on the waveform data, and evaluation means for evaluating the grout filling status based on the restoration data and the waveform data are provided. A grout filling status evaluation system is disclosed.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] On the other hand, when making a determination using electromagnetic waves, impact sounds, impact elastic waves, or ultrasonic waves in the grout non-filling inspection, the result of this determination is affected by various factors such as the material and position of the PC steel material used, the position of the reinforcing bars, the sheath material and position, the filling rate, the influence of the structure edge, the change in the member thickness of the structure, the concrete propagation speed of the elastic wave, and the attached hardware. Also, when the member surface with respect to the structure is not parallel, etc., the elastic wave diverges, resulting in a weakening of the resonance frequency intensity obtained, etc., and the measurement accuracy decreases. For these reasons, when evaluating this grout filling situation using the disclosed technology of Patent Document 1, in order to improve the evaluation accuracy, a large number of learning data are required.

[0007] However, in the disclosed technology of Patent Document 1, in order to obtain this large number of learning data, it is necessary to perform inspections under various conditions, which has the problem that a great deal of cost and labor are required.

[0008] Therefore, the present invention has been devised in view of the above-described problems, and its object is to provide a grout filling learning data generation system, a grout filling learning data generation program, a concrete void evaluation learning data generation system, and a concrete void evaluation learning data generation program capable of generating learning data for evaluating the voids of concrete such as the filling situation of grout.

Means for Solving the Problems

[0009] The learning data generation system for grout filling according to the first invention includes a first acquisition means for acquiring generation waveform data obtained by non-destructive inspection of a cable sheath of a PC structure and condition data regarding grout filling conditions, and a generation means for generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired by the first acquisition means.

[0010] The learning data generation system for grout filling according to the second invention is the same as the first invention, wherein the generation means refers to a generation model learned using learning data with the generation waveform data and condition data as input data and the learning waveform data as output data, and generates the learning waveform data based on the generation waveform data and condition data acquired by the first acquisition means.

[0011] The learning data generation system for grout filling according to the third invention is the same as the first invention, further including a second acquisition means for acquiring evaluation waveform data obtained by non-destructive inspection of a cable sheath for evaluating the grout filling status, an output means for outputting restored data obtained by referring to a restoration model that is generated by machine learning using the learning waveform data generated by the generation means as learning data, limiting the frequencies having characteristics in the input evaluation waveform data, and restoring the evaluation waveform data based on the limited characteristics, and an evaluation means for evaluating the grout filling status based on the restored data output by the output means and the evaluation waveform data acquired by the second acquisition means.

[0012] The learning data generation system for grout filling according to the fourth invention is the same as the first invention, wherein the first acquisition means acquires the condition data including configuration data regarding components of the PC structure.

[0013] The grout filling learning data generation system according to the fifth invention is characterized in that, in the first invention, the first acquisition means acquires the condition data including shape data regarding the cross-sectional shape of the PC structure.

[0014] The grout filling learning data generation system according to the sixth invention is characterized in that, in the first invention, the first acquisition means acquires the condition data including depth data regarding the depth from the surface of the PC structure to the cable sheath.

[0015] The grout filling learning data generation system according to the seventh invention is characterized in that, in the first invention, the acquisition means acquires the condition data including incidental data regarding the cable sheath.

[0016] The grout filling learning data generation program according to the eighth invention causes a computer to execute a first acquisition step of acquiring generation waveform data obtained by non-destructive inspection of a cable sheath of a PC structure and condition data regarding grout filling conditions, and a generation step of generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired in the first acquisition step.

[0017] The learning data generation system for concrete void evaluation according to the ninth invention includes a first acquisition means for acquiring generation waveform data obtained by non-destructive inspection of a concrete structure and condition data regarding the conditions of the concrete structure, and a generation means for generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired by the first acquisition means.

[0018] The learning data generation system for concrete void evaluation according to the 10th invention, in the 9th invention, includes a second acquisition means for acquiring evaluation waveform data obtained by non-destructive inspection of concrete for evaluating voids, and a learning waveform data generated by the generation means is used as learning data and generated by machine learning, and a restoration model that limits the frequency having characteristics in the input evaluation waveform data and outputs restoration data obtained by restoring the evaluation waveform data based on the limited characteristics is referred to, and an output means for outputting restoration data based on the evaluation waveform data acquired by the second acquisition means, and an evaluation means for evaluating the voids of the concrete based on the restoration data output by the output means and the evaluation waveform data acquired by the second acquisition means.

[0019] The learning data generation program for concrete void evaluation according to the 11th invention causes a computer to execute a first acquisition step of acquiring generation waveform data obtained by non-destructive inspection of a concrete structure and condition data regarding the conditions of the concrete structure, and a generation step of generating learning waveform data based on the conditions based on the generation waveform data and the condition data acquired in the first acquisition step.

Advantages of the Invention

[0020] According to the 1st to 11th inventions, pseudo learning waveform data based on conditions is generated based on the generation waveform data and the condition data. As a result, it becomes possible to generate learning waveform data that reflects the conditions in the generation waveform data. Therefore, it is possible to generate learning data for evaluating the voids of concrete such as the grout filling status.

[0021] In particular, according to the 2nd invention, a generation model is referred to, and learning waveform data is generated based on the generation waveform data and the condition data. As a result, it becomes possible to generate learning waveform data that reflects the conditions with higher accuracy.

[0022] In particular, according to the third invention, with reference to a restoration model generated by machine learning using learning waveform data as learning data, restoration data based on evaluation waveform data is output, and the grout filling status is evaluated based on the restoration data and the evaluation waveform data. Thereby, for a cable sheath for newly evaluating the grout filling status, waveform data obtained by performing non-destructive inspection is input into the restoration model, and based on the output restoration data and the waveform data, it becomes possible to evaluate the abnormal value when comparing the newly acquired waveform data with the waveform data when the grout has a predetermined filling degree. Thereby, even when evaluating the grout filling status in the case where a frequency is prominent or in a structure in which a plurality of cable sheaths are arranged, it becomes possible to evaluate the grout filling status with high accuracy.

[0023] In particular, according to the fourth invention, condition data including configuration data regarding the constituent members of the PC structure is acquired. Thereby, it becomes possible to generate optimal learning waveform data according to the constituent members of the PC structure, and it is possible to evaluate the grout filling status with higher accuracy.

[0024] In particular, according to the fifth invention, condition data including shape data regarding the cross-sectional shape of the PC structure is acquired. Thereby, it becomes possible to generate optimal learning waveform data according to the cross-sectional shape, and it is possible to evaluate the grout filling status with higher accuracy.

[0025] In particular, according to the sixth invention, condition data including depth data regarding the depth from the surface of the PC structure to the cable sheath is acquired. Thereby, it becomes possible to generate optimal learning waveform data according to the depth, and it is possible to evaluate the grout filling status with higher accuracy.

[0026] In particular, according to the seventh invention, condition data including incidental data regarding the cable sheath is acquired. Thereby, it becomes possible to generate optimal learning waveform data according to the incidental data, and it is possible to evaluate the grout filling status with higher accuracy.

Brief Description of the Drawings

[0027]

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

Embodiments for Carrying Out the Invention

[0028] Hereinafter, the concrete void evaluation system to which the present invention is applied will be described in detail with reference to the drawings.

[0029] FIG. 1 is a block diagram showing the overall configuration of a concrete void evaluation system 1 to which the present invention is applied. The concrete void evaluation system 1 evaluates the voids of a concrete structure, such as the grout filling condition in a cable sheath 71 of a concrete structure 7, by non-destructive inspection. The concrete void evaluation system 1 includes a non-destructive inspection unit 8, a conversion device 9 connected to the non-destructive inspection unit 8, an evaluation device 2 connected to the conversion device 9, and a database 3 connected to the evaluation device 2.

[0030] The concrete structure 7 is a PC structure such as a bridge, viaduct, or building in which PC steel materials and a cable sheath 71 are disposed inside, but is not limited thereto, and may be any concrete structure in which PC steel materials and a cable sheath 71 are not disposed inside.

[0031] The cable sheath 71 is disposed in a tensioned state with a PC steel material (not shown) such as a PC steel bar or a large number of PC steel wires inside, and is spaced apart from the inner wall surface of the cable sheath 71. Incidentally, in the present embodiment, since a post-tensioned concrete structure 7 is taken as an example for explanation, in such a case, after the cable sheath 71 is disposed in the concrete structure 7, the concrete is filled and cured, and then a PC steel material (not shown) is inserted into the cable sheath 71 to apply a tensile stress. Further, thereafter, the cable sheath 71 is filled with grout and cured.

[0032] The non-destructive inspection unit 8 applies elastic waves to the concrete structure 7 by non-destructive inspection methods such as the impact elastic wave method, the impact echo method, and the ultrasonic method, and detects the reflected waveform as inspection data. The non-destructive inspection unit 8 transmits the detected inspection data to the conversion device 9.

[0033] The conversion device 9 is composed of electronic devices such as a PC (personal computer), a smartphone, a tablet terminal, and a wearable terminal. This conversion device 9 converts the acquired inspection data into waveform data including at least the intensity and frequency for each frequency for evaluating the grout filling status in the cable sheath 71, and transmits the converted waveform data to the evaluation device 2.

[0034] The conversion device 9 may transmit the acquired inspection data to the evaluation device 2 as waveform data without any special analysis. For example, when waveform data indicating the frequency and the intensity for each frequency is acquired as inspection data, the conversion device 9 may transmit the inspection data as it is as waveform data to the evaluation device 2 without any special analysis.

[0035] Alternatively, the conversion device 9 may analyze the acquired inspection data, convert it into waveform data, and transmit it to the evaluation device 2. For example, when waveform numerical data indicating the relationship between time and amplitude is acquired as inspection data, the conversion device 9 may analyze this and, after converting it into waveform data indicating the frequency and the intensity for each frequency, transmit it to the evaluation device 2. By analyzing the inspection data composed of waveform numerical data, the conversion device 9 will appropriately process it into an optimal one for evaluating the grout filling degree in the cable sheath 71.

[0036] For example, this conversion device 9 may convert the waveform numerical data indicating the relationship between the acquired time and the amplitude into waveform data indicating the relationship between the frequency and the intensity for each frequency by performing FFT (Fast Fourier Transform). Further, the conversion device 9 performs Fourier transform on the waveform numerical data indicating the relationship between the acquired time and the amplitude to convert it into waveform numerical data indicating the relationship between time and frequency, then removes a specific frequency region, and further performs inverse Fourier transform to convert it into waveform numerical data indicating the relationship between time and amplitude with the specific frequency region remaining. Further, the conversion device 9 may perform Fourier transform on the waveform numerical data indicating the relationship between time and amplitude with the specific frequency region remaining to convert it into waveform data indicating the relationship between the frequency with the specific frequency region remaining and the intensity for each frequency. Further, the conversion device 9 may convert the waveform numerical data indicating the relationship between the acquired time and the amplitude into waveform data indicating the relationship between the intensity and the frequency such as a power spectrum.

[0037] The conversion device 9 may perform wavelet transform on the waveform numerical data as the acquired inspection data to convert it into waveform numerical data indicating the relationship between time and frequency. By performing wavelet transform, it is possible to retain the time characteristics that are lost during Fourier transform. For this reason, by performing wavelet transform, it will be converted into waveform numerical data indicating the relationship between time and frequency.

[0038] The conversion device 9 may convert the acquired waveform numerical data as inspection data into waveform numerical data indicating a cepstrum obtained by taking the logarithm of the value of the power spectrum obtained by Fourier-transforming the waveform numerical data and then performing an inverse Fourier transform. Further, the conversion device 9 may further extract waveform numerical data indicating a spectral envelope, which is a lower-order cepstrum, or a spectral fine structure, which is a higher-order cepstrum, from the waveform numerical data indicating the cepstrum. For example, the waveform numerical data indicating the spectral envelope may be extracted by determining the cepstrum order. This cepstrum order may take any value such as 20, 100, etc. Further, in the waveform numerical data indicating the spectral envelope, the coefficients of each cepstrum order may be extracted. Also, the conversion device 9 may convert the acquired waveform image into waveform numerical data indicating a formant, which is a peak when the amplitude cut out from a certain time region is converted into the frequency domain, into waveform numerical data. When the frequency bands of the peaks are arranged from the lowest, as the first formant, the second formant, ···, the waveform numerical data may indicate, for example, the relationship between the first formant and the second formant and the relationship between frequencies. Also, the conversion device 9 may perform AFTE (Auditory filterbank temporal Envelope) conversion on the acquired waveform numerical data as inspection data.

[0039] In this way, the conversion device 9 may convert the acquired inspection data into two-dimensional waveform data as shown in FIG. 2. The two-dimensional waveform data may indicate, for example, the relationship between two of time, amplitude, frequency, intensity, spectrum, cepstrum, formant, etc. Also, the reciprocals of these may be taken.

[0040] Also, the conversion device 9 may perform, for example, a spectrogram or the like to convert the acquired inspection data into three-dimensional waveform numerical data. The three-dimensional waveform numerical data may indicate, for example, the relationship between three of time, amplitude, frequency, intensity, spectrum, cepstrum, formant, etc.

[0041] Incidentally, this conversion device 9 can display each inspection data via a display unit composed of, for example, a display (not shown). Further, the conversion device 9 can store these respective data in a storage, display these data on the display unit based on an instruction by a user, or write these data to a portable memory. The user can remove this portable memory from the conversion device 9 and carry it around freely. Furthermore, the conversion device 9 can also transfer these respective data to other electronic devices via a public communication network.

[0042] Note that the configuration of this conversion device 9 is not essential in the present invention and may be omitted. In such a case, the inspection data output from the non-destructive inspection unit 8 is directly transmitted to the evaluation device 2.

[0043] The database 3 stores various data. In the database 3, data sent via a public communication network or data input by a user of this system is accumulated. Further, the database 3 transmits this accumulated data to the evaluation device 2 based on a request from the evaluation device 2.

[0044] The evaluation device 2 is composed of an electronic device such as a personal computer (PC), for example. However, in addition to a PC, it may be embodied by any other electronic device such as a mobile phone, a smartphone, a tablet terminal, a wearable terminal, etc. The user can evaluate whether the filling of grout in the cable sheath 71 is sufficient or whether there are voids in the concrete structure 7 by obtaining the evaluation result of voids such as the filling state of grout as a search solution by this evaluation device 2. And, for example, when the filling of grout in the cable sheath 71 is insufficient, an operation of filling the cable sheath 71 with grout by operating a pressure pump (not shown) is performed.

[0045] Figure 3 shows a specific configuration example of the evaluation device 2. This evaluation device 2 includes a control unit 24 for controlling the entire evaluation device 2, an operation unit 25 for inputting various control commands via operation buttons, a keyboard, etc., a communication unit 26 for performing wired or wireless communication, a search unit 27 for searching for optimal design conditions, a storage unit 28 represented by a hard disk, etc., for storing programs for performing searches to be executed, and a generation unit 29 for generating learning waveform data, which are respectively connected to an internal bus 21. Further, a display unit 23 as a monitor for actually displaying data is connected to this internal bus 21.

[0046] The control unit 24 is a so-called central control unit for controlling each component implemented in the evaluation device 2 by transmitting a control signal via the internal bus 21. Also, this control unit 24 transmits various control commands via the internal bus 21 in response to operations via the operation unit 25.

[0047] The operation unit 25 is embodied by a keyboard or a touch panel, and an execution command for executing a program is input from the user. When an execution command is input from the user, this operation unit 25 notifies the control unit 24 of it. The control unit 24 that receives this notification starts with the search unit 27 and executes a desired processing operation in cooperation with each component. Also, various data may be input from the user to the operation unit 25. The operation unit 25 may include, for example, configuration data regarding the constituent members of the concrete structure 7, shape data regarding the cross-sectional shape of the concrete structure 7, or additional data including depth data regarding the depth from the surface of the concrete structure 7 to a cable sheath for newly evaluating the grout filling status, etc., which may be input by the user.

[0048] The exploration unit 27 explores the evaluation results of the grout filling status. When executing the exploration operation, this exploration unit 27 reads out various data stored in the storage unit 28 and various data stored in the database 3 as necessary data. This exploration unit 27 may be controlled by artificial intelligence. This artificial intelligence may be based on any well-known artificial intelligence technology.

[0049] The display unit 23 is composed of a graphic controller that creates a display image based on the control by the control unit 24. This display unit 23 is realized by, for example, a liquid crystal display (LCD) or the like.

[0050] When the storage unit 28 is composed of a hard disk, based on the control by the control unit 24, predetermined data is written to each address and read out as necessary. Also, a program for executing the present invention is stored in this storage unit 28. This program will be read out and executed by the control unit 24.

[0051] The operation of the concrete void evaluation system 1 having the above-described configuration will be described.

[0052] In the concrete void evaluation system 1, inspection data obtained by performing a non-destructive inspection on a concrete structure 7 for evaluating voids is converted into waveform data, the converted waveform data is input into a restoration model, restoration data is output, and based on the output restoration data and the waveform data, voids such as the grout filling status are evaluated. The processing operation flow of this concrete void evaluation system 1 is shown in FIG. 4. Hereinafter, the detailed processing in each step of FIG. 4 will be described.

[0053] First, in step S11, the non-destructive inspection unit 8 detects inspection data of the concrete structure 7 by a non-destructive inspection method such as the impact elastic wave method, the impact echo method, or the ultrasonic method. In such a case, as shown in FIG. 1, the frequency of the elastic wave to be excited may be changed according to the depth Y from the surface of the concrete structure 7 to the cable sheath 71 or the void in the concrete structure 7. For example, when evaluating the cable sheath 71 with a large depth Y, a low frequency is input. The non-destructive inspection unit 8 outputs the detected inspection data to the conversion device 9. Also, an elastic wave with a frequency f shown in the following formula (1) may be excited. In such a case, V represents the propagation speed of the elastic wave.

Equation

[0054] Next, in step S12, the conversion device 9 performs various analyses on the inspection data detected by the non-destructive inspection unit 8 as necessary, and also processes the waveform data as necessary to facilitate the search by the subsequent search device. Incidentally, if various analyses and processing are not performed, this step S12 can be omitted. The conversion device 9 outputs the converted waveform data as evaluation waveform data to the evaluation device 2. The evaluation waveform data is waveform data used to evaluate the grout filling status.

[0055] Also, in step S12, the conversion device 9 may normalize the evaluation waveform data. The conversion device 9 may normalize the evaluation waveform data, for example, with the maximum value of the intensity for each frequency being 1 and the minimum value being 0. Also, the conversion device 9 may normalize the intensity at the common frequency points at frequencies other than the frequency corresponding to the depth Y, with the maximum value being 1 and the minimum value being 0.

[0056] Next, proceed to step S13, refer to a restoration model that compresses the input evaluation waveform data and outputs restoration data obtained by restoring the compressed evaluation waveform data, and output restoration data based on the evaluation waveform data output from the conversion device 9 in step S12.

[0057] FIG. 5 is a diagram showing a restoration model. The restoration model is a model that compresses the input evaluation waveform data and outputs restored data obtained by restoring the compressed evaluation waveform data. The restoration model includes an encoder that compresses the input waveform and a decoder that restores the compressed evaluation waveform data. Further, the restoration model may be a learned model generated by machine learning using past evaluation waveform data obtained by non-destructive inspection of the concrete structure 7 as learning data. In such a case, the restoration model is generated by compressing the evaluation waveform data of the learning data, extracting the features of the compressed evaluation waveform data, and accumulating the features of the evaluation waveform data. Thereby, the restoration model can compress the input evaluation waveform data and restore the compressed evaluation waveform data based on the features of the learning data. The restoration model may be generated by machine learning such as an AutoEncoder, for example. Further, the features of the evaluation waveform data are data that can reproduce the input evaluation waveform data. Further, the compression of the evaluation waveform data is to convert the evaluation waveform data into data of a lower dimension. For example, the evaluation waveform data may be represented by frequency and intensity for each frequency, but in such a case, the compression of the evaluation waveform data is to limit the frequencies having features. Further, the restoration of the evaluation waveform data is to convert the evaluation waveform data into data of a higher dimension. For example, the evaluation waveform data may be represented by frequency and frequency intensity, but in such a case, the restoration of the evaluation waveform data is to convert the data from the frequencies with limited features to the original frequencies.

[0058] As the evaluation waveform data serving as learning data, for example, past evaluation waveform data obtained by non-destructive inspection of the concrete structure 7 may be used. This evaluation waveform data may use, for example, a plurality of past evaluation waveform data obtained by non-destructive inspection by exciting elastic waves at frequencies of 10 Hz each from 2500 to 50000 Hz. Further, as the evaluation waveform data, for example, artificial evaluation waveform data obtained by pseudo-generation assuming the sheath or the concrete propagation speed of elastic waves for the evaluation waveform data of frequencies of 10 Hz each from 2500 to 50000 Hz may be used.

[0059] Also, the training data may be classified into normal data and abnormal data according to the void condition such as the grout filling condition of the concrete structure 7 when nondestructive inspection is performed. For example, when the grout filling condition of the cable sheath 71 at the time of nondestructive inspection is a filling degree equal to or higher than a preset threshold value, or when there is no void in the concrete structure 7, the evaluation waveform data obtained may be regarded as normal data, and when the filling degree is lower than the threshold value, the evaluation waveform data obtained may be regarded as abnormal data. Specifically, when the filling degree is 100%, or when there is no void in the concrete structure 7, the evaluation waveform data obtained may be regarded as normal data, and when the filling degree is less than 100%, or when there is a void in the concrete structure 7, the evaluation waveform data obtained may be regarded as abnormal data. Also, when the grout filling condition of the cable sheath 71 at the time of nondestructive inspection is a plurality of types of filling degrees equal to or higher than a preset threshold value, the plurality of evaluation waveform data obtained may be regarded as normal data, and when the filling degrees are less than the threshold value, the evaluation waveform data obtained may be regarded as abnormal data. Specifically, assuming the threshold value is 80%, when the filling degrees are 100% and 90%, the respective evaluation waveform data obtained may be regarded as normal data, and when the filling degrees are 70% and 50%, the respective evaluation waveform data obtained may be regarded as abnormal data. Also, the training data may be normalized evaluation waveform data.

[0060] Further, the restoration model may generate new evaluation waveform data by varying the intensity and frequency band for each frequency of the past evaluation waveform data, and use the generated evaluation waveform data as further training data. For example, the restoration model may generate evaluation waveform data having an intensity of 70% to 130% of the intensity for each frequency of the evaluation waveform data measured in the past, and use the generated evaluation waveform data as further training data. This is a value considering the variation in the input location of the device that inputs elastic waves. Further, the restoration model may use, as training data, the evaluation waveform data generated artificially in addition to the past evaluation waveform data. The evaluation waveform data generated artificially is data of a waveform having a virtual peak calculated using the assumed sheath depth and the assumed concrete propagation speed of the elastic wave. Further, the evaluation waveform data generated artificially includes the evaluation waveform data created artificially. Further, the restoration model may newly generate evaluation waveform data having a frequency with a difference of 200 Hz or less from the frequency of the past evaluation waveform data, and obtain the generated evaluation waveform data as training data. This is a value considering that the concrete propagation speed of the elastic wave to be evaluated is not uniform but has variations in the concrete structure 7, and that there are errors in the design and construction of the positions of the voids and the cable sheath 71. Thereby, even when there are variations in the intensity and frequency for each frequency of the evaluation waveform data to be obtained when evaluating voids such as the grout filling condition, it is possible to evaluate accurately.

[0061] The number of nodes 61 in the input layer and the output layer of the restoration model is the same, and the number of nodes 61 in the intermediate layer is less than the number of nodes 61 in the input layer. For example, the number of nodes 61 in the input layer and the output layer may be 4750, and the number of nodes 61 in the intermediate layer may be 8. Further, in the restoration model, each node 61 in the input layer is connected to each node 61 in the intermediate layer with a correlation of three or more levels, and each node 61 in the intermediate layer is connected to each node 61 in the output layer with a correlation of three or more levels. Further, a plurality of hidden layers may be provided between the input layer and the intermediate layer, or between the intermediate layer and the output layer, respectively. Each node 61 in the hidden layer is connected to each node 61 in the adjacent layer with a correlation of three or more levels.

[0062] The degree of correlation indicates the accuracy between each node 61. For example, for node 61a in the input layer that takes evaluation waveform data A as input, node 61b and node 61c in the hidden layer are connected with a degree of correlation of 30% and 60% respectively. In such a case, node 61c with a higher degree of correlation is closer to making an accurate judgment compared to node 61b. This degree of correlation may be indicated, for example, by five levels or a percentage.

[0063] Also, as learning data used for learning the restoration model, pseudo-generated learning waveform data may be used. The generation method of this learning waveform data will be described. FIG. 6 is a flowchart showing the operation of the concrete void evaluation system 1 to which the present invention is applied for generating learning waveform data.

[0064] First, in step S1, the non-destructive inspection unit 8 detects inspection data of the concrete structure 7 by a non-destructive inspection method such as the impact elastic wave method, the impact echo method, or the ultrasonic method. The non-destructive inspection unit 8 detects inspection data in the same manner as in step S11. Also, in such a case, as the concrete structure 7 to be used, the concrete structure 7 that actually evaluates the grout filling status used in step S11 may be used, but it is not limited thereto, and any concrete structure 7 may be used.

[0065] Also, in step S1, the evaluation device 2 acquires condition data regarding the conditions of the concrete structure 7. The condition data is data regarding the conditions of the concrete structure 7. Also, the condition data may be data regarding the conditions of the grout filling. The condition data is, for example, data such as the material and position of the PC steel material 72 to be used, the position of the reinforcing bars, the sheath material and position, the filling rate, the influence of the edge of the concrete structure 7, the change in the member thickness of the concrete structure 7, the concrete propagation speed of elastic waves, and the data of the attached hardware. Also, the condition data may be data of the inclination of the surface of the member where the non-destructive inspection unit 8 inspects the concrete structure 7. Also, the condition data may include any of the form data, incidental data, depth data, configuration data, and shape data of the cable sheath 71. In step S1, the evaluation device 2 may acquire condition data such as the type of the PC steel material 72, the sheath material, the reinforcing bar position from the input drawing or bridge survey report, the edge position and shape of the concrete structure 7 from the drawing or visual inspection, the material and position of the attached hardware, and the concrete propagation speed of elastic waves from the extraction inspection.

[0066] Next, in step S2, the conversion device 9 performs various analyses on the inspection data detected by the non-destructive inspection unit 8 as needed, and processes the waveform data as needed to facilitate the search by the subsequent search device, in the same manner as in step S12. The conversion device 9 outputs the converted waveform data as generation waveform data to the evaluation device 2. The generation waveform data is waveform data for generating learning waveform data to be used as learning data for the restoration model.

[0067] Next, in step S3, the evaluation device 2 generates waveform data for learning. The waveform data for learning is waveform data used for learning the restoration model. The waveform data for learning is waveform data that is pseudo-generated by reflecting the conditions of the conditional data in the waveform data for generation. The waveform data for learning is waveform data that includes features based on conditions. The evaluation device 2 refers to a generation model learned using, for example, learning data in which the waveform data for generation and the conditional data are input data and the waveform data for learning is output data, and based on the waveform data for generation and the conditional data acquired in steps S1 and S2, generates the waveform data for learning.

[0068] As a method for generating the generation model, for example, machine learning using a neural network as a model may be used to generate the generation model. The generation model may be learned using, for example, machine learning using a neural network such as a CNN (Convolution Neural Network), or any model may be used. Further, as a method for generating the generation model, for example, Retrieval-Augmented Generation (RAG), seq2seq (Sequence To Sequence) linear discrimination, support vector machine, k-nearest neighbor method, random forest, deep learning, etc. may be used to generate the generation model.

[0069] In such a case, as shown in FIG. 7, for example, the generation model stores a correlation having a degree of correlation between the waveform data for generation and the conditional data, which are the input data, and the waveform data for learning, which is the output data. The degree of correlation indicates the degree of connection between the input data and the output data. For example, it can be determined that the higher the degree of correlation, the stronger the connection between the respective data. The degree of correlation is indicated by, for example, three or more values or three or more levels such as a percentage, or may be indicated by two values or two levels. Further, the conditional data and the waveform data for learning used for learning the degree of correlation are, for example, the waveform data for generation, the conditional data, and the waveform data for learning for use in learning data acquired in advance, but are not limited thereto, and data acquired at any timing may be used.

[0070] For example, the relevance is constructed based on the degree of connection among a plurality of input data, pairs, and a plurality of output data. The relevance is appropriately updated in the process of machine learning, and represents, for example, a classifier using a function optimized based on the plurality of input data and the plurality of output data. Note that the relevance may have a plurality of degrees of relevance indicating the degree of connection among each data, for example. The degree of relevance can be made to correspond to a weight variable, for example, when the database is constructed by a neural network. The relevance may indicate the degree of connection between a plurality of input data and a plurality of output data, for example, as shown in FIG. 7. In this case, by using the relevance, for each input data from "waveform data A for generation and condition data A" to "waveform data C for generation and condition data C" in FIG. 7, the degree of relationship with the plurality of output data from "waveform data A for learning" to "waveform data C for learning" can be linked and stored. Therefore, for example, a plurality of input data can be linked to one output data via the relevance. Thereby, it is possible to realize a multi-faceted selection of output data for the input data. Further, the input data and the output data are not limited to this, and any type of data may be further used.

[0071] The relevance has a plurality of degrees of relevance that link each input data and each output data, respectively. The degree of relevance is indicated by three or more levels such as a percentage, a 10-level scale, or a 5-level scale, and is indicated by, for example, a feature of a line (such as thickness). For example, "waveform data A for generation and condition data A" included in the input data indicates a degree of relevance AA of "73%" with "waveform data A for learning" included in the output data, and a degree of relevance AB of "12%" with "waveform data B for learning" included in the output data. That is, the "degree of relevance" indicates the degree of connection between each data. For example, the higher the degree of relevance, the stronger the connection between each data.

[0072] Obtain in advance the correlation of three or more levels as shown in FIG. 7. That is, when actually discriminating the solution, accumulate the input data and the output data that have been adopted and evaluated, and the past data sets, and create the correlation shown in FIG. 7 by analyzing and analyzing these.

[0073] For example, assume that in the past, for the input data of "generation waveform data B and condition data B", "learning waveform data B" was determined to be the most suitable and evaluated. By collecting and analyzing such data sets, the correlation between the input data and the output data becomes stronger.

[0074] This analysis and analysis may be performed by artificial intelligence. In such a case, for example, when there are many cases where "learning waveform data B" is estimated for the input data of "generation waveform data B and condition data B", set a higher correlation between this "generation waveform data B and condition data B" and "learning waveform data B".

[0075] Also, this correlation may be composed of the nodes of a neural network in artificial intelligence. That is, the weighting coefficient of this neural network node with respect to the output will correspond to the above-mentioned correlation. Also, it may be composed of any decision-making factor that constitutes artificial intelligence, not limited to neural networks.

[0076] Also, the generation model may be provided with at least one or more hidden layers between the input data and the output data and be machine-learned. The above-mentioned correlation is set in either or both of the input data or the hidden layer data, which becomes the weighting of each data, and the output is selected based on this. And when this correlation exceeds a certain threshold value, the output may be selected.

[0077] Such a degree of correlation serves as learning data in the context of artificial intelligence. By pre-learning such learning data, in actuality at step S3, the evaluation device 2 will output learning waveform data based on newly generated waveform data and condition data. When outputting, for example, refer to the degree of correlation shown in Fig. 7 obtained in advance. For example, when the newly acquired waveform data for generation and condition data are the same as or similar to "waveform data A for generation and condition data A", they are associated via the degree of correlation with the degree of correlation AA "73%" with "learning waveform data A" and the degree of correlation AB "12%" with "learning waveform data B". In this case, select "waveform data A for generation and condition data A" with the highest degree of correlation as the optimal solution. However, it is not essential to select the one with the highest degree of correlation as the optimal solution; instead, it is also possible to select "learning waveform data B" whose degree of correlation is low but whose relevance itself is recognized as the optimal solution. Additionally, it goes without saying that it is also possible to select an output solution that is not connected by an arrow. As long as it is based on the degree of correlation, it can be selected in any other priority order.

[0078] By referring to such a degree of correlation, even when the waveform data for generation and condition data are dissimilar, in addition to the case where they are the same as or similar to the input data, it is possible to quantitatively select output data suitable for the input data.

[0079] The generation model may be, for example, similar to the restoration model, a model that compresses the input waveform data for generation and outputs the learning waveform data obtained by restoring the compressed waveform data for generation. In such a case, according to the condition data acquired in advance, using different learning data respectively, from a plurality of generated generation models, refer to the generation model corresponding to the condition data acquired in step S1, compress the input waveform data for generation, and output the learning waveform data obtained by restoring the compressed waveform data for generation.

[0080] Also, in step S3, the evaluation device 2 may further generate new learning waveform data based on the generated learning waveform data. In such a case, for example, waveform data obtained by changing the intensity of the generated learning waveform data from 0.1 or more to within 10 times may be newly obtained as learning waveform data. Also, for example, new learning waveform data may be generated based on the generated learning waveform data and the newly obtained condition data. In such a case, for example, as the condition data, a condition in which the position of the PC steel material 72 of the cable sheath 71 is shifted by a distance within 3 times the diameter of the cable sheath 71 may be used to generate new learning waveform data.

[0081] FIG. 8(a) is a diagram showing a cross section of the cable sheath 71 with a grout filling rate of 100%. FIG. 8(b) is a diagram showing a cross section of the cable sheath 71 with a grout filling rate of 50%. FIG. 8(c) is a diagram showing a cross section of the cable sheath 71 with a grout filling rate of less than 100%. FIG. 8(d) is a diagram showing a cross section of the cable sheath 71 when there is a space 73 between the PC steel materials 72. In step S3, the evaluation device 2 may newly generate learning waveform data using, for example, condition data with a filling rate as shown in FIG. 8 as a condition. Also, the evaluation device 2 may classify the generated learning waveform data.

[0082] The evaluation device 2 may classify, for example, learning waveform data generated based on condition data with a condition of a filling rate of less than 100% as abnormal data. Also, the evaluation device 2 may classify the learning waveform data in association with the condition data. In such a case, the evaluation device 2 may, for example, associate and classify the learning waveform data for each form data, accessory data, depth data, configuration data, and shape data included in the condition data.

[0083] Further, the evaluation device 2 may refer to the generation model and newly generate learning waveform data based on the generated learning waveform data and the condition data. Further, the evaluation device 2 refers to a determination model learned using, for example, learning data in which the learning waveform data is input data and the determination result is output data, and determines the learning waveform data based on the learning waveform data generated in step S3. The determination result is information indicating whether there is an abnormality in the learning waveform data or whether it is normal. Thereby, since it is possible to determine whether there is an abnormality in the generated learning waveform data, the learning waveform data can be generated with higher accuracy.

[0084] Through the above-described steps S1 to S3, the operation of generating the learning waveform data is completed. The evaluation device 2 uses the generated learning waveform data as learning data for the restoration model and performs learning of the restoration model. Thereby, it becomes possible to generate learning data for evaluating the grout filling state.

[0085] The concrete void evaluation system 1 may generate the above-described restoration model in advance, store it in the database 3, and output the restoration model to the evaluation device 2 in response to a request from the evaluation device 2. In step S13, the evaluation device 2 refers to the above-described restoration model and outputs restoration data based on the evaluation waveform data output in step S12.

[0086] Further, in step S13, the evaluation device 2 may refer to the restoration model generated using only the learning waveform data associated with specific condition data as learning data and output the restoration data. Further, in step S13, the evaluation device 2 may refer to the restoration model generated using only normal data as learning data, for example, and output the restoration data. Thereby, since the restoration data is restored based on the characteristics of the normal data, it becomes evaluation waveform data based on the characteristics of the normal data. Thereby, comparison between the evaluation waveform data and the normal data becomes possible, and the grout filling state can be evaluated with high accuracy.

[0087] Also, in step S13, the evaluation device 2 may refer to the restoration model generated using only abnormal data as learning data and output the restored data. As a result, since the restored data is restored based on the characteristics of the abnormal data, it becomes evaluation waveform data based on the characteristics of the abnormal data. Thereby, comparison between the evaluation waveform data and the abnormal data becomes possible, and voids such as the grout filling status can be evaluated with high accuracy.

[0088] Further, the database 3 may store a plurality of restoration models associated with the characteristics of the learning data respectively. The database 3 may store a plurality of restoration models associated with the condition data used when generating the learning waveform data respectively. For example, the restoration models may be stored respectively associated with the configuration data regarding the constituent members of the concrete structure 7 included in the condition data input from the user or acquired in step S1 when each learning data was obtained. In such a case, in step S13, the evaluation device 2 acquires, for example, the configuration data input from the user via the operation unit 25, and refers to the restoration model learned using the learning waveform data associated with the same or corresponding configuration data as the acquired configuration data, and outputs the restored data. Thereby, it becomes possible to refer to appropriate restoration models respectively according to the configuration data, and even when the constituent members of the concrete structure 7 are different, it becomes possible to accurately evaluate the grout filling status. Note that the configuration data regarding the constituent members of the concrete structure 7 means data indicating, for example, structural forms such as T-girders, I-girders, box girders, and parts such as cantilever floor slabs, webs, and flanges.

[0089] FIG. 9 is a diagram showing a cross-section of a T-shaped concrete structure 9. When the concrete structure 9 is T-shaped with edges as shown in FIG. 9 for example, when a non-destructive inspection is performed, a combined waveform of a waveform A reflected from the edge 7T and a waveform B reflected from the PC steel material 72 is detected. Therefore, it is necessary to consider the influence of the waveform A reflected from the edge 7T. In step S13, the evaluation device 2 acquires the input configuration data, refers to the restoration model learned using the learning waveform data associated with the same or corresponding configuration data as the acquired configuration data, and outputs the restoration data, thereby outputting the restoration data considering the influence of the waveform A reflected from the edge 7T.

[0090] FIG. 10(a) is a diagram showing a state of non-destructive inspection on an I-shaped concrete structure 7. FIG. 10(b) is a diagram showing waveform data when non-destructive inspection is performed on the I-shaped concrete structure 7. When non-destructive inspection is performed on the I-shaped concrete structure 7, a waveform P1 reflected from the end and a waveform P2 reflected from the cable sheath 71 without cloud filling are detected. Therefore, it is necessary to consider the influence of the waveform P1 reflected from the end and the waveform P2 reflected from the cable sheath 71 without cloud filling. In step S13, the evaluation device 2 acquires the input configuration data, refers to the restoration model learned using the learning waveform data associated with the same or corresponding configuration data as the acquired configuration data, and outputs the restoration data, thereby outputting the restoration data considering the influence of the waveform P1 reflected from the end and the waveform P2 reflected from the cable sheath 71 without cloud filling.

[0091] Further, the restoration model may be stored in association with the shape data regarding the cross-sectional shape of the concrete structure 7 included in the condition data input from the user or acquired in step S1 when each piece of learning data was obtained. In such a case, in step S13, the evaluation device 2 acquires, for example, via the operation unit 25, the shape data included in the condition data input from the user or acquired in step S1, and refers to the restoration model learned using the learning waveform data associated with the shape data identical or corresponding to the said shape data, and outputs the restoration data. Thereby, it becomes possible to refer to an appropriate restoration model according to the shape data, and even when the cross-sectional shapes of the concrete structures 7 are different, it becomes possible to accurately evaluate the grout filling status. Note that the shape data regarding the cross-sectional shape of the concrete structure 7 is, for example, the thickness, width, length, etc. of the member, and also includes the distance from the member end.

[0092] Further, the restoration model may be stored in association with the depth data from the surface of the concrete structure 7 to the cable sheath 71 for newly evaluating the grout filling status included in the condition data input from the user or acquired in step S1 when each piece of learning data was obtained. In such a case, in step S13, the evaluation device 2 acquires, for example, via the operation unit 25, the depth data included in the condition data input from the user or acquired in step S1, and refers to the restoration model learned using the learning waveform data associated with the depth data identical or corresponding to the said depth data, and outputs the restoration data. Thereby, it becomes possible to refer to an appropriate restoration model according to the depth data, and even when the depth from the surface of the concrete structure 7 to the target of the void such as the cable sheath 71 is different, it becomes possible to accurately evaluate the voids of the concrete structure 7 such as the grout filling status.

[0093] Also, the restoration model may store, in association with the restoration model, the incidental data included in the condition data input by the user or acquired in step S1 when each learning data is acquired. The incidental data includes any one or more of the form data of the cable sheath 71, the arrangement data of the cable sheath 71, the tendon data regarding the tendon (PC steel material) provided in the concrete structure 7 in which the cable sheath 71 is provided, the grout data regarding the grout filled in the cable sheath 71, the concrete data placed in the concrete structure 7 in which the cable sheath 71 is provided, the data of the test conditions for the non-destructive inspection of the cable sheath 71, the surface condition data of the concrete structure 7 in which the cable sheath 71 is provided, and the configuration data within the concrete structure 7 in which the cable sheath 71 is provided.

[0094] As the form data of the cable sheath 71, there are the length of the cable sheath 71, the diameter of the cable sheath 71, the material of the cable sheath, the degree of corrosion of the cable sheath, and the disappearance of the cable sheath due to corrosion, etc. The arrangement data of the cable sheath 71 includes all data related to the arrangement form of the cable sheath 71 in the concrete structure 7. Examples of this arrangement data of the cable sheath 71 include, for example, data on whether the cable sheaths are arranged in parallel in the front view and the intervals between the cable sheaths 71, etc. The tendon data includes the length of the PC steel material, the type of PC steel material, the tension, etc. The grout data is the Young's modulus of the grout, etc. The concrete data is the Young's modulus of the concrete placed in the concrete structure 7 and the presence or absence of cover concrete, etc. The data of the test conditions for non-destructive inspection includes, in the case of the impact elastic wave method, data related to the impact force of the steel ball that strikes the inspection object, etc. The data of the test conditions for non-destructive inspection includes, in the case of the impact echo method, data related to the diameter of the impact steel ball for striking the inspection object, etc. The data of the test conditions for non-destructive inspection includes, in the case of the ultrasonic method, data related to the frequency of the ultrasonic wave applied to the inspection object, etc. The surface condition data of the concrete structure 7 includes data on the presence or absence of unevenness, the cracking situation, the dry-wet situation, etc. The component data inside the concrete structure 7 is the arrangement interval of the reinforcing bars, etc.

[0095] Figure 6(a) is a diagram showing the restored data obtained by referring to the restored model generated using the evaluation waveform data and the normal data acquired when the grout filling degree is 100% as learning data and restoring the compressed evaluation waveform data. Figure 6(b) is a diagram showing the restored data obtained by referring to the restored model generated using the evaluation waveform data and the normal data acquired when the grout filling degree is 0% as learning data and restoring the compressed evaluation waveform data.

[0096] In step S14, as shown in FIGS. 6(a) and 6(b), the evaluation device 2 evaluates the grout filling status based on the evaluation waveform data converted in step S12 and the restored data output in step S13. The evaluation device 2 may calculate an abnormal value based on, for example, the difference between the evaluation waveform data converted in step S12 and the restored data output in step S13. In such a case, for example, the cumulative value of the squared difference values of the intensities for each frequency of the evaluation waveform data and the restored data may be used as the abnormal value. Further, the evaluation device 2 may calculate an abnormal value based on, for example, the ratio between the evaluation waveform data converted in step S12 and the restored data output in step S13. In such a case, for example, the cumulative value of the squared values obtained by adding -1 to the ratio of the intensities for each frequency of the evaluation waveform data and the restored data may be used as the abnormal value. Since this abnormal value increases according to the difference between the evaluation waveform data and the restored data, it serves as an index indicating the difference between the evaluation waveform data and the restored data. Since the restored data is the evaluation waveform data reflecting the characteristics of the learning data of the restoration model referred to in step S13, a large abnormal value indicates that the evaluation waveform data does not contain the characteristics of the learning data. For example, as shown in FIGS. 6(a) and 6(b), when referring to a restoration model generated using normal data as the learning data, the abnormal value in FIG. 6(b) into which the evaluation waveform data obtained when the grout filling degree is 100% is input is larger than that in FIG. 6(a). As a result, for example, it becomes possible to evaluate the similarity between the evaluation waveform data and the normal data by using the learning data as the normal data, and it becomes possible to evaluate the similarity between the evaluation waveform data and the abnormal data by using the learning data as the abnormal data. Thereby, even when evaluating voids such as the grout filling status in a case where a frequency is prominent or in a concrete structure 7 in which a plurality of cable sheaths 71 are arranged, it becomes possible to evaluate the voids with high accuracy.

[0097] Also, in step S14, the evaluation device 2 may calculate respective outlier values based on a comparison between a plurality of evaluation waveform data and respective restoration data obtained from the evaluation waveform data, and set the average of the calculated plurality of outlier values as the outlier value.

[0098] Next, the process proceeds to step S15, and the grout filling status in the cable sheath 71 evaluated in step S14 is displayed via the display unit 23. As a result, the user can immediately grasp the status of the voids in the concrete structure 7 by visually checking the display unit 23.

[0099] Furthermore, in the present invention, the above-described degree of association may be updated. This update may, for example, reflect data provided via a public communication network including the Internet. When new findings are discovered regarding the relationship between the input parameters and the output solution (void evaluation result) through site data, writings, etc. that can be acquired from the public communication network, the degree of association is increased or decreased according to the findings.

[0100] This update of the degree of association may be performed manually or automatically on the system side or the user side based on research data, papers, conference presentations, newspaper articles, books, etc. by experts, in addition to being based on data that can be acquired from the public communication network. Artificial intelligence may be utilized in these update processes.

[0101] Next, the evaluation result of the concrete void evaluation system 1 to which the present invention is applied will be described with reference to the drawings.

[0102] FIG. 12 is a schematic diagram of a concrete specimen 70. The concrete specimen 70 is a specimen composed of a concrete member that mimics a structure. The concrete specimen 70 is, for example, a rectangular parallelepiped having a rectangle with a short side of 400 mm and a long side of 1000 mm as a cross section. The concrete specimen 70 includes a cable sheath 71 inside. Y indicates the depth from the lower surface 70a of the concrete specimen 70 to the cable sheath 71.

[0103] The evaluation waveform data obtained by non-destructive inspection that excites elastic waves from the Z direction to the lower surface 70a of the concrete specimen 70 is input into the concrete void evaluation system 1 to which the present invention is applied. At this time, the depth Y is set to 120 mm, and evaluation is performed using the evaluation waveform data obtained for each grout filling degree of 0% and 100% of the cable sheath 71. Further, at this time, a restoration model generated using the evaluation waveform data obtained when the grout filling degree is 100% as learning data is used.

[0104] FIG. 13 is a graph showing outliers with respect to the number of data for each grout filling degree. In FIG. 13, the triangular legend indicates the outlier when the grout filling degree is 100%, and the circular legend indicates the outlier when the grout filling degree is 0%. The large legend indicates the average of the outliers for each grout filling degree.

[0105] As shown in FIG. 13, when comparing the cases where the grout filling degree is 0% and 100%, the outlier is lower when the grout filling degree is 100%, and the outlier is larger when the grout filling degree is 0%. From this, it can be seen that the restored data is restored based on the characteristics of the normal data obtained when the grout filling degree is 100%. Thus, by using the concrete void evaluation system 1 to which the present invention is applied, it becomes possible to highly accurately evaluate the void situation such as the grout filling situation.

Explanation of Signs

[0106] 1 Concrete void evaluation system 2 Evaluation device 3 Database 7 Concrete structure 8 Non-destructive inspection unit 9 Conversion device 21 Internal bus 23 Display unit 24 Control unit 25 Operation unit 26 Communication unit 27 Search unit 28 Storage unit 29 Generation unit 61 nodes 70 concrete specimens 71 cable sheaths 72 PC steel materials 73 spaces 107 PC structures 171 cable sheaths

Claims

1. First acquisition means for acquiring generation waveform data obtained by non-destructive inspection of a cable sheath of a PC structure and condition data regarding grout filling conditions; Comprising generation means for generating learning waveform data based on the generation waveform data and condition data acquired by the first acquisition means; A learning data generation system for grout filling, characterized by the above.

2. The generation means refers to a generation model learned using learning data with the generation waveform data and condition data as input data and the learning waveform data as output data, and based on the generation waveform data and condition data acquired by the first acquisition means, generates the learning waveform data. The learning data generation system for grout filling according to claim 1, characterized by the above.

3. Second acquisition means for acquiring evaluation waveform data obtained by non-destructive inspection of a cable sheath for evaluating grout filling status; Referring to a restoration model that is generated by machine learning using the learning waveform data generated by the generation means as learning data, limits the frequency having characteristics in the input evaluation waveform data, and outputs restoration data obtained by restoring the evaluation waveform data based on the limited characteristics, and output means for outputting restoration data based on the evaluation waveform data acquired by the second acquisition means; Comprising evaluation means for evaluating the grout filling status based on the restoration data output by the output means and the evaluation waveform data acquired by the second acquisition means. The learning data generation system for grout filling according to claim 1, characterized by the above.

4. The first acquisition means acquires the condition data including configuration data regarding the constituent members of the PC structure. The learning data generation system for grout filling according to claim 1, characterized by the above.

5. The first acquisition means acquires the condition data including shape data regarding the cross-sectional shape of the PC structure. The learning data generation system for grout filling according to claim 1, characterized by the above.

6. The first acquisition means acquires the condition data including depth data regarding the depth from the surface of the PC structure to the cable sheath. The learning data generation system for grout filling according to claim 1, characterized by the above.

7. The first acquisition means acquires the condition data including additional data regarding the cable sheath. The grout filling learning data generation system according to claim 1, characterized by

8. A first acquisition step of acquiring generation waveform data obtained by non-destructive inspection of a cable sheath of a PC structure and condition data regarding grout filling conditions, and Causing a computer to execute a generation step of generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired in the first acquisition step A grout filling learning data generation program, characterized by

9. A first acquisition means for acquiring generation waveform data obtained by non-destructive inspection of a concrete structure and condition data regarding the conditions of the concrete structure, and Comprising a generation means for generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired by the first acquisition means A learning data generation system for concrete void evaluation, characterized by

10. A second acquisition means for acquiring evaluation waveform data obtained by non-destructive inspection of concrete for evaluating voids, and A restoration model that is generated by machine learning using the learning waveform data generated by the generation means as learning data, limits the frequency having characteristics in the input evaluation waveform data, and outputs restoration data obtained by restoring the evaluation waveform data based on the limited characteristics. An output means for outputting restoration data based on the evaluation waveform data acquired by the second acquisition means; and An evaluation means for evaluating the voids of the concrete based on the restoration data output by the output means and the evaluation waveform data acquired by the second acquisition means The learning data generation system for concrete void evaluation according to claim 9, characterized by

11. A first acquisition step of acquiring generation waveform data obtained by non-destructive inspection of a concrete structure and condition data regarding the conditions of the concrete structure, and Causing a computer to execute a generation step of generating learning waveform data based on the conditions based on the generation waveform data and condition data acquired in the first acquisition step A concrete void evaluation learning data generation program, characterized by

Images