Structural diagnostic device, structural diagnostic method, program, and recording medium

Abstract

To provide a structure diagnostic device, a method for diagnosing a structure, a program, and a recording medium that can collect striking sound data effectively and can determine degraded parts and healthy parts highly accurately from striking sound data.SOLUTION: A structure diagnostic device 1 includes: a sound collecting unit 10; a striking sound signal processing unit 22; and a striking sound determination unit 23. The sound collecting unit 10 collects striking sound when a protrusion unit 16 of a rotation impact body 15 strikes the structure. The striking sound signal processing unit 22 analyzes the collected striking sound. The striking sound determination unit 23 analyzes the analyzed striking sound data by a machine learning and diagnoses defects of the structure.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a structure diagnostic device, a structure diagnostic method, a program, and a recording medium. [Background technology]

[0002] If cracks, gaps, peeling, swelling, or other deterioration occurs on the surface or inside of the wall of a building structure or tiles attached to the wall, the tile and its surrounding area may fall off. However, it is very difficult to detect such deterioration just by visually inspecting the surface of the structure. Therefore, a commonly used method is called the hammering method, in which the surface of a structure is struck with an inspection hammer and an expert uses their hearing to distinguish between deteriorated and sound areas based on the subtle differences in the sounds (reverberations). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 3742032 Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, for inspection using the hammering method, an effective hammering tool has been proposed as shown in Patent Document 1, and the efficiency of the hammering work has been improved. However, since the assessment of structural deterioration relies on the tapping and hearing of skilled personnel, securing skilled personnel is a major issue. Furthermore, in recent years, the types of structures to be inspected have been increasing, making it even more difficult to determine the deterioration of a wide variety of structures.

[0005] Therefore, the present invention has been made in consideration of the above-mentioned problems, and an object of the present invention is to provide a structural diagnosis device, a structural diagnosis method, a program, and a recording medium that can efficiently collect hammering sound data and further determine deteriorated areas and sound areas from the hammering sound data with high accuracy. [Means for solving the problem]

[0006] Form 1: One or more embodiments of the present invention propose a structure diagnosis device that strikes a structure and diagnoses defects in the structure from hammering sounds, the structure diagnosis device comprising: a sound collection unit that collects hammering sounds generated when a rotary impactor having a plurality of protrusions arranged on the circumferential surface of a circular rotor strikes the structure as the rotary impactor rotates; a hammering sound signal processing unit that analyzes the hammering sound data collected by the sound collection unit; and a hammering sound determination unit that inputs the hammering sound data analyzed in the hammering sound signal processing unit and determines a diagnosis result by machine learning.

[0007] The structure diagnosis device according to this embodiment includes a sound collection unit, a hammering sound signal processing unit, and a hammering sound determination unit. The sound collection unit collects hammering sounds generated when a protrusion of a rotating impactor strikes a structure. The hammering sound signal processing unit analyzes the collected hammering sounds. The hammering sound determination unit analyzes the analyzed hammering sound data using machine learning to diagnose defects in the structure. According to this configuration, hammering sound data can be collected efficiently, and furthermore, by analyzing the hammering sound data using machine learning, it is possible to determine deteriorated and healthy areas with high accuracy.

[0008] Mode 2: One or more embodiments of the present invention propose a structural diagnostic device that is characterized by having multiple rotating impactors, which are arranged with a certain offset angle so that the protrusions do not strike each other simultaneously. According to this configuration, a plurality of rotary impactors are provided, and the rotary impactors are arranged so that they do not strike the protrusions simultaneously, so that hitting sounds can be picked up with high accuracy. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0009] Mode 3: One or more embodiments of the present invention propose a structural diagnostic device characterized in that the number of microphones in the sound collection section is equal to or less than the number of rotating impactors. According to this configuration, the number of microphones disposed in the sound collection section is equal to or less than the number of rotating impactors, so that the impact sound data can be collected efficiently. Therefore, the number of microphones to be installed can be reduced, which reduces the cost of the structural diagnostic device.

[0010] Mode 4: One or more embodiments of the present invention propose a structural diagnosis device, wherein the hammering sound signal processing unit divides hammering sound data for each strike of a protrusion, and performs frequency analysis for each of the divided hammering sound data. According to this configuration, the impact sound data is divided for each impact and frequency analysis is performed, so that deteriorated and sound areas can be determined with high accuracy.

[0011] Mode 5: One or more embodiments of the present invention propose a structure diagnostic device characterized in that the hammering sound judgment unit uses hammering sound data that has been determined in advance to be good or bad as training data. According to this configuration, the hitting sound determination section uses hitting sound data that has been determined in advance to be good or bad as training data. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0012] Mode 6: One or more embodiments of the present invention propose a structure diagnostic device characterized in that the hammering sound judging unit changes the judging criteria for each type of structure. According to this configuration, the hitting sound determination section changes the determination criteria for each type of structure. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0013] Mode 7: One or more embodiments of the present invention propose a structure diagnosis method executed by a structure diagnosis device that includes a sound collection unit, a hammering sound signal processing unit, and a hammering sound determination unit, and that strikes a structure and diagnoses defects in the structure from the hammering sounds, the structure diagnosis method comprising: a first step in which the sound collection unit collects hammering sounds that are generated when the protrusions strike the structure as the rotating impactor, which has a plurality of protrusions arranged on the circumferential surface of a circular rotor, rotates; a second step in which the hammering sound signal processing unit analyzes the collected hammering sound data; and a third step in which the hammering sound determination unit inputs the analyzed hammering sound data and determines a diagnosis result by machine learning.

[0014] The structure diagnosis method according to this embodiment includes a sound collection unit, a hammering sound signal processing unit, and a hammering sound determination unit. The sound collection unit includes a first step of collecting hammering sounds generated when a protrusion of a rotating impactor strikes a structure. The hammering sound signal processing unit includes a second step of analyzing the collected hammering sounds. The hammering sound determination unit includes a third step of analyzing the analyzed hammering sound data by machine learning and diagnosing defects in the structure. According to this diagnostic method, hammering sound data can be collected efficiently. Furthermore, it is possible to determine deteriorated and healthy areas with high accuracy from the hammering sound data.

[0015] Mode 8: One or more embodiments of the present invention propose a structure diagnosis method, characterized in that in the second step, the hammering sound data is divided for each strike on a protrusion, and frequency analysis is performed for each of the divided hammering sound data. According to this diagnostic method, frequency analysis is performed for each piece of hammering sound data. Therefore, it is possible to determine deteriorated and healthy areas with high accuracy from the hammering sound data.

[0016] Mode 9: In one or more embodiments of the present invention, in the third step, hammering sound data that has been determined to be good or bad in advance is used as training data. According to this configuration, the hammering sound data, which has been determined in advance to be good or bad, is used as training data. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0017] Mode 10: One or more embodiments of the present invention propose a structure diagnostic method characterized in that in the third step, the judgment criteria are changed for each type of structure. According to this configuration, the determination criteria are changed for each structure. Therefore, it is possible to determine deteriorated and healthy areas with high accuracy from the hammering sound data.

[0018] Form 11: One or more embodiments of the present invention propose a program for causing a computer to execute a structure diagnostic method in a structure diagnostic device that includes a sound collection unit, a hitting sound signal processing unit, and a hitting sound determination unit, and that hits a structure and diagnoses defects in the structure from the hitting sounds, the program causing a computer to execute the following steps: a first step in which the sound collection unit collects hitting sounds that are generated when the protrusions hit the structure as the rotating impactor, which has a plurality of protrusions arranged on the circumferential surface of a circular rotor, rotates; a second step in which the hit sound signal processing unit analyzes the collected hitting sound data; and a third step in which the hit sound determination unit inputs the analyzed hitting sound data and determines a diagnosis result by machine learning.

[0019] The program according to this embodiment includes a sound collection unit, a hitting sound signal processing unit, and a hitting sound determination unit. The sound collection unit causes a computer to execute a first step of collecting hitting sounds generated when a protrusion of a rotating impactor strikes a structure. The hitting sound signal processing unit causes the computer to execute a second step of analyzing the collected hitting sounds. The hitting sound determination unit causes the computer to execute a third step of analyzing the analyzed hitting sound data by machine learning and diagnosing defects in the structure.

[0020] This program makes it possible to efficiently collect hammering sound data. Furthermore, a program can be executed on a computer that can determine deteriorated and sound locations with high accuracy from hammering sound data.

[0021] Mode 12: One or more embodiments of the present invention propose a program characterized in that in the second step, the hitting sound data is divided for each hit on a protrusion, and a frequency analysis is performed for each of the divided hitting sound data. According to this configuration, it is possible to cause a computer to execute a program that can determine deteriorated and sound locations with high accuracy from hammering sound data.

[0022] Mode 13: One or more embodiments of the present invention propose a program characterized in that in the third step, hammering sound data that has been determined to be good or bad in advance is used as training data. According to this program, hammering sound data that has been determined to be good or bad in advance is used as training data. Therefore, the computer can be made to determine deteriorated and healthy locations with high accuracy.

[0023] Mode 14: One or more embodiments of the present invention propose a program characterized in that in the third step, the judgment criteria are changed for each type of structure. According to this program, the hammering sound data is judged based on the judgment criteria for each structure. Therefore, regardless of the type of structure, the computer can be made to determine deteriorated and sound areas with high accuracy.

[0024] Form 15: One or more embodiments of the present invention propose a computer-readable non-transitory recording medium having recorded thereon a program for causing a computer to execute a structure diagnostic method in a structure diagnostic device that includes a sound collection unit, a hitting sound signal processing unit, and a hitting sound determination unit, and that hits a structure and diagnoses defects in the structure from the hitting sounds, the recording medium having recorded thereon a program for causing a computer to execute: a first step in which the sound collection unit collects the hitting sounds that are generated when a rotating impactor having a plurality of protrusions arranged on the circumferential surface of a circular rotor strikes the structure as the rotating impactor rotates; a second step in which the hitting sound signal processing unit analyzes the collected hitting sound data; and a third step in which the hitting sound determination unit inputs the analyzed hitting sound data and determines a diagnosis result by machine learning.

[0025] The recording medium has recorded thereon a program that causes a computer to execute the following steps: a first step in which a sound collection unit collects impact sounds generated when a protrusion of a rotating impact body strikes a structure; a second step in which a sound signal processing unit analyzes the collected impact sounds; and a third step in which a sound determination unit analyzes the analyzed impact sound data by machine learning and diagnoses defects in the structure.

[0026] This recording medium allows a computer to execute a program that can determine deteriorated and sound locations with high accuracy from hammering sound data. [Effects of the Invention]

[0027] According to one or more embodiments of the present invention, it is possible to provide a structure diagnosis device, a structure diagnosis method, a program, and a recording medium that can efficiently collect hammering sound data and further determine deteriorated parts and sound parts from the hammering sound data with high accuracy. [Brief explanation of the drawings]

[0028] [Figure 1] 1 is a block diagram showing the configuration of a structural diagnosis device according to a first embodiment of the present invention. [Figure 2]1 is a perspective view illustrating a sound collection unit according to a first embodiment of the present invention. [Figure 3] 1 is a side view illustrating a sound collection unit according to a first embodiment of the present invention. [Figure 4] 4 is a waveform illustrating hitting sound data collected by a sound collection unit according to the first embodiment of the present invention. [Figure 5] 3 is a flowchart relating to a structure diagnostic method according to the first embodiment of the present invention. [Figure 6] FIG. 2 is a diagram illustrating an example of division of hitting sounds processed by the hitting sound signal processing unit according to the first embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0029] [First embodiment] (Configuration of structural diagnostic device 1) The configuration of a structural diagnostic device 1 according to this embodiment will be described below with reference to FIG. As shown in FIG. 1, the structure diagnostic device 1 includes a sound pickup unit 10 and a determination unit 20.

[0030] The sound collection unit 10 collects the hitting sound generated when the structure is struck. The determination unit 20 divides the hammering sounds collected by the sound collection unit 10 into individual hammering sounds and performs signal processing for each of the divided hammering sounds. The signal-processed hammering sound data is analyzed using machine learning, and the deterioration state of the structure is determined based on the analysis results.

[0031] (Configuration of sound pickup unit 10) 1, sound collection unit 10 is provided with sound collection microphones 11a and 11b that collect impact sounds generated when a structure is struck. The impact sounds collected by sound collection microphones 11a and 11b are transferred to data receiving unit 21 of determination unit 20 via data transmitting unit 18.

[0032] Here, the data transmitting unit 18 and the data receiving unit 21 may transmit the hitting sound data from the sound collecting unit 10 to the determining unit 20, and may transmit and receive data using, for example, BT (Bluetooth).

[0033] (Configuration of the determination unit 20) As shown in FIG. 1, the determination unit 20 includes a data receiving unit 21, a hitting sound signal processing unit 22, a hitting sound determination unit 23, a determination criterion learning data storage unit 24, and a determination result display unit 25. The data receiving unit 21 receives the data transmitted from the data transmitting unit 18 and transfers the data to the hitting sound signal processing unit 22. Then, the hitting sound signals collected by the sound collecting microphones 11a and 11b are input to the hitting sound signal processing unit 22, where predetermined signal processing is performed. The hammering sound judgment unit 23 diagnoses the hammering sound data by performing machine learning using hammering sound data that has been determined to be good or bad in advance, which is teacher data used for machine learning stored in the judgment criterion learning data storage unit 24. The judgment result display section 25 displays the diagnosis result of the structure diagnosed by the hammering sound judgment section 23.

[0034] (Structure of the sound pickup unit 10) An example of the structure of the sound collection unit 10 that collects the hitting sounds will be described with reference to FIG. As shown in FIG. 2, the sound collection unit 10 is configured to include at least a sound collection microphone 11, a support rod 12, a frame 13, a rotation shaft 14, and a rotation impactor 15. In this embodiment, two rotary impactors 15 are provided (in the drawing, rotary impactors 15a and 15b).

[0035] The U-shaped frame 13 has bearings 17a and 17b at both ends, and a rotary impactor 15 fixed to a rotary shaft 14 is attached to the bearings. Additionally, sound-collecting microphones 11a and 11b are fixed to the frame 13 to collect the impact sound produced when the rotary impactor 15 strikes the structure.

[0036] Although the frame 13 in this embodiment has a U-shaped configuration, it may be changed to a U-shaped configuration, a V-shaped configuration, or the like. Furthermore, it is sufficient that the sound pickup microphones can reliably pick up the impact sounds, and the number of sound pickup microphones may be less than the number of rotating impactors. Furthermore, the position of the sound pickup microphone is not limited to the positions shown in the embodiment, as long as it can reliably pick up the hitting sound.

[0037] (Structure of the rotating impactor 15) The structure of the rotary impactor 15 will be described in detail with reference to FIG. The rotary impact body 15 is a rotary body in which a rotary disc is fixed to a rotary shaft 14 . The circular rotor has projections 16 arranged at equal intervals on its circumference.

[0038] The rotary impactor 15 has two rotary impactors 15a and 15b fixed to the rotary shaft 14, and furthermore, the rotary impactors 15a and 15b are provided with protrusions 16a and 16b, respectively. At this time, the rotary impactors 15a and 15b have the same shape. Moreover, the protrusion 16 is formed of metal with a rounded tip, for example. Here, the shape and material of the tip of the protrusion 16 are not limited to the above as long as they reliably generate a striking sound and are durable.

[0039] Here, when the rotary shaft 14 rotates and the protrusions 16a and 16b are struck simultaneously, the two striking sounds overlap, making it impossible to collect accurate striking sound data. Therefore, the adjacent rotary impactors 15a and 15b are fixed to the rotary shaft 14 with a certain offset angle θ so that the positions of the protrusions 16a and 16b do not overlap.

[0040] In this embodiment, an example is shown in which eight protrusions are arranged on one rotary impactor, and the angle between the arranged protrusions from the axis is 45 degrees. Here, the number of protrusions is not limited to eight, and may be changed depending on the radius of the circular rotating body, the surface roughness of the structure to be inspected, and the like. Similarly, the deviation angle θ is not a limited angle, and may be changed as long as the angle does not cause the hammering sounds to overlap and the hammering sound diagnosis can be completed within a time period equivalent to the deviation angle θ.

[0041] When collecting the hitting sound, the rotary impactor 15 is arranged so that the protrusion 16 comes into contact with the structure, and the sound collecting part 10 is moved in a direction perpendicular to the surface of the structure relative to the rotation axis 14 of the rotary impactor 15. By moving the sound collection unit 10, the rotary impactor 15 rotates, and the rotation generates continuous striking sounds of the protrusions 16. In this embodiment, by the rotation of the rotary impactor 15 having two rotary impactors, it is possible to collect hitting sound data in two directions at once. The number of rotary impactors is not limited to two, and three or more rotary impactors may be provided.

[0042] FIG. 4 shows the waveform of the impact sound produced by the rotary impactor 15 having the two rotary impactors described above. The timing of occurrence of the hitting sounds in the waveforms of the hitting sounds picked up by the sound pickup microphones 11a and 11b is shifted by the above-mentioned shift angle θ.

[0043] (Explanation of structural diagnostic methods) The structural diagnostic method will be described with reference to FIG. The protrusion 16 arranged on the rotating impactor 15 of the sound collection unit 10 is placed against the surface of the structure to be diagnosed, and the entire sound collection unit 10 is moved while the protrusion 16 is placed in the direction of the support rod 12, and the impact sound is collected by the sound collection microphone 11 (step S1).

[0044] The hitting sound signal collected by the microphone 11 is converted into digital data in the data transmitting unit 18, and the digital data is transferred to the data receiving unit 21 of the determining unit 20. The format of the hammering sound signal converted into digital data can be, for example, a WAVE file. In this embodiment, two sound collection microphones are provided, and therefore, two WAVE files of the hit sound signals collected by the process in step S1 are generated.

[0045] Next, the WAVE file received by the data receiving unit 21 is transferred to the hitting sound signal processing unit 22. Then, the hitting sound signal processing unit 22 divides the multiple hitting sound signals included in the WAVE file into data for each hitting of the protrusion 16 (step S2).

[0046] The division process of the hitting sound data executed in step S2 will be described with reference to FIG. The WAVE file of the impact sound signal collected by the sound collection microphone 11 a contains data of continuous impact sounds of the protrusion 16 accompanying the rotary impactor 15 . As shown in FIG. 6, in the hitting sound signal processing unit 22, a plurality of hitting sound signals included in the hitting sound signal WAVE file are divided into data for each hitting sound signal (hitting sound data 1 to hitting sound data 5 in the drawing). In this embodiment, since two sound collecting microphones are provided, the same processing is also performed on the WAVE file of the hitting sound signal collected by the sound collecting microphone 11b.

[0047] In the present embodiment, the hitting sound data is divided by a method in which a period from the generation of a hitting sound signal to the generation of the next hitting sound signal is divided into one piece of hitting sound data. However, the method for dividing the hitting sound data is not limited to the above method as long as the divided hitting sound data can be used to perform a highly accurate hitting sound analysis.

[0048] Next, the hitting sound signal processing unit 22 applies a window to the divided hitting sound data, and the hitting sound data extracted by the window is Fourier transformed and converted into data in the frequency domain (step S3).

[0049] Furthermore, in the hitting sound signal processing unit 22, waveform amplitude data at a plurality of frequency points is extracted from the converted hitting sound data in the same manner as the teacher data used for machine learning in the hitting sound determination unit 23. The extracted data is then transferred to the hitting sound determination unit 23.

[0050] Next, the hitting sound determination unit 23 performs machine learning using the teacher data stored in the determination criterion learning data storage unit 24, thereby diagnosing the structure (step S4).

[0051] Teacher data learned from structures whose good and bad properties are known in advance is stored in the judgment criterion learning data storage unit 24. The data (teacher data) stored in the judgment criterion learning data storage unit 24 may be replaced with data for each type of structure to be diagnosed. Furthermore, by using data obtained from a structure with predefined defect levels as training data, it is possible to present not only a pass / fail judgment but also a diagnostic result indicating the defect level, such as excellent, good, fair, or poor.

[0052] When replacing the teacher data stored in the judgment criterion learning data storage unit 24 as described above, the teacher data stored in the judgment criterion learning data storage unit 24 may be rewritten, for example, by receiving data from outside the diagnostic device or by reading data from an SD memory or the like.

[0053] Next, the diagnostic information determined by machine learning is transferred to the determination result display unit 25, and the diagnostic information is displayed on the display unit (step S5). The transferred diagnostic information includes information such as the pass / fail determination result and the hammering sound waveform. (Action and effect)

[0054] The structural diagnostic device 1 according to the first embodiment of the present invention includes a sound collection unit 10 that collects impact sounds generated when the protrusions 16 strike a structure as the rotating impactor 15, which has a plurality of protrusions 16 arranged on the circumferential surface of a circular rotor, rotates, a sound signal processing unit 22 that analyzes the impact sound data collected by the sound collection unit 10, and a sound judgment unit 23 that inputs the analyzed impact sound data in the sound signal processing unit 22 and judges the diagnosis result by machine learning.

[0055] The sound collection section 10 collects the impact sound produced when the protrusion 16 of the rotary impactor 15 strikes the structure. The hammering sound signal processing unit 22 divides the collected hammering sound data and performs signal processing. The hammering sound determination unit 23 performs machine learning on the hammering sound data that has undergone signal processing using the training data stored in the determination criterion learning data storage unit 24, performs data analysis, and diagnoses the structure.

[0056] Therefore, the structure diagnosis device 1 can efficiently collect hammering sound data. Furthermore, by analyzing the hammering sound data using machine learning, it is possible to determine with high accuracy which parts of the structure are deteriorated and which parts are sound.

[0057] The structural diagnostic device 1 according to the first embodiment of the present invention is characterized in that it comprises a plurality of rotating impactors 15, which are arranged with a certain offset angle θ so that the protrusions 16 do not strike simultaneously.

[0058] In other words, the plurality of rotary impactors 15 are fixed to the rotary shaft 14 with a certain offset angle θ so that the protrusions 16 of the plurality of rotary impactors 15 do not strike simultaneously. Therefore, with this configuration, it is possible to collect the hitting sounds with high accuracy. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0059] The structure diagnostic device 1 according to the first embodiment of the present invention is characterized in that the number of sound collection microphones 11 arranged in the sound collection section 10 is equal to or less than the number of rotating impactors 15 arranged. According to this configuration, the number of sound collection microphones 11 arranged in the sound collection section 10 is equal to or less than the number of rotating impactors 15 . Therefore, the hammering sound data can be collected efficiently. Furthermore, by reducing the number of sound pickup microphones 11 to be installed, the cost of the structure diagnostic device can be reduced.

[0060] In the structure diagnosis device 1 according to the first embodiment of the present invention, the hammering sound signal processing unit 22 divides the hammering sound data for each strike of the protrusion 16, and performs frequency analysis for each of the divided hammering sound data. According to this configuration, the hitting sound data is divided for each hit and frequency analysis is performed. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0061] The structure diagnostic device 1 according to the first embodiment of the present invention is characterized in that the hammering sound judging section 23 uses hammering sound data that has been determined in advance to be good or bad as training data. According to this configuration, the hitting sound determination unit 23 uses hitting sound data that has been determined in advance to be good or bad as training data. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0062] The structure diagnostic device 1 according to the first embodiment of the present invention is characterized in that the hammering sound judging section 23 changes the judging criteria for each type of structure. According to this configuration, the hitting sound determination unit 23 changes the determination criteria for each type of structure. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0063] The structure diagnostic method according to the first embodiment of the present invention includes a first step in which the sound pickup unit 10 picks up impact sounds generated when the protrusions 16 strike the structure as the rotating impactor 15, which has a plurality of protrusions 16 arranged on the circumferential surface of a circular rotor, rotates; a second step in which the impact sound signal processing unit 22 analyzes the picked-up impact sound data; and a third step in which the impact sound determination unit 23 inputs the analyzed impact sound data and determines the diagnosis result by machine learning.

[0064] Therefore, according to this structural diagnostic method, hammering sound data can be efficiently collected. Furthermore, by analyzing the hammering sound data using machine learning, it is possible to determine with high accuracy which parts of the structure are deteriorated and which parts are sound.

[0065] Furthermore, the structural diagnostic method according to the first embodiment of the present invention is characterized in that it comprises a plurality of rotating impactors 15, which are arranged with a certain offset angle θ so that the protrusions 16 do not strike simultaneously.

[0066] In other words, the plurality of rotary impactors 15 are fixed to the rotary shaft 14 with a certain offset angle θ so that the protrusions 16 of the plurality of rotary impactors 15 do not strike simultaneously. Therefore, according to this structure diagnosis method, it is possible to collect the hitting sounds with high accuracy. Furthermore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0067] The structure diagnostic method according to the first embodiment of the present invention is characterized in that the hammering sound data is divided for each strike of the protrusion 16, and frequency analysis is performed for each of the divided hammering sound data. According to this structural diagnostic method, a frequency analysis is performed for each piece of hammering sound data. Therefore, it is possible to determine deteriorated and healthy areas with high accuracy from the hammering sound data.

[0068] Moreover, in the structure diagnostic method according to the first embodiment of the present invention, hammering sound data that has been determined to be good or bad in advance is used as training data in the third step. According to this structural diagnostic method, hammering sound data that has been determined to be good or bad in advance is used as training data. Therefore, it is possible to determine with high accuracy whether a location is deteriorated or healthy.

[0069] The structure diagnostic method according to the first embodiment of the present invention is characterized in that in the third step, the determination criteria are changed for each type of structure. According to this structure diagnostic method, the judgment criteria are changed for each structure. Therefore, it is possible to determine deteriorated and healthy areas with high accuracy from the hammering sound data.

[0070] The program according to the first embodiment of the present invention is a program for causing a computer to execute a structure diagnostic method in a structure diagnostic device that includes a sound collection unit 10, a hammering sound signal processing unit 22, and a hammering sound determination unit 23, and that strikes a structure to diagnose defects in the structure from the hammering sounds. The program causes a computer to execute: a first step in which the sound collection unit 10 collects hammering sounds generated when the protrusions 16 strike the structure as the rotating impactor 15, which has a plurality of protrusions 16 arranged on the circumferential surface of a circular rotor, rotates; a second step in which the hammering sound signal processing unit 22 analyzes the collected hammering sound data; and a third step in which the hammering sound determination unit 23 inputs the analyzed hammering sound data and determines a diagnosis result by machine learning.

[0071] This program makes it possible to efficiently collect hammering sound data. Furthermore, the computer can be made to determine deteriorated and sound locations with high accuracy from the hammering sound data.

[0072] The program according to the first embodiment of the present invention is characterized in that in the second step, the hitting sound data is divided for each hit of the protrusion 16, and a process of performing a frequency analysis for each of the divided hitting sound data is performed. This program allows a computer to determine deteriorated and sound areas with high accuracy from hammering sound data.

[0073] Moreover, in the program according to the first embodiment of the present invention, in the third step, hitting sound data that has been determined to be good or bad in advance is used as training data. According to this program, hammering sound data that has been determined to be good or bad in advance is used as training data. Therefore, the computer can be made to determine deteriorated and healthy locations with high accuracy.

[0074] The program according to the first embodiment of the present invention is characterized in that in the third step, the determination criteria are changed for each type of structure. According to this program, the hammering sound data is judged based on the judgment criteria for each structure. Therefore, regardless of the type of structure, the computer can be made to determine deteriorated and sound areas with high accuracy.

[0075] The recording medium having recorded thereon a program according to the first embodiment of the present invention is a non-transitory computer-readable recording medium having recorded thereon a program for causing a computer to execute a structure diagnostic method in a structure diagnostic device that includes a sound collection unit 10, a hitting sound signal processing unit 22, and a hitting sound determination unit 23, and that hits a structure and diagnoses defects in the structure from the hitting sounds. The recording medium has recorded thereon a program for causing a computer to execute: a first step in which the sound collection unit 10 collects hitting sounds generated when the protrusions 16 hit the structure as the rotating impactor 15, which has a plurality of protrusions 16 arranged on the circumferential surface of a circular rotor, rotates; a second step in which the hit sound signal processing unit 22 analyzes the collected hitting sound data; and a third step in which the hit sound determination unit 23 inputs the analyzed hitting sound data and determines a diagnosis result by machine learning.

[0076] In other words, it is a recording medium having recorded thereon a program that causes a computer to execute the following steps: a first step in which the sound collection unit 10 collects the impact sound generated when the protrusion 16 of the rotating impact body 15 strikes the structure; a second step in which the impact sound signal processing unit 22 analyzes the collected impact sound; and a third step in which the impact sound determination unit 23 analyzes the analyzed impact sound data by machine learning and diagnoses defects in the structure. This recording medium allows a computer to execute a program that can determine deteriorated and sound locations with high accuracy from hammering sound data. [Other embodiments] The present invention is not limited to the above-described embodiment, and various modifications are possible without departing from the spirit and scope of the present invention.

[0077] The distance between the two rotating impactors 15a and 15b arranged in the sound collection unit 10 may be changed based on the size and arrangement of tiles or the like attached to the surface of the structure to be diagnosed. For example, by preparing several types of sound collection units 10 with different intervals between rotating impactors and exchanging the rotating impactors to match the surface of the structure to be diagnosed, it is possible to respond to changes in the surface of the structure.

[0078] Also, although not shown, the structure may be such that when the sound collection unit 10 is moved along the structure, the distance between the two rotating impactors repeatedly becomes closer and farther apart in accordance with the amount of movement of the sound collection unit 10. This structure can accommodate a wide range of changes in the surface of the structure.

[0079] Furthermore, a movable body with a controllable movement to which the sound collection unit 10 is attached may move on the surface of the structure to collect the hitting sound. With this configuration, it is possible to diagnose a structure by moving the device over the surface of the structure, such as a high place where people cannot directly pick up sound. Furthermore, by storing data relating the position information of the moving body to the diagnosis results, it is possible to display not only the presence or absence of defects but also the defective locations of the structure superimposed on a diagram of the structure.

[0080] As mentioned above, the judgment criteria learning data is changed for each type of structure to be diagnosed. However, even for the same structure, the impact sounds will change depending on whether it is wet due to rain or dry. Therefore, the training data may be changed to suit not only the type of structure but also the diagnosis environment.

[0081] The structure diagnosis device 1 of the present invention can be realized by recording the processing of the structure diagnosis device 1 on a recording medium that can be read by a computer system, and having the structure diagnosis device 1 read and execute the program recorded on this recording medium. The computer system here includes hardware such as an OS and peripheral devices.

[0082] "Computer system" also includes the homepage provision environment (or display environment) if the WWW (World Wide Web) system is used. The above program may be transmitted from a computer system that stores the program in a storage device or the like to another computer system via a transmission medium or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) like the Internet or a communication line (communication line) like a telephone line.

[0083] The program may also be a program for implementing some of the above-mentioned functions, or may be a so-called differential file (differential program) that can implement the above-mentioned functions in combination with a program already stored in the computer system. [Explanation of symbols]

[0084] 1. Structural diagnostic equipment 10; Sound pickup section 11. Sound pickup microphone 12;Support rod 13;frame 14; Rotation axis 15; Rotating impact body 16; Protrusion 17; bearing 18; Data transmission section 20: Hitting sound detection section 21: Data receiving section 22: Hitting signal processing section 23; Judgment section 24: Criteria learning data storage unit 25;Judgment result display section

Claims

1. A structure diagnostic device that strikes a structure and diagnoses defects in the structure from striking sounds, a sound pickup unit that picks up impact sounds generated when a rotary impactor having a plurality of protrusions disposed on the circumferential surface of a circular rotor strikes the structure as the rotary impactor rotates; a hitting sound signal processing unit that analyzes the hitting sound data collected by the sound collection unit; a hammering sound determination unit that receives the analyzed hammering sound data in the hammering sound signal processing unit and determines a diagnosis result by machine learning; Equipped with A structural diagnostic device comprising a plurality of the rotary impactors, the rotary impactors being arranged with a certain angular offset so as not to strike the protrusions simultaneously.

2. 2. A structural diagnostic device according to claim 1, wherein the number of microphones provided in said sound collection section is equal to or less than the number of said rotating impactors provided.

3. 3. The structural diagnosis device according to claim 1, wherein the hammering sound signal processing unit divides the hammering sound data for each strike of the protrusion, and performs frequency analysis for each of the divided hammering sound data.

4. 4. The structural diagnostic device according to claim 1, wherein the hammering sound determination unit uses the hammering sound data, which has been determined in advance to be good or bad, as training data.

5. 5. The structure diagnostic device according to claim 1, wherein the hammering sound determination unit changes the determination criteria for each type of the structure.

Images