Hammering sound inspection device, hammering sound inspection method, and hammering sound inspection program

Abstract

To provide a hammering sound inspection device, a hammering sound inspection method, and a hammering sound inspection program capable of obtaining a precise result by less prior preparation than a conventional method.SOLUTION: A hammering sound inspection device 1 comprises: an acquisition section 11 for acquiring waveform data of a hammering sound of an inspection object; a first calculation section 12 for calculating a frequency spectrum by performing Fourier transform of waveform data; a second calculation section 13 for calculating a mel spectrum obtained by performing dimensional compression of a frequency spectrum by a mel scale simulating characteristics of a human pitch perception; a storage section 14 for storing a comparison mel spectrum calculated in advance from a hammering sound of a sound inspection object; and an output section 17 for outputting a comparison result between an unknown mel spectrum calculated from a hammering sound of an inspection object whose soundness is unknown and a comparison mel spectrum.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a hammering inspection device, a hammering inspection method, and a hammering inspection program. [Background technology]

[0002] Tile materials are used as exterior finishing materials for buildings. Tiles can become loose, crack, or peel over time. Therefore, tapping tests are performed to check for tile peeling, in which the tiles are tapped with a tapping rod.

[0003] In the tile deterioration diagnosis method described in Patent Document 1, the deterioration rate of a tile is calculated by calculating the ratio of the time during which a percussion sound from a normal part, which indicates that the tile is not deteriorated, occurs to the time during which a percussion sound from an abnormal part, which indicates that the tile is deteriorated, occurs during a specified analysis time of the recorded percussion sound. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2020-153938 Summary of the Invention [Problem to be solved by the invention]

[0005] Incidentally, in the tile deterioration diagnosis method described in Patent Document 1, three types of patterns are calculated in advance: a reference frequency pattern indicating a normal portion, a reference frequency pattern indicating a base lifted portion, and a reference frequency pattern indicating a tile lifted portion, and then matching is performed with the reference frequency patterns. Therefore, three types of reference frequency patterns must be calculated in advance for each tile, which makes advance preparations complicated. Note that the same problem exists not only for tiles, but also for any object to be inspected that undergoes a hammering test. [Means for solving the problem]

[0006] A hammering inspection device that solves the above-mentioned problems includes an acquisition unit that acquires waveform data of hammering sounds of an object to be inspected, a first calculation unit that performs a Fourier transform on the waveform data to calculate a frequency spectrum, a second calculation unit that calculates a mel spectrum by dimensionally compressing the frequency spectrum using a mel scale that simulates the characteristics of human pitch perception, a storage unit that stores a comparison mel spectrum that is calculated in advance from hammering sounds of an object to be inspected that is healthy, and an output unit that outputs a comparison result between the unknown mel spectrum calculated from hammering sounds of an object to be inspected, whether or not it is healthy.

[0007] According to the above configuration, the condition of the object to be inspected can be quantified by calculating a Mel spectrum from the waveform data of the hammering sounds of the object to be inspected and outputting the comparison result between the unknown Mel spectrum and the comparative Mel spectrum. Furthermore, by calculating the comparative Mel spectrum from the hammering sounds of a healthy object to be inspected, highly accurate results can be obtained with less advance preparation than conventional methods.

[0008] In the hammering inspection device, the comparison result is preferably a degree of abnormality obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum.

[0009] The hammering inspection device preferably includes a determination unit that determines that the object to be inspected is sound when the degree of abnormality is smaller than a predetermined value, and that the object to be inspected is unsound when the degree of abnormality is equal to or greater than the predetermined value.

[0010] A hammering inspection method that solves the above-described problems is a hammering inspection method in which a computer inspects hammering sounds of an object to be inspected, and includes an acquisition process that acquires waveform data of the hammering sounds of the object to be inspected, a first calculation process that performs a Fourier transform on the waveform data to calculate a frequency spectrum, a second calculation process that calculates a mel spectrum by dimensionally compressing the frequency spectrum using a mel scale that simulates the characteristics of human pitch perception, a storage process that stores a comparison mel spectrum that has been calculated in advance from hammering sounds of an object to be inspected that is healthy, and an output process that outputs a comparison result between an unknown mel spectrum calculated from hammering sounds of an object to be inspected, the unknown mel spectrum being calculated from hammering sounds of an object to be inspected that is healthy or unknown.

[0011] According to the above method, the condition of the object to be inspected can be quantified by calculating a Mel spectrum from the waveform data of the hammering sounds of the object to be inspected and outputting the comparison result between the unknown Mel spectrum and the comparative Mel spectrum. Furthermore, by calculating the comparative Mel spectrum from the hammering sounds of a healthy object to be inspected, highly accurate results can be obtained with less advance preparation than conventional methods.

[0012] In the hammering test method, the comparison result is preferably a degree of abnormality obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum.

[0013] The hammering inspection method preferably includes a determination process of determining that the object to be inspected is healthy when the degree of abnormality is smaller than a predetermined value, and determining that the object to be inspected is unhealthy when the degree of abnormality is equal to or greater than the predetermined value.

[0014] A hammering test program that solves the above-described problems is a hammering test program for testing hammering sounds of an object to be tested, and causes a computer to execute the following steps: an acquisition process for acquiring waveform data of the hammering sounds of the object to be tested; a first calculation process for calculating a frequency spectrum by Fourier transforming the waveform data; a second calculation process for calculating a mel spectrum by dimensionally compressing the frequency spectrum using a mel scale that simulates the characteristics of human pitch perception; a storage process for storing a comparative mel spectrum that has been calculated in advance from hammering sounds of an object to be tested that is healthy; and an output process for outputting a comparison result between an unknown mel spectrum calculated from hammering sounds of an object to be tested, the object being healthy or unknown.

[0015] According to the above program, the condition of the object to be inspected can be quantified by calculating a Mel spectrum from waveform data of the hammering sounds of the object to be inspected and outputting the comparison result between the unknown Mel spectrum and the comparative Mel spectrum. Furthermore, by calculating the comparative Mel spectrum from hammering sounds of a healthy object to be inspected, highly accurate results can be obtained with less advance preparation than conventional methods.

[0016] In the hammering test program, the comparison result is preferably a degree of abnormality obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum.

[0017] With regard to the hammering test program, it is preferable to cause the computer to execute a determination process in which the object to be tested is determined to be healthy when the degree of abnormality is smaller than a predetermined value, and the object to be tested is determined to be unhealthy when the degree of abnormality is equal to or greater than the predetermined value. [Effects of the Invention]

[0018] According to the present invention, highly accurate results can be obtained with less prior preparation than conventional methods. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 10 is an image diagram showing a hammering test. [Figure 2] 1 is a block diagram showing a schematic configuration of an embodiment of a hammering inspection device. [Figure 3] FIG. 10 is a diagram showing waveform data of hammering sounds in the hammering inspection method of the embodiment. [Figure 4] FIG. 4 is a diagram showing an amplitude spectrum of hammering sounds in the hammering inspection method according to the embodiment. [Figure 5] FIG. 2 is a diagram showing a Mel filter bank in the hammering test method according to the embodiment. [Figure 6] FIG. 2 is a diagram showing a Mel spectrum of hammering sounds obtained in the hammering inspection method according to the embodiment. [Figure 7] FIG. 10 is a diagram showing learning data of hammering sounds in the hammering inspection method of the embodiment. [Figure 8] FIG. 2 is a diagram showing a Mel spectrum of hammering sounds obtained in the hammering inspection method according to the embodiment. [Figure 9] FIG. 10 is a diagram showing a comparison between the Mel spectrum of hammering sounds and learning data in the hammering inspection method of the embodiment. [Figure 10] 4 is a flowchart showing a hammering test method according to the embodiment. [Figure 11] 4 is a flowchart showing a hammering test method according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0020] (Present embodiment) Hereinafter, an embodiment of a hammering inspection device, a hammering inspection method, and a hammering inspection program will be described with reference to FIGS.

[0021] As shown in Figure 1, the object to be inspected in a hammering test is a tile 2 that constitutes the exterior wall of a building. The test is carried out based on the hammering sound, which is the sound produced when an inspector taps the tile 2 with a tapping rod 3. The hammering test is used to check for peeling of the tile 2. Peeling of the tile 2 can be tile loosening, which is peeling between the tile and the applied mortar, or base loosening, which is peeling between the base material and the base mortar.

[0022] (Hit sound inspection device 1) As shown in FIG. 2, the hammering inspection device 1 includes an information processing device 10, a microphone 20, a display device 21, and a speaker 22. The microphone 20, the display device 21, and the speaker 22 are connected to the information processing device 10. The microphone 20 collects hammering sounds, converts the collected hammering sounds into electrical signals, and outputs the electrical signals to the information processing device 10. It is desirable to install the microphone 20 in a location that does not interfere with the work of an inspector, such as near the chest. The display device 21 displays the inspection method, the inspection results, etc. The speaker 22 provides audio guidance on the inspection method and outputs the inspection results as sound. The hammering inspection device 1 may be a dedicated device.

[0023] The information processing device 10 may be configured as one or more processors that execute various processes according to a computer program (software). The processes executed by the information processing device 10, i.e., the processor, include a hammering test method. The hammering test method includes an acquisition process, a first calculation process, a second calculation process, a storage process, an output process, and a judgment process, which will be described later. Note that the information processing device 10 may also be configured as a circuit including one or more dedicated hardware circuits, such as an application-specific integrated circuit (ASIC), or a combination thereof, that execute at least some of the various processes. The processor includes a CPU and memories such as RAM and ROM. The memory stores program code or instructions configured to cause the CPU to execute the processes. The memory, i.e., a computer-readable medium, includes any available medium accessible by a general-purpose or dedicated computer. The program stored in the computer-readable medium includes a hammering test program. The hammering test program causes a computer to execute an acquisition process, a first calculation process, a second calculation process, a storage process, an output process, and a judgment process.

[0024] The information processing device 10 includes an acquisition unit 11, a first calculation unit 12, a second calculation unit 13, a storage unit 14, a comparison unit 15, a determination unit 16, and an output unit 17. The acquiring unit 11 acquires waveform data of the hammering sounds of the test object. As shown in Fig. 3, the waveform data of the hammering sounds of the test object has time on the horizontal axis and sound pressure on the vertical axis.

[0025] The first calculation unit 12 calculates a frequency spectrum by Fourier transforming the waveform data acquired by the acquisition unit 11. As shown in Fig. 4, in the frequency spectrum, the horizontal axis represents frequency. The second calculation unit 13 calculates a mel spectrum by compressing the frequency spectrum into dimensions using a mel scale that simulates the characteristics of human pitch perception.

[0026] It is known that human hearing has high frequency resolution in the low range and low frequency resolution in the high range. The Mel scale was designed to mimic this characteristic of human pitch perception. For example, the formula used to convert frequency f to the Mel scale mel is the one used in the speech recognition toolkit called HTK (Hidden Markov Model Toolkit), defined in equation (1). The Mel scale allows for dimensionality reduction based on hearing characteristics.

[0027]

number

[0028] The Mel spectrum is a frequency spectrum converted to the Mel scale. The Mel spectrum M can be calculated from the product of the Mel filter bank F and the frequency spectrum X as shown in equations (2) and (3). Here, i represents a Mel bin and f represents a frequency bin. Equation (3) shows the value of the j-th Mel spectrum M of any of the i Mel bins.

[0029]

number

[0030]

number

[0031] A Mel filter bank is a collection of bandpass filters configured with triangular windows in the frequency domain. Figure 5 shows a Mel filter bank. The center frequencies of each filter are designed to be equally spaced on the Mel scale converted using equation (1). This allows for detailed detection of low-frequency changes and rough detection of high-frequency changes. To align the scale of each filter's output relative to the input, the area of ​​each filter is normalized to 1.

[0032] By converting the frequency spectrum X into the Mel spectrum M, it is possible to simulate the characteristics of human pitch perception. At the same time, the number of dimensions of the frequency spectrum X can be compressed to the number of Melvin dimensions. Figure 6 shows the Mel spectrum M. The dashed line indicates the frequency spectrum X.

[0033] The memory unit 14 compares the Mel spectrum M calculated in advance from the hammering sounds of a sound test object in which no peeling has occurred with the Mel spectrum M c As shown in Figure 7, the comparison Mel spectrum M c is a set of Mel spectra M calculated from hammering sounds of multiple healthy test objects. The solid line shows the comparison Mel spectrum M c The average value of

[0034] The second calculation unit 13 calculates the Mel spectrum M of n hit sounds of a healthy tile (healthy tile). Then, the sample mean of each Melvin is calculated from equation (4), and the unbiased standard deviation of each Melvin is calculated from equation (5). M k,j Equation (4) represents the value of the j-th mel spectrum M of the k-th i-th melvin among n mel spectra M. Equation (4) represents the average value of the n mel spectrum M of the j-th melvin. Similarly, equation (5) represents the unbiased standard deviation of the n mel spectrum M of the j-th melvin.

[0035]

number

[0036]

number

[0037] As shown in Fig. 8, when inspecting a tile (unknown tile) that is not known to be sound or not, the unknown tile is struck with a tapping rod 3, and a Mel spectrum (unknown Mel spectrum) is calculated from the hammering sounds. In detail, an acquisition unit 11 acquires waveform data from a microphone 20, a first calculation unit 12 calculates a frequency spectrum X indicated by the dashed line from the waveform data, and a second calculation unit 13 calculates a Mel spectrum M indicated by the solid line from the frequency spectrum X. The Mel spectrum M obtained from the hammering sounds of the unknown tile is called the unknown Mel spectrum M'.

[0038] The comparison unit 15 compares the unknown Mel spectrum M' with the comparison Mel spectrum M c That is, the comparison unit 15 compares the unknown Mel spectrum M' with the comparison Mel spectrum M c The degree of anomaly is obtained from the sum of the absolute values ​​of the Z score z at each frequency. The degree of anomaly is the result of the comparison. The Z score z is a value converted so that the relative position of each individual in the set can be determined. The Z score z is also a value converted so that the average is "0" and the standard deviation is "1". As shown in Figure 9, in the case of an unhealthy tile, the comparison Mel spectrum M c There will be many areas that are out of alignment.

[0039] The comparison unit 15 calculates the Z score z of each Melvin using equation (6).

[0040]

number

[0041] The comparison unit 15 calculates the absolute values ​​of the Z scores z, and adds them up to obtain the abnormality level a(M') using equation (7).

[0042]

number

[0043] The determination unit 16 determines that the object to be inspected is healthy when the degree of abnormality is smaller than a predetermined value, and determines that the object to be inspected is unhealthy when the degree of abnormality is equal to or greater than the predetermined value. th The following judgment is made based on the predetermined value a th may be stored in advance in the storage unit 14, or may be set by an inspector at the time of inspection.

[0044] a(M')≧a th Then, M' is abnormal. a(M') th Then, M' is normal The output unit 17 outputs the determination result obtained from the abnormality degree a(M'). The output unit 17 displays the determination result on the display device 21, or outputs the result as sound from the speaker 22. Note that only one of these may be used.

[0045] (Action of this embodiment) Next, the processing procedure of the hammering inspection method will be described with reference to FIGS.

[0046] First, healthy tiles that are free from peeling are learned as shown in Figure 10. Learning of healthy tiles is performed at the inspection site. The hammering inspection device 1 acquires waveform data of hammering sounds of a sound tile (step S11). That is, the acquisition unit 11 acquires waveform data of hammering sounds collected by the microphone 20. Step S11 corresponds to an acquisition process.

[0047] The hammering inspection device 1 calculates a frequency spectrum (step S12). That is, the first calculation unit 12 calculates a frequency spectrum by Fourier transforming the waveform data acquired by the acquisition unit 11. Step S12 corresponds to the first calculation process.

[0048] ​The hammering inspection device 1 calculates a Mel spectrum M (step S13). The second calculation unit 13 calculates a Mel spectrum M by compressing the dimension of the frequency spectrum X calculated by the first calculation unit 12 on the Mel scale. Step S13 corresponds to the second calculation process.

[0049] The hammering test device 1 is a comparative Mel spectrum M c The storage unit 14 stores the Mel spectrum M calculated in advance from the hitting sounds of a plurality of healthy tiles as the comparison Mel spectrum M c Step S14 corresponds to a storage process.

[0050] Next, we examine the unknown tiles, which may be healthy or unknown, as shown in FIG. The hammering inspection device 1 acquires waveform data of hammering sounds of the unknown tile (step S21). That is, the acquisition unit 11 acquires waveform data of hammering sounds collected by the microphone 20. Step S21 corresponds to an acquisition process.

[0051] The hammering inspection device 1 calculates the frequency spectrum X (step S22). That is, the first calculation unit 12 performs a Fourier transform on the waveform data acquired by the acquisition unit 11 to calculate the frequency spectrum X. Step S22 corresponds to the first calculation process.

[0052] The hammering inspection device 1 calculates an unknown Mel spectrum M' (step S23). The second calculation unit 13 calculates a Mel spectrum M by dimensionally compressing the frequency spectrum X calculated by the first calculation unit 12 on the Mel scale. The Mel spectrum M calculated from the hammering sounds of the unknown tile is defined as the unknown Mel spectrum M'. Step S23 corresponds to the second calculation process.

[0053] The hammering test device 1 measures the unknown Mel spectrum M' and the comparison Mel spectrum M c That is, the determination unit 16 compares the unknown Mel spectrum M' with the comparison Mel spectrum M c The degree of abnormality a(M') is obtained from the average value of the absolute values ​​of the Z scores at each frequency. Step S24 corresponds to a comparison process.

[0054] The hammering inspection device 1 judges the comparison result (step S25). That is, the judgment unit 16 judges whether the abnormality degree a(M') is equal to or greater than the predetermined value a th When the anomaly degree a(M') is smaller than the predetermined value a, the unknown tile is judged to be healthy. th If the above is true, the unknown tile is determined to be unsound. Step S25 corresponds to the determination process.

[0055] The hammering inspection device 1 outputs the determination result (step S26). That is, the output unit 17 outputs the determination result of the determination unit 16 to the display device 21 and the speaker 22. Step S26 corresponds to the output process.

[0056] (Effects of this embodiment) Next, the effects of this embodiment will be described. (1) Calculate the Mel spectrum M from the waveform data of the tapping sound of the tile to be inspected, and compare it with the unknown Mel spectrum M'. c By outputting the comparison results with the tile, it is possible to quantify the peeling state of the tile being inspected. c By calculating this, highly accurate results can be obtained with less advance preparation than with conventional methods.

[0057] (2) The result of the comparison is the degree of anomaly a(x') obtained from the sum of the absolute values ​​of the Z scores. Therefore, the deviation of the calculated Mel spectrum M from the mean can be evaluated on the scale of standard deviation.

[0058] (3) Abnormality level a(M') and predetermined value a th The system determines whether the tile being inspected is sound or unsound by comparing it with the actual tile. Therefore, the inspector can understand the condition of the tile being inspected simply by looking at the judgment result.

[0059] (Other embodiments) The above embodiment can be modified as follows: The above embodiment and the following modifications can be combined with each other within the scope of technical compatibility.

[0060] In the above embodiment, the display device 21 may be a bracelet-type display device. In the above embodiment, the display device 21 may be a lighting unit provided on the tip or handle of the tapping rod 3. The inspection result may be displayed by the color or lighting manner of the lighting unit.

[0061] In the above embodiment, an information processing device such as a smartphone or tablet terminal may operate according to a hammering test program and function as a hammering test device. Also, eyeglasses such as smart glasses may collect hammering sounds and operate according to a hammering test program to function as a hammering test device.

[0062] In the above embodiment, the hammering test was performed by the information processing device 10 provided in the hammering test device 1. However, the hammering sounds collected by the microphone 20 may be transmitted to a server via a network, and the information processing device of the server may operate according to a hammering test program to function as the hammering test device.

[0063] In the above embodiment, the hammering inspection device 1 is provided with a microphone 20. However, as long as waveform data of the hammering sounds of the inspection object can be acquired, the hammering inspection device 1 does not need to be provided with a microphone. Also, the hammering inspection device 1 is provided with a display device 21 and a speaker 22. However, as long as inspection results can be output, the hammering inspection device 1 does not need to be provided with a display device and a speaker.

[0064] In the above embodiment, the comparison result is the degree of abnormality obtained from the sum of the absolute values ​​of the Z scores. However, the comparison result may be the degree of abnormality obtained from the average value of the absolute values ​​of the Z scores. In the above embodiment, in the learning of healthy tiles, a plurality of Mel spectra M calculated from the tapping sounds of a plurality of healthy tiles are compared. cHowever, in the learning of healthy tiles, one Mel spectrum M calculated from the hitting sound of one healthy tile is compared with the Mel spectrum M c For example, n=1 may be used in equations (4) and (5).

[0065] In the above embodiment, learning of healthy tiles is performed at the site of inspection, but learning data may be stored in advance. Then, a comparison Mel spectrum M selected from the stored learning data is c may be compared by

[0066] In the above embodiment, the determination unit 16 determines whether the object to be inspected is healthy or unhealthy by comparing the degree of abnormality with a predetermined value. However, the determination unit 16 may be omitted. In this case, the output unit 17 outputs the unknown Mel spectrum M' calculated from the hammering sounds of the object to be inspected, whether healthy or unhealthy, and the comparison Mel spectrum M c The comparison result is output.

[0067] In the above embodiment, the tapping sound is generated by tapping the test object with the tapping rod 3. However, the tapping sound may be generated by rolling the tapping rod 3 over the surface of the test object. In the above embodiment, the tapping sound is generated by the tapping rod 3. However, the tapping sound may be generated by any other means other than the tapping rod 3, as long as the tapping sound is generated from the object to be inspected.

[0068] In the above embodiment, the object to be inspected is the tile 2. However, the object to be inspected may be an exterior wall member such as a concrete wall or concrete block, or a mounting member such as a bolt. [Explanation of symbols]

[0069] 1...Heat testing device 2...Tile to be inspected 3...Sounding rod 10...Information processing device 11…Acquisition part 12...1st calculation section 13…Second calculation section 14...Storage section 15...Comparison section 16…Judgment section 17...Output section 20...Mike 21...Display device 22...Speaker

Claims

1. an acquisition unit that acquires waveform data of hammering sounds of the test object; a first calculation unit that performs a Fourier transform on the waveform data to calculate a frequency spectrum; a second calculation unit that calculates a mel spectrum by compressing the frequency spectrum in dimensions using a mel scale that simulates the characteristics of human pitch perception; a storage unit that stores a comparative Mel spectrum calculated in advance from hammering sounds of a healthy test object; an output unit that outputs a comparison result between an unknown Mel spectrum calculated from the hammering sound of an object to be inspected, whether the object is healthy or unknown, and the comparative Mel spectrum. Hammering inspection device.

2. The comparison result is an abnormality degree obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum. The hammering inspection device according to claim 1.

3. a determination unit that determines that the inspection object is healthy when the degree of abnormality is smaller than a predetermined value, and that the inspection object is unhealthy when the degree of abnormality is equal to or greater than the predetermined value; The hammering inspection device according to claim 2.

4. A hammering inspection method in which a computer inspects hammering sounds of an object to be inspected, an acquisition process for acquiring waveform data of the hammering sound of the test object; a first calculation process of calculating a frequency spectrum by Fourier transforming the waveform data; a second calculation process for calculating a mel spectrum by compressing the frequency spectrum in dimensions using a mel scale that simulates the characteristics of human pitch perception; a storage process for storing a comparative Mel spectrum calculated in advance from the hammering sounds of a healthy test object; and an output process for outputting a comparison result between an unknown Mel spectrum calculated from the hammering sound of an object to be inspected, whether healthy or unknown, and the comparative Mel spectrum. Hammering test method.

5. The comparison result is an abnormality degree obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum. The hammering inspection method according to claim 4.

6. and determining that the object to be inspected is healthy when the degree of abnormality is smaller than a predetermined value, and determining that the object to be inspected is unhealthy when the degree of abnormality is equal to or greater than the predetermined value. The hammering inspection method according to claim 5.

7. A hammering inspection program for inspecting hammering sounds of an object to be inspected, an acquisition process for acquiring waveform data of the hammering sound of the test object; a first calculation process of calculating a frequency spectrum by Fourier transforming the waveform data; a second calculation process for calculating a mel spectrum by compressing the frequency spectrum in dimensions using a mel scale that simulates the characteristics of human pitch perception; a storage process for storing a comparative Mel spectrum calculated in advance from the hammering sounds of a healthy test object; and outputting a comparison result between an unknown Mel spectrum calculated from the hammering sound of an object to be inspected, whether healthy or unknown, and the comparative Mel spectrum. Hammering test program.

8. The comparison result is an abnormality degree obtained from the sum of absolute values ​​of Z scores or the average value of absolute values ​​of Z scores at each frequency of the unknown Mel spectrum and the comparison Mel spectrum. The hammering test program according to claim 7.

9. The computer is caused to execute a determination process of determining that the inspection object is healthy when the degree of abnormality is smaller than a predetermined value, and determining that the inspection object is unhealthy when the degree of abnormality is equal to or greater than the predetermined value. The hammering test program according to claim 8.

Images