Acoustic analysis equipment, acoustic analysis methods, acoustic analysis programs

Abstract

To analyze hitting sound in consideration of environmental sounds, such as ambient noise and noise.SOLUTION: An acoustic analyzer 200 for extracting a sound to be analyzed in order to evaluate the soundness of a structure includes: a data acquisition part 201 for acquiring sound pressure data generated when checking the percussion of a wall surface; a hitting sound specific part 203 for dividing the sound pressure data included in a time to be analyzed in the acquired sound pressure data into a plurality of first division segments larger than the interval of a hitting sound to specify the maximum absolute value of each first division segment as the hitting sound; and a representative value calculation part 207 for calculating the representative value of the hitting sound in the time to be analyzed.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] This invention relates to an acoustic analysis device, an acoustic analysis method, and an acoustic analysis program for extracting sounds to be analyzed for the purpose of evaluating the structural integrity of a structure.

[0002] Traditionally, when mortar, tiles, or other materials attached to or adhered to the walls of buildings peel, deteriorate, expand, or become contaminated with air or foreign matter, the affected parts can fall, potentially causing accidents. Therefore, periodic inspections are conducted. The inspection method involves using a hammer-shaped inspection tool (striking rod) to tap the wall of the building and determining the presence or absence of peeling, deterioration, expansion, air or foreign matter contamination of the mortar, etc., based on the reverberation sound. This method is generally called the percussion method.

[0003] When inspection points are at high altitudes, a lifting gondola is sometimes suspended from the roof of the building, and workers riding in the gondola strike the surface with an inspection hammer to perform the inspection. However, in recent years, devices have been developed that allow for inspection without the need for scaffolding or other equipment for human direct inspection, such as robots equipped with a motor-driven rotating striking rod and a mechanism that moves along the wall surface on wheels while maintaining close contact with the wall surface using a pressure exhaust fan, striking the exterior wall tiles, or devices that attach a motor-driven rotating striking rod to a telescopic rod.

[0004] Traditionally, determining the soundness of materials by percussion often involved sensory testing, where a person would empirically assess the sound by listening to the percussion. However, there are also discussions about using alternative algorithms to perform this assessment, thereby enabling inspections that are not dependent on the skill level of the assessor.

[0005] For example, Patent Document 1 discloses a technique for tapping inspection of concrete structures, in which a polygonal head is applied to the surface of the concrete structure and moved, and the change in the intensity of the tapping sound with respect to the frequency of the tapping sound generated at that time is analyzed, and a technique is disclosed to determine whether or not the wall surface is sound according to the intensity of the tapping sound.

[0006] Furthermore, Patent Document 2 discloses a technique for inspecting concrete structures by tapping, which focuses on the fact that the attenuation of tapping sounds is reduced in the delamination area. The technique determines whether or not the wall surface is sound based on the timing at which the sound pressure of the tapping sound detected by a microphone exceeds a threshold, and the elapsed time since the threshold was exceeded.

[0007] Furthermore, Patent Document 3 discloses a technique for determining whether a wall surface is sound by calculating the area of ​​the sound pressure waveform in multiple regions and comparing the waveform area of ​​a specific part with the average waveform area. [Prior art documents] [Patent Documents]

[0008] [Patent Document 1] Japanese Patent Publication No. 2013-253947 [Patent Document 2] Japanese Patent Publication No. 2007-132720 [Patent Document 3] Japanese Patent Publication No. 2016-080592 [Overview of the project] [Problems that the invention aims to solve]

[0009] However, none of the patent documents 1, 2, and 3 take into account the presence of ambient noise such as noise and static. Therefore, when performing a tapping inspection of a wall surface using a robot or other device, the tapping sound includes sounds other than the tapping sound, such as mechanical noise and static, making analysis based solely on the tapping sound difficult.

[0010] This invention has been made in view of these circumstances, and aims to provide an acoustic analysis device, an acoustic analysis method, and an acoustic analysis program that can perform percussion sound analysis while taking into account environmental sounds such as noise and static. [Means for solving the problem]

[0011] (1) In order to achieve the above objective, the present invention employs the following means. Specifically, the acoustic analysis device of the present invention is an acoustic analysis device for extracting sounds to be analyzed for structural integrity evaluation, and is characterized by comprising: a data acquisition unit that acquires sound pressure data generated during a tapping inspection of a wall surface; a sound identification unit that divides the sound pressure data included in the analysis target time from the acquired sound pressure data into a plurality of first division sections larger than the tapping interval, and identifies the maximum absolute value of each of the first division sections as the tapping sound; and a representative value calculation unit that calculates a representative value of the tapping sound during the analysis target time. In this way, by dividing into sections larger than the tapping interval and identifying the maximum absolute value as the tapping sound, the tapping sound produced by the tapping tool can be accurately extracted.

[0012] (2) Furthermore, in the acoustic analysis apparatus described in (1) above, the first division interval is characterized in that it is between 2 and 4 times the interval between impact sounds. By specifying the first division interval in this way, even if the interval between impact sounds is irregular, the impact sounds can be included in each first division interval in most cases.

[0013] (3) In addition, in the acoustic analysis apparatus described in (1) or (2) above, the representative value calculation unit is characterized in that it calculates a representative value of the impact sound using a numerical value within a predetermined range from among the impact sounds. By calculating a representative value of the impact sound using a numerical value within a predetermined range from among the impact sounds in this way, intervals that do not contain impact sounds can be excluded from the analysis target.

[0014] (4) Furthermore, the acoustic analysis apparatus described in any of (1) to (3) above is further equipped with an ambient sound identification unit that divides the sound pressure data included in the analysis target time from the acquired sound pressure data into a plurality of second division intervals smaller than the interval between sounds, and identifies the maximum absolute value of each second division interval as an ambient sound candidate, and the representative value calculation unit calculates a representative value of the ambient sound based on the ambient sound candidate. In this way, ambient sounds can be extracted by dividing into intervals smaller than the interval between sounds and identifying the maximum absolute value as an ambient sound candidate.

[0015] (5) Further, in the acoustic analysis device described in (4) above, the second divided section is characterized in that it is 1 / 18 times or more and 1 / 6 times or less of the interval between hitting sounds. By specifying the second divided section in this way, a short section including the hitting sound can be distinguished from other sections.

[0016] (6) Further, in the acoustic analysis device described in (4) or (5) above, the representative value calculation unit is characterized in that it calculates the representative value of the ambient sound using numerical values within a predetermined range among the ambient sound candidates. Thereby, the sound pressure data of the section including the hitting sound can be excluded, and the sound pressure data of the section including only the ambient sound can be extracted.

[0017] (7) Further, in the acoustic analysis device according to any one of (1) to (6) above, the acquired sound pressure data is the sound pressure data collected using a percussion inspection robot provided with an exhaust pressure fan for adhering to the wall surface. Thereby, even when the inspection location is at a high place, the hitting sound can be acquired, and the soundness evaluation of the wall surface can be performed.

[0018] (8) Further, in the acoustic analysis device according to any one of (1) to (7) above, an analysis evaluation unit that calculates a hitting sound index value based on the representative value of the hitting sound and the representative of the ambient sound, and evaluates the soundness of the wall surface based on the comparison result between the hitting sound index value and the first threshold value, is further provided. Thereby, the soundness evaluation of the wall surface considering the ambient sound can be performed.

[0019] (9) Further, in the acoustic analysis device according to any one of (1) to (7) above, an analysis evaluation unit that calculates a hitting sound index value using the representative values of the hitting sound and the ambient sound, and evaluates the soundness of the wall surface based on the comparison result between the hitting sound index value and the second threshold value, is further provided. Thereby, the soundness evaluation of the wall surface considering the ambient sound can be performed.

[0020] (10) The acoustic analysis device described in any of (1) to (9) above is further characterized by comprising an analysis and evaluation unit that evaluates the integrity of the wall surface by machine learning using the sound pressure data, using a machine learning model that has been trained on the sound pressure data of the wall surface when it is healthy and when it is unhealthy as training data. This makes it possible to easily evaluate the integrity of the wall surface from the acquired sound pressure data and the waveform of the sound pressure data.

[0021] (11) The present invention also relates to an acoustic analysis method for extracting sounds to be analyzed for structural integrity evaluation, comprising the steps of: acquiring sound pressure data generated during a tapping inspection of a wall surface; and dividing the sound pressure data included in the analysis target time from the acquired sound pressure data into a plurality of first division sections larger than the tapping interval, and identifying the maximum absolute value of each of the first division sections as the tapping sound. In this way, by dividing into sections larger than the tapping interval and identifying the maximum absolute value as the tapping sound, the tapping sound produced by the tapping tool can be accurately extracted.

[0022] (12) The present invention also relates to an acoustic analysis program for extracting sounds to be analyzed for structural integrity evaluation, characterized in that it causes a computer to perform the following processes: acquiring sound pressure data generated during a tapping inspection of a wall surface; and dividing the sound pressure data included in the analysis target time from the acquired sound pressure data into a plurality of first division sections larger than the tapping interval, and identifying the maximum absolute value of each of the first division sections as the tapping sound. In this way, by dividing into sections larger than the tapping interval and identifying the maximum absolute value as the tapping sound, the tapping sound produced by the tapping tool can be accurately extracted. [Effects of the Invention]

[0023] Thus, according to the present invention, it becomes possible to perform impact sound analysis while taking into account environmental sounds such as noise and other sounds. [Brief explanation of the drawing]

[0024] [Figure 1] (a), (b), and (c) are the front, rear, and side views of the inspection device, respectively. [Figure 2] This diagram shows the general structure of the hitting area. [Figure 3] This diagram shows the process of inspecting the wall surface of a structure using an inspection device. [Figure 4] This is a block diagram showing an acoustic analysis device according to this embodiment. [Figure 5] Figures (a) to (d) show the procedures for dividing the intervals of percussion sounds and ambient sounds and calculating the average value, respectively. [Figure 6] This figure shows the average values ​​of recorded sound pressure data for percussion and ambient sounds. [Figure 7] Figures (a) and (b) show the sound pressure ratios of a sound wall surface (sound portion) and a delaminating wall surface (delaminating portion), respectively. [Figure 8] (a) and (b) are diagrams showing the results of the wall surface integrity assessment, respectively. [Figure 9] This flowchart shows the operation of the acoustic analysis device. [Figure 10] (a) and (b) are graphs showing sound pressure data for the sound and delaminated areas of wall surface A, respectively. [Figure 11] (a) and (b) are graphs showing sound pressure data for the sound and delaminated areas of wall surface B, respectively. [Figure 12] (a) and (b) are graphs showing the exclusion rate of the maximum absolute value, the standard deviation σ1 of the maximum absolute value, and the sound pressure corresponding to the divided intervals of the sound of impact on the sounded portion of wall A and the delaminated portion of wall B, respectively. [Figure 13] (a) and (b) are graphs showing the relationship between the effective sound pressure range and exclusion rate for the sound portion of wall A and the delaminated portion of wall B, respectively. [Figure 14] (a) and (b) are graphs showing the exclusion rate of the maximum absolute value, the standard deviation σ² of the maximum absolute value, and the sound pressure corresponding to the divided intervals of the sound of impact on the sounded portion of wall A and the delaminated portion of wall B, respectively. [Figure 15] (a) and (b) are graphs showing the relationship between the effective sound pressure range and exclusion rate for the sound portion of wall A and the delaminated portion of wall B, respectively. [Modes for carrying out the invention]

[0025] Embodiments of the present invention will be described below with reference to the drawings. To facilitate understanding of the description, the same reference numerals are used for identical components in each drawing, and redundant descriptions are omitted.

[0026] [Inspection device] Figures 1(a), (b), and (c) are the front, rear, and side views, respectively, of the inspection device. Figure 2 is a diagram showing the schematic configuration of the striking section. The inspection device 100 comprises at least a plurality of striking sections 101, a drive motor (not shown), a pressure exhaust fan 111, a microphone 113, and a tire 115. Each striking section 101 is equipped with a support shaft 103 and a striking element 105.

[0027] The ball-turning element 105 is formed as a polyhedron with a regular hexagonal cross-section and a roughly circular or elliptical longitudinal cross-section, and is rotatably supported at the tip of the support shaft 103. The cross-section is the cross-section perpendicular to the support shaft 103, and the longitudinal cross-section is the cross-section parallel to the support shaft 103. The outer circumferential surface of the ball-turning element 105 has six ball-turning surfaces 105a and corners 105b.

[0028] The rolling pin 105 is moved while the rolling pin 101 is rotated, and the presence or absence of peeling or other defects at the inspection point can be checked by the sound of the rolling pin 105. In this embodiment, there are three rolling pins 101 as an example, but it is not limited to this. There may be one rolling pin 101, or there may be two, four, or more.

[0029] The exhaust fan 111 rotates with a drive motor, creating an airflow from the wall surface of the structure to the outside of the exhaust fan 111, thereby keeping the inspection device from moving away from the wall surface of the structure. The inspection device moves along the wall surface of the structure as the tires 115 rotate. In this way, having the exhaust fan 111 and tires 115 allows the inspection device 100 to be moved while maintaining stable contact with the wall surface of the structure.

[0030] Microphone 113 acquires the sound of the turning mechanism 105. The microphone used in this embodiment is a monaural microphone and does not have an active noise cancellation function. Sound-absorbing material can be attached to the sound-collecting part of the microphone to adjust the sound pressure level. This prevents distortion from occurring when the sound pressure of the acquired sound pressure data exceeds the sound pressure that can be acquired by the microphone. The microphone may have a recording function or a function to transfer the sound pressure data acquired by the microphone to a remote location.

[0031] Figure 3 shows a tapping inspection of the wall surface of a structure using the inspection device shown in Figure 1. As shown in Figure 3, the inspection device 100 of this embodiment can perform a tapping inspection while swinging the inspection device 100 from side to side and gradually moving it from bottom to top. In this way, even if the inspection point is out of reach due to the height of the structure 300, the inspection device can be suspended from the roof of the structure or elsewhere and a tapping inspection can be performed. In addition to being used while suspended, the inspection device can also be controlled and driven from a remote location using a remote controller.

[0032] [Acoustic analysis device] Figure 4 is a block diagram of the acoustic analysis device according to this embodiment. Figures 5(a) to 5(d) are diagrams showing the procedure for dividing the intervals of impact sounds and ambient sounds and calculating the average value, respectively. The acoustic analysis device 200 includes a data acquisition unit 201, an impact sound identification unit 203, an ambient sound identification unit 205, a representative value calculation unit 207, and an analysis and evaluation unit 209. The acoustic analysis device 200 is a processing device such as a PC, and is composed of a processor that executes processing and memory or a hard disk that stores programs and data. For example, the acoustic analysis device 200 receives user input from an input device such as a keyboard or mouse and outputs the processing results to an output device such as a display. The acoustic analysis device 200 may also be a server device located on the cloud.

[0033] The data acquisition unit 201 acquires sound pressure data generated during tapping inspections of the wall surface of a structure. The inspection device 100 and the acoustic analysis device 200 can communicate with each other via a wired or wireless network and send and receive data to and from each other. Alternatively, they can send and receive data via a recording medium on which they output or input data. The acoustic analysis device 200 may also be incorporated into the inspection device 100 and integrated as a measuring device.

[0034] The impact sound identification unit 203 identifies impact sounds from the sound pressure data acquired by the data acquisition unit 201. As shown in Figure 5(a), the sound pressure data acquired by the data acquisition unit 201 that is included in the analysis target time is divided into multiple sections (hereinafter also referred to as the first division section) at time intervals longer than the impact sound interval of the turner. The peak that takes the maximum absolute value of the sound pressure in each first division section is identified as the impact sound. Due to the condition of peeling tiles on the wall surface, the height of the tiles etc. may differ, or the turner may misfire, resulting in irregular occurrences of impact sounds and their magnitude. Therefore, it is necessary to divide the acquired sound pressure data into appropriate sections. The length [s] of the first division section is preferably 2 to 4 times the impact sound interval of the turner, and more preferably 3 times. By identifying the first division section, impact sounds can be included in each first division section in most cases, even if the interval between impact sound occurrences is irregular.

[0035] Here, "impact sound" refers to the sound produced when the wall surface is tapped with a tapping tool. "Environmental sound" refers to sounds other than the impact sound, such as the operating sound of the exhaust fan, the sound of tires driving, and other noises. The analysis target time is an arbitrary time interval extracted from the acquired sound pressure data. In the examples shown in Figures 5(a) and (b), the sound pressure data for 1 second is shown from the sound pressure data acquired as the analysis target time. The reason for selecting a time of 1 second for easier visualization of the waveform is that the actual analysis target time is not limited to this. The impact sound interval is the average impact sound interval of the tapping tool when an inspection is performed using the inspection device. The impact sound generated when tapping with a tapping tool does not necessarily occur at a constant interval due to reasons such as differences in the height of tiles, etc., or dry tapping of the tapping tool. Therefore, it is preferable to use the average impact sound interval of the tapping tool.

[0036] The ambient sound identification unit 205 identifies ambient sounds from the sound pressure data acquired by the data acquisition unit 201. As shown in Figure 5(c), the sound pressure data included in the analysis target time from the sound pressure data acquired by the data acquisition unit 201 is divided into multiple sections (hereinafter also referred to as second division sections) at time intervals smaller than the interval between the sound strikes of the turner. The peak that takes the maximum absolute value of the sound pressure in each second division section is then identified as an ambient sound candidate. The reason it is called an ambient sound candidate rather than an ambient sound is that the maximum absolute value of the second division section may also be a sound strike (hereinafter simply referred to as ambient sound). The length [s] of the second division section is preferably 1 / 18 to 1 / 6 times the interval between the sound strikes of the turner, and more preferably 1 / 12 times. This makes it possible to exclude sound strikes as much as possible from the second division section and extract sound pressure data of ambient sounds only.

[0037] As described in the explanation of the impact sound identification unit 203, the analysis target time is an arbitrary time interval extracted from the acquired sound pressure data. In the examples shown in Figures 5(c) and (d), the sound pressure data for 0.1 seconds of the sound pressure data acquired as the analysis target time is shown. The reason for using 0.1 seconds for the extracted time is to make the waveform easier to understand, and the actual analysis target time is not limited to this. By dividing the analysis target time in this way, impact sounds can be removed and sound pressure data of ambient sounds can be accurately extracted.

[0038] The representative value calculation unit 207 calculates representative values ​​for the percussion sounds and ambient sounds, respectively, from the percussion sounds and ambient sounds identified by the percussion sound identification unit 203 and the ambient sound identification unit 205. The representative value for percussion sounds is the value calculated from the percussion sounds of each first division section identified as percussion sounds by the percussion sound identification unit 203. The representative value for ambient sounds is the value calculated from the ambient sounds of each second division section identified as ambient sound candidates by the ambient sound identification unit 205. The representative values ​​include the mean and median.

[0039] The method for calculating the representative value of the sound is explained (Figure 5(b)). First, the average value of the maximum absolute value of the sound pressure in each first division section (P 10 ) is calculated. The average value of the maximum absolute value of the sound pressure in each of these first division intervals (P 10 ) may be used as a representative value for the sound of impact. However, the average value of the maximum absolute value of this sound pressure (P 10 ) includes sudden sounds and impact sounds such as slippage of the striking part. Therefore, more preferably, in each first division section, sections in which the maximum absolute value of the sound pressure falls outside a predetermined range are excluded, and the average value of the maximum absolute value of the sound pressure in each first division section (P) is calculated again for the remaining data. 11 ) calculate this average value (P 11 ) may be used as a representative value for the sound of impact. The predetermined range is the average value (P 10 This refers to the range obtained by ± an arbitrary magnification × σ1 (σ1: standard deviation). The arbitrary magnification is preferably 2.0 to 3.5, and more preferably 2.5. This allows for the inclusion of low-sound-pressure percussion sounds in the calculation, while excluding intervals that do not contain percussion sounds.

[0040] Next, a method for calculating a representative value of environmental sound will be described (Fig. 5(d)). First, the average value (P 20 ) of the maximum absolute value of the sound pressure in each second divided interval is obtained. This average value (P 20 ) of the maximum absolute value of the sound pressure in each second divided interval may be used as the representative value of the environmental sound. However, this average value (P 20 ) of the maximum absolute value of the sound pressure includes knocking sounds. In order to accurately identify the sound pressure of the environmental sound, it is preferable to calculate it based on the sound pressure in the section without knocking sounds. Therefore, more preferably, in each second divided interval, an interval where the maximum absolute value of the sound pressure is outside a predetermined range is excluded, and for the remaining data, the average value (P 21 ) of the maximum absolute value of the sound pressure in each second divided interval is obtained again, and this average value (P 21 ) may be used as the representative value of the knocking sound. The predetermined range refers to a range obtained by average value (P 20 ) ± arbitrary magnification × σ2 (σ2: standard deviation). The arbitrary magnification is preferably 0.6 to 2.0, and more preferably 1.0.

[0041] Fig. 6 is a diagram showing the average values of the knocking sound and environmental sound of the recorded sound pressure data, and is the average values of the knocking sound and environmental sound of the recorded sound pressure data obtained based on the above-described calculation method.

[0042] The analysis and evaluation unit 209 evaluates the soundness of the wall surface based on statistical values including the average value obtained at each analysis target time. As a method for evaluating the soundness of the wall surface, for example, there is a method of setting a threshold value for soundness evaluation and evaluating based on the comparison result between the representative value of the knocking sound and the threshold value. It is more preferable to set the threshold value for soundness evaluation by directly performing percussion inspection on a part of the wall surface and using the determined result.

[0043] Furthermore, the analysis and evaluation unit 209 may pre-store sound pressure data and waveforms of sound pressure data from similar structures and wall surfaces, as well as statistical values ​​such as average values ​​and thresholds obtained for each analysis target time, and other data acquired or judged to date. By accumulating and analyzing various types of data, the setting of thresholds for soundness evaluation can be made more accurate without relying on human judgment, and the soundness of the wall surface can be examined from various perspectives, enabling the evaluation of the structural integrity.

[0044] Figures 7(a) and 7(b) show the sound pressure ratios of a sound wall surface (sound portion) and a wall surface with delamination (delamination portion), respectively. Figures 8(a) and 8(b) show the evaluation results of the wall surface's soundness, respectively. Note that a wall surface with delamination (delamination portion) refers to an unsound wall surface (unsound portion).

[0045] First, the sound pressure ratio is calculated for each analysis time in both the sound and delaminated sections, and the average sound pressure ratio for the sound and delaminated sections is calculated (Figures 7(a) and (b)). The sound pressure ratio is the ratio of the average sound pressure of the impact noise to the average sound pressure of the ambient noise for each analysis time, calculated by the representative value calculation unit (average sound pressure / average ambient noise). Such a sound pressure ratio is used as an impact noise index value when evaluating the soundness of the wall surface by comparing it with a threshold. As an impact noise index value, the sound pressure difference between the representative value of the impact noise and the representative value of the ambient noise may also be used. Alternatively, the sound pressure ratio or sound pressure difference calculated between the impact noise and the representative value of the ambient noise, which are identified for each sound pressure peak, may also be used.

[0046] Next, a threshold is set at the midpoint between the average sound pressure ratio of the sound pressure area and the average sound pressure ratio of the delaminated area. The soundness of the wall surface is then evaluated based on the comparison between the sound pressure ratio at each analysis time and the threshold. As shown in Figures 8(a) and (b), the soundness of the wall surface can be evaluated based on the magnitude of the sound pressure ratio at the analysis time relative to the calculated threshold. In other words, if the sound pressure ratio at the analysis time is less than the threshold, the wall surface can be judged to be sound, and if the sound pressure ratio at the analysis time is greater than or equal to the threshold, it can be judged that delamination has occurred on the wall surface.

[0047] In addition to the sound pressure ratio, it is also possible to set the threshold value using the sound pressure difference. The sound pressure difference is the difference between the average value of the knocking sound of each analysis target time calculated by the representative value calculation unit and the average value of the ambient sound (average value of knocking sound - average value of ambient sound). First, in the sound-healthy part and the peeling part, the sound pressure difference for each analysis target time is calculated.

[0048] Next, the midpoint between the average value of the sound pressure difference of the sound-healthy part and the average value of the sound pressure difference of the peeling part is set as the threshold value, and the soundness of the wall surface is evaluated based on the comparison result between the average value of the knocking sound of each analysis target time and the threshold value. The soundness of the wall surface can also be evaluated based on the magnitude of the sound pressure difference of the analysis target time with respect to the calculated threshold value. That is, when the sound pressure difference of the analysis target time is less than the threshold value, it can be determined that the wall surface is sound, and when the sound pressure difference of the analysis target time is greater than or equal to the threshold value, it can be determined that peeling has occurred on the wall surface.

[0049] Furthermore, the threshold value calculated from the sound pressure ratio and the sound pressure difference may be calculated by the following formula using a coefficient a where 0 < a < 1, in addition to the arithmetic mean described above. Let the sound pressure of the knocking sound and the ambient sound in the sound-healthy part be N hit 、N back respectively, and the sound pressure of the knocking sound and the ambient sound in the peeling part be A hit 、A back respectively. When the sound pressure of the knocking sound S hit and the ambient sound pressure S back obtained when a certain wall surface is tapped are used, the threshold value T1 calculated by the sound pressure ratio is expressed by Equation (1), and the soundness of the wall surface can be evaluated based on the magnitude of the value of (S hit / S back ) with respect to the threshold value T1.

Equation

[0050] Alternatively, the threshold value T2 calculated by the sound pressure difference is expressed by Equation (2), and the soundness of the wall surface can be evaluated based on the magnitude of the value of (S hit - S back ) with respect to the threshold value T2. Note that representative values can be used for the sound pressure in each equation.

Equation

[0051] Alternatively, the sound pressure of the impact sound can be compared with each of the aforementioned thresholds, and the sound quality of the wall surface can be evaluated based on the relative magnitudes of the values ​​for each threshold and the impact sound pressure.

[0052] The analysis and evaluation unit 209 may also have AI functions, such as a machine learning model. For example, a machine learning model consisting of a neural network with an input layer, a hidden layer, and an output layer can be used. By training the machine learning model with sound pressure data and waveforms of sound pressure data from the sound of impacts on a healthy and unhealthy wall surface as training data, the health of the wall surface can be evaluated from the acquired sound pressure data and waveforms of sound pressure data without calculating a threshold. Alternatively, representative values ​​of impact sounds or ambient sounds calculated from the acquired sound pressure data and waveforms of sound pressure data at each analysis target time may be used as training data. In this way, by accumulating the acquired sound pressure data, waveforms of sound pressure data, and the values ​​calculated from this data as training data, the accuracy of evaluating the health of the wall surface can be further improved.

[0053] [Procedure for assessing the structural integrity of walls] Figure 9 is a flowchart illustrating the operation of the acoustic analysis device. The flowchart in Figure 9 shows the acoustic analysis method as part of the procedure for evaluating the structural integrity of a wall surface. As shown in Figure 9, first, sound pressure data generated during a tapping inspection of the wall surface is acquired (step S1). The sound pressure data includes not only the sound pressure data of the tapping sound but also sound pressure data of ambient noise such as background noise.

[0054] Next, the sound pressure data included in the analysis time from the acquired sound pressure data is divided into multiple first division intervals larger than the sound impact interval (step S2), and the maximum absolute value of each first division interval is identified as the sound impact (step S3). A representative value of the sound impact during the analysis time is calculated (step S4).

[0055] Next, the sound pressure data included in the analysis time from the acquired sound pressure data is divided into multiple second division intervals smaller than the interval between impacts (step S5), and the maximum absolute value of each second division interval is identified as a candidate for ambient sound (step S6). A representative value of the ambient sound during the analysis time is calculated (step S7).

[0056] Based on the representative values ​​of the impact sound and the representative values ​​of the ambient sound, an impact sound index value is calculated (step S8), and the impact sound index value is compared with a threshold (step S9). Alternatively, in step S8, the impact sound index value may be calculated using the impact sound and ambient sound identified in each division interval, and then compared with a threshold.

[0057] In this way, the impact sound index value is compared with a threshold value. If the impact sound index value is below the threshold value, the wall surface is determined to be sound, and the process is terminated. On the other hand, if the comparison result is above the threshold value, the wall surface is determined to be unsound (peeling has occurred), and the process is terminated. The above procedure can be performed by having a computer execute a program.

[0058] [Examples] The acoustic analysis was verified as follows. First, for each wall surface judged to be sound by sensory inspection and for wall surfaces judged to have detached tiles, the inspection device shown in Figure 1 was operated for approximately 10 seconds, and the tapping sound during that time was recorded.

[0059] Specifically, two types of wall surfaces, A and B, were prepared. Wall surfaces A and B were constructed using pressure-bonded ceramic tiles measuring 45mm x 95mm, and both walls had both sound and delaminated sections. Since an inspection device equipped with a pressure exhaust fan, tires, and a motor was used on both wall surfaces A and B, the sound pressure data collected included noise such as the motor sound for rotating the tapping part of the inspection device, the operating sound of the pressure exhaust fan for contacting the inspection device with the wall surface, and the running sound of the tires moving along the wall surface. In addition, wall surface B had air conditioning equipment nearby, so the collected sound pressure data also included noise originating from the air conditioning equipment. The inspection device was driven on wall surfaces A and B for approximately 10 seconds, and the tapping sound during that time was recorded.

[0060] Figures 10(a) and (b) are graphs showing sound pressure data for the sound and delaminated areas of wall surface A, respectively. Figures 11(a) and (b) are graphs showing sound pressure data for the sound and delaminated areas of wall surface B, respectively. In the data shown in Figures 10(a) and (b) and Figures 11(a) and (b), only 1 second of sound pressure data is extracted from the 10 seconds of recorded sound pressure data. The sound pressure of the recorded sound pressure data has been normalized so that the maximum sound pressure that the microphone can pick up is 1.

[0061] (Analysis of percussive sounds) Next, we performed an analysis of the impact sounds and ambient sounds using the extracted sound pressure data. First, we analyzed the impact sounds. The recorded sound pressure data was divided into segments of several times the average impact interval of the turner.

[0062] The average sound interval of the tapping hammer was calculated by recording the sound produced when tapping the wall surface using the inspection device shown in Figure 1, calculating the average number of taps per second from the waveform of the recorded sound pressure data, and taking the reciprocal of the calculated average number. As a result, the average sound interval of the tapping hammer of the inspection device was 0.06 seconds on average, so the sound pressure data was divided into intervals of 0.18 seconds, which is three times the average, and the maximum absolute value of the sound pressure was determined for each interval. In this embodiment, the average sound interval of the tapping hammer was used to determine the division width, but this is not the only method. For example, it may be determined based on the rotation speed of the tapping hammer that taps the wall surface.

[0063] Furthermore, the wall surface used for tapping to calculate the average tapping interval of the tapping tool was a ceramic tile commonly used for walls. In addition, the average tapping interval mainly depends on the material (rigidity), rotation speed, number of tapping tools, and shape of the tapping tools, and the change in the average tapping interval due to the wall material, such as cement-based materials like mortar, or tiles made from ceramics, hard plastics, etc., is small. Therefore, the wall material is not considered when calculating the average tapping interval.

[0064] As a result, the mean and standard deviation of the maximum absolute value for each interval are as shown in Table 1. To exclude intervals where there was no impact sound due to sudden loud noises or slippage of the striking part, the effective sound pressure range was defined as the range in which the maximum absolute value of the sound pressure fell outside the range of mean sound pressure ± 2.5 × standard deviation σ1, and the average value of the remaining sound pressure data was taken again to determine the impact sound pressure.

[0065] [Table 1]

[0066] First, we examined the appropriate width of the division interval. Figures 12(a) and (b) are graphs showing the exclusion rate of the maximum absolute value (the percentage of data that did not fall within the range of the average sound pressure ± 2.5σ1 and were therefore not considered valid) as well as the standard deviation σ1 of the maximum absolute value and the sound pressure when the division interval of the sound of impact noise for the sound of

[0067] As shown in Figures 12(a) and (b), the exclusion rate and deviation increase as the division width decreases. This is because the occurrence of percussion sounds is irregular, and reducing the division width results in many sections that do not contain percussion sounds. In this embodiment, since the average value of percussion sounds is used as the sound pressure, it is undesirable for many sections that do not contain percussion sounds to occur.

[0068] On the other hand, increasing the division width increases the sound pressure, which then plateaus. This is because, when multiple percussion sounds are contained within the same interval, only the percussion sound with the highest sound pressure within that interval is used in the calculation. In this case, only the percussion sounds with the highest sound pressure are evaluated, which is undesirable when determining the average sound pressure of the percussion sounds.

[0069] Based on the above, it was found that, within the scope of the study, a range of approximately 2 to 4 times the average sound interval is preferable for obtaining reasonable evaluation results, and more preferably 3 times the sound interval.

[0070] Next, we investigated the appropriate effective sound pressure range. Figures 13(a) and (b) are graphs showing the relationship between the effective sound pressure range and the exclusion rate for the sound portion of wall A and the delaminate portion of wall B, respectively. The graphs show the results when the ratio of the effective sound pressure range to the standard deviation was changed to 1.5, 2.0, 2.5, 3.0, and 3.5 times. In this investigation, as examples of impact sound data, we used data from the sound portion of wall A, which had the lowest impact sound level, and data from the delaminate portion of wall B, which had the highest impact sound level.

[0071] As shown in Figure 13, if the effective sound pressure range is reduced, low-sound-pressure percussion sounds will be excluded from the calculation. On the other hand, if the effective sound pressure range is made too large, if there is a section that does not contain percussion sounds, the sound pressure of ambient noise will be included in the calculation. However, in this embodiment, percussion sounds were included in every section, so the sound pressure of ambient noise was not included in the calculation.

[0072] Based on the above, it was found that in order to obtain reasonable evaluation results, an effective sound pressure range of approximately 2.0 to 3.5 times the standard deviation is preferable, and more preferably 2.5 times the standard deviation.

[0073] (Analysis of ambient sounds) Next, the ambient sound was analyzed. In the ambient sound analysis, the collected sound pressure data was divided into segments that were sufficiently shorter than the average interval between taps of the tapping device. Since the average interval between taps of the tapping device used in this embodiment was 0.06 seconds, the sound pressure data was divided into segments of 0.005 seconds (1 / 12 of the interval), and the maximum absolute value of the sound pressure was calculated for each segment. As a result, the mean and standard deviation of the maximum absolute value for each segment are shown in Table 2. The ambient sound pressure was evaluated based on the sound pressure of segments that did not contain tapping sounds. To exclude segments containing tapping sounds, the effective sound pressure range was defined as the segment in which the maximum absolute value of the sound pressure fell outside the range of mean ± 1.0 × standard deviation σ² of the sound pressure, and the average value of the remaining sound pressure data was taken again to determine the ambient sound pressure.

[0074] [Table 2]

[0075] The division width of the second division section was examined. Figures 14(a) and (b) are graphs showing the exclusion rate of the maximum absolute value (the percentage of data that did not fall within the range of the average sound pressure ± 1.0σ² and were therefore not considered valid) as well as the standard deviation σ² of the maximum absolute value and the sound pressure when the division section of the sound

[0076] As shown in Figures 14(a) and (b), increasing the division width increases the proportion of sections containing percussive sounds, and the weight of those sections increases in the maximum absolute value, causing the sound pressure to be overestimated compared to the actual ambient sound pressure. On the other hand, if the intervals are too small, sections with low sound pressure within the ambient sound are often evaluated, resulting in a large deviation and an underestimation of the ambient sound pressure.

[0077] Based on the above, within the scope of our investigation, it was found that a range of approximately 1 / 18 to 1 / 6 times the average sound interval is preferable for obtaining reasonable evaluation results, and more preferably 1 / 12 times the sound interval.

[0078] Next, we examined the appropriate effective sound pressure range. Figures 15(a) and (b) are graphs showing the relationship between the effective sound pressure range and the exclusion rate for the sound portion of wall A and the delaminated portion of wall B, respectively, and show the results when the magnification of the standard deviation is changed to 0.5, 1.0, 1.5, 2.0, and 2.5 times.

[0079] As shown in Figure 15, reducing the effective sound pressure range excludes not only percussive sounds but also sections of ambient sound with low sound pressure, resulting in a higher exclusion rate. On the other hand, increasing the effective sound pressure range also picks up percussive sounds. The sound pressure is lowest when the magnification is set to 1.0 on both wall surfaces A and B, suggesting that the average value of ambient sound excluding percussive sounds is being evaluated.

[0080] Based on the above, it was found that, within the scope of the study, a range of approximately 0.6 to 2.0 times the standard deviation is preferable as the effective sound pressure range in order to obtain reasonable evaluation results, and more preferably 1.0 times the standard deviation.

[0081] Using the impact sound pressure and ambient sound pressure obtained in this way, impact sound index values ​​and thresholds for sound integrity can be calculated. Possible methods for calculating impact sound index values ​​include taking the ratio of impact sound pressure to ambient sound pressure (sound pressure ratio) or the difference between impact sound pressure and ambient sound pressure (sound pressure difference). Here, the thresholds (referred to as "threshold (ratio)" and "threshold (difference)," respectively) were set by taking the arithmetic mean of the sound pressure ratio and sound pressure difference between the healthy and detached parts. Table 3 summarizes these values.

[0082] [Table 3]

[0083] For each of wall surfaces A and B, five locations were selected that were judged as sound in the sensory evaluation and five locations that were judged as delaminating. Three sets of one-second recordings were made at each location (resulting in 30 data points: sound area / delaminating area x 5 locations x 3 sets). The same judgments as in the sensory evaluation were then checked using threshold (ratio) and threshold (difference). Table 4 shows the judgment results.

[0084] [Table 4]

[0085] As shown in Table 4, the overall accuracy rate was high, and even on wall surface B with ambient noise, an accuracy rate of approximately 90% was achieved. The present invention's method of determining impact sound pressure and noise sound pressure from waveforms and using them as indicators for judgment proved useful for evaluating the structural integrity of structures.

[0086] After conducting a structural integrity assessment using the method described above, any areas of the wall identified as having delamination may be replaced. Alternatively, before replacing the wall, the areas identified as having delamination may be inspected again by a person using a tapping instrument such as a tapping hammer, and a final determination of whether or not they are indeed delamination areas may be made. In this way, by using an inspection device, the areas that require direct tapping inspection by a person can be limited, thereby reducing the workload.

[0087] In this study, determining wall surface A using a threshold (sound pressure ratio) resulted in a higher accuracy rate, while determining wall surface B using a threshold (sound pressure difference) resulted in a higher accuracy rate. While this study used simple indicators based solely on sound pressure ratios and differences, further improvements in accuracy are expected by combining these with machine learning techniques.

[0088] This specification describes a method for determining the soundness of a structural wall surface using delamination (detached portion) as an example, but is not limited to this. It can also be applied to cases where mortar, tiles, etc., attached to or glued to the wall surface of a building have delaminated, or where deterioration, expansion, air or foreign matter has entered, etc. have occurred.

[0089] As described above, this embodiment makes it possible to perform impact sound analysis while taking into account environmental sounds such as noise and other sounds. [Explanation of symbols]

[0090] 100 Inspection device 101 Batting Department 103 Support shaft 105 Turning batting 105a Rolling surface 105b Corner 111 Exhaust fan 113 Mike 115 tires 200 Acoustic analysis equipment 201 Data Acquisition Unit 203 Percussion Sound Identification Section 205 Environmental sound identification section 207 Representative Value Calculation Unit 209 Analysis and Evaluation Department 300 Structures

Claims

1. An acoustic analysis device for extracting sounds to be analyzed in order to evaluate the structural integrity of a structure, A data acquisition unit that acquires sound pressure data generated during a tapping inspection of a wall surface, which includes the sound of the tapping and other ambient sounds. A sound identification unit divides the sound pressure data included in the analysis target time from the acquired sound pressure data into multiple first division intervals larger than the sound impact interval, and identifies the maximum absolute value of each of the first division intervals as the sound impact. A representative value calculation unit calculates a representative value of the impact sound during the analysis period and a representative value of the ambient sound from sound pressure data excluding the identified impact sound. An acoustic analysis device characterized by being equipped with the following features.

2. An acoustic analysis device for extracting sounds to be analyzed for the purpose of evaluating the soundness of a structure, A data acquisition unit that acquires sound pressure data generated during wall surface tapping inspection, A sound identification unit divides the sound pressure data included in the analysis target time from the acquired sound pressure data into multiple first division intervals larger than the sound impact interval, and identifies the maximum absolute value of each of the first division intervals as the sound impact. An ambient sound identification unit divides the sound pressure data included in the analysis target time from the acquired sound pressure data into multiple second division intervals smaller than the sound interval, and identifies the maximum absolute value of each second division interval as an ambient sound candidate. A representative value calculation unit calculates a representative value of the tapping sound during the analysis period and a representative value of the ambient sound based on the ambient sound candidates. An acoustic analysis device characterized by being equipped with the following features.

3. The acoustic analysis apparatus according to claim 1 or 2, characterized in that the first division interval is two times or more and four times or less the interval between impact sounds.

4. The acoustic analysis apparatus according to claim 1 or 2, characterized in that the representative value calculation unit calculates a representative value of the impact sound using a numerical value within a predetermined range from among the impact sounds.

5. The system further includes an ambient sound identification unit that divides the sound pressure data included in the analysis target time from the acquired sound pressure data into a plurality of second division intervals smaller than the sound interval, and identifies the maximum absolute value of each second division interval as an ambient sound candidate. The acoustic analysis apparatus according to claim 1, characterized in that the representative value calculation unit calculates a representative value of the ambient sound based on the ambient sound candidates.

6. The acoustic analysis apparatus according to claim 5, characterized in that the second division interval is 1 / 18 times or more and 1 / 6 times or less of the sound interval.

7. The acoustic analysis apparatus according to claim 1 or 2, characterized in that the representative value calculation unit calculates a representative value of the ambient sound using a numerical value within a predetermined range from among the ambient sound candidates.

8. The acoustic analysis apparatus according to claim 1 or 2, characterized in that the acquired sound pressure data is sound pressure data collected using a tapping inspection robot equipped with a pressure relief fan for making close contact with the wall surface.

9. The acoustic analysis apparatus according to claim 1 or 2, further comprising: an analysis and evaluation unit that calculates an impact sound index value based on the representative value of the impact sound and the representative value of the ambient sound, and evaluates the soundness of the wall surface based on the comparison result of the impact sound index value and a threshold value.

10. The acoustic analysis apparatus according to claim 1 or 2, further comprising: an analysis and evaluation unit that calculates an impact sound index value using the impact sound and the ambient sound representative values, and evaluates the soundness of the wall surface based on the comparison result between the impact sound index value and a threshold value.

11. The acoustic analysis apparatus according to claim 1 or 2, further comprising: an analysis and evaluation unit that evaluates the health of the wall surface by machine learning using the sound pressure data, using a machine learning model trained on the sound pressure data of the wall surface when it is healthy and when it is unhealthy as training data.

12. An acoustic analysis method for extracting sounds to be analyzed for structural integrity assessment, A step of acquiring sound pressure data generated during a tapping inspection of a wall surface, which includes the sound of the tapping and other ambient sounds, The steps include dividing the sound pressure data included in the analysis time from the acquired sound pressure data into multiple first division intervals larger than the sound impact interval, and identifying the maximum absolute value of each of the first division intervals as the sound impact, A step of calculating a representative value of the impact sound during the analysis period and a representative value of the ambient sound from sound pressure data excluding the identified impact sound. An acoustic analysis method characterized by including the following.

13. An acoustic analysis program for extracting sounds to be analyzed in order to assess the structural integrity of a structure, A process for acquiring sound pressure data generated during a tapping inspection of a wall surface, which includes the sound of the tapping and other ambient sounds. The process involves dividing the sound pressure data included in the analysis time from the acquired sound pressure data into multiple first division intervals larger than the sound impact interval, and identifying the maximum absolute value of each of the first division intervals as the sound impact. A process for calculating a representative value of the impact sound during the analysis period and a representative value of the ambient sound from sound pressure data excluding the identified impact sound. An acoustic analysis program characterized by having a computer execute it.

Images