Inspection device and method for estimating sound source location

Abstract

To provide an inspection device capable of locating a position of an abnormal sound source even if an abnormal sound of a rotating machine has periodicity.SOLUTION: An inspection device includes: multiple microphones detecting sound generated from each of multiple rotating machines; a sound waveform processing unit obtaining a feature amount of a sound waveform detected by the multiple microphones; an abnormality detection unit detecting an abnormal sound on the basis of the feature amount; a waveform time difference calculation unit calculating the minimum time difference of sound waveforms among the multiple microphones; and an abnormal sound source position locating unit locating a rotating machine that generates an abnormal sound from among the multiple rotating machines on the basis of the minimum time difference.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] Embodiments of the present invention relate to an inspection device and a method for estimating the location of a sound source. [Background technology]

[0002] Numerous rotating machines, including prime movers, transmissions, generators, control motors, and pumps, are used in factories and power plants. While sensors such as vibration meters are installed on major rotating machines to monitor their condition, it is difficult to install sensors on all rotating machines and monitor their condition due to the complexity of the monitoring system. Therefore, non-major rotating machines are often inspected periodically by humans, and corrective measures are taken if any abnormalities are found.

[0003] In recent years, there has been a desire to automate the inspection of rotating machinery within plants in order to reduce the labor costs required for inspection work. However, as mentioned above, it is difficult to install sensors such as vibration meters on all rotating machinery. Therefore, it is known that multiple microphones are installed in the plant to simultaneously acquire sound waveforms emitted by multiple rotating machines in order to determine whether there is an abnormality or to identify the abnormal rotating machine.

[0004] As described above, when inspecting rotating machinery using multiple microphones, a technique has been proposed to generate white noise with a wide range of frequency components from an experimental sound source, determine the phase difference spectrum of the sound waveforms between the microphones, and pinpoint the sound source location. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2022-17037 [Patent Document 2] Japanese Patent Publication No. 2003-337164 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] Generally, the sound waveforms emitted by rotating machines are periodic. Therefore, among the waveforms of abnormal sounds of rotating machines, there can also be waveforms having periodicity instead of general attenuation waveforms. Even when the abnormal sound has periodicity, by using the above technique, the distance difference from the abnormal sound source between microphones can be uniquely determined. However, it is difficult to intentionally generate sounds of arbitrary frequencies for each of a plurality of types of rotating machines installed in an actual plant.

[0007] The problem to be solved by the present invention is to provide an inspection device capable of specifying the position of an abnormal sound source and a method for estimating the sound source position even when the abnormal sound of a rotating machine has periodicity.

Means for Solving the Problem

[0008] [[ID=!13]] An inspection device according to an embodiment includes a plurality of microphones that detect sounds respectively emitted from a plurality of rotating machines, a sound waveform processing unit that obtains feature amounts of the sound waveforms detected by the plurality of microphones, an abnormality detection unit that detects an abnormal sound based on the feature amounts, a waveform time difference calculation unit that calculates a minimum time difference of the sound waveforms between the plurality of microphones when the abnormal sound is detected, and an abnormal sound source position specification unit that specifies a rotating machine that has emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference.

Effect of the Invention

[0009] According to the present embodiment, even when the abnormal sound of a rotating machine has periodicity, it becomes possible to specify the position of the abnormal sound source.

Brief Description of the Drawings

[0010] [Figure 1] It is a block diagram showing a schematic configuration of an inspection device according to the first embodiment. [Figure 2] It is a diagram showing an example of the frequency spectrum of a rotating machine in the first embodiment. [Figure 3] It should be noted that in the translation of the text in item

[14] , it seems there is a tag problem in the original text you provided. The tag seems incorrect. I translated it as as it is, but please check if it's a correct tag in the original context.This figure shows another example of the frequency spectrum of a rotating machine in the first embodiment. [Figure 4] This flowchart shows an example of a method for estimating the location of an abnormal sound source according to the first embodiment. [Figure 5] This figure shows an example of a waveform of an abnormal sound from rotating machinery. [Figure 6] This figure illustrates the process for determining the distance difference from an abnormal sound source between microphones in the first embodiment. [Figure 7] This is a block diagram showing the schematic configuration of the inspection device according to the second embodiment. [Figure 8] This figure shows an example of data used to estimate the cause of an anomaly. [Figure 9] An example of the frequency spectrum of a rotating machine in the third embodiment is shown. [Figure 10] This flowchart shows an example of a method for estimating the location of an abnormal sound source according to the third embodiment. [Figure 11] This figure shows an example of distance difference data. [Figure 12] An example of the frequency spectrum of a rotating machine in the fourth embodiment is shown. [Figure 13] This flowchart shows an example of a method for estimating the location of an abnormal sound source according to the fourth embodiment. [Figure 14] This is an explanatory diagram of the process for determining the distance difference from an abnormal sound source between microphones in the fourth embodiment. [Figure 15] This is an example of sound waveforms measured by each microphone when the operating conditions of an abnormally rotating machine are changed. [Figure 16] This flowchart shows an example of a method for estimating the location of an abnormal sound source according to the fifth embodiment. [Modes for carrying out the invention]

[0011] Embodiments of the present invention will be described below with reference to the drawings. The embodiments described below are not intended to limit the present invention.

[0012] (First Embodiment) Figure 1 is a block diagram showing a schematic configuration of the inspection device according to the first embodiment. The inspection device 10 according to this embodiment is installed in the plant 11. The plant 11 also has several types of rotating machinery 9A to 9E installed. Each of the rotating machines is, for example, a prime mover, a transmission, a generator, a control motor, and a pump.

[0013] The inspection device 10 shown in Figure 1 comprises microphones 1 and 2, sound waveform processing units 3 and 4, anomaly detection unit 5, waveform time difference calculation unit 6, and an anomaly sound source location identification unit 7. In Figure 1, to simplify the explanation of the embodiment of the invention, all of the rotating machines 9A to 9E are arranged in a straight line at a parallel distance L from microphones 1 and 2. However, the parallel distance from microphones 1 and 2 may differ between the rotating machines. In this case, as described in Japanese Patent Application Publication No. 2022-17037, the location of the anomaly sound source is not fixed to a single point, but rather a hyperbola that is a candidate for the anomaly sound source is determined. Therefore, by further increasing the number of microphones and determining the distance difference dt between the microphones with respect to the sound source, the location of the anomaly sound source can be fixed to a single point.

[0014] Microphones 1 and 2 are installed at a distance h from each other, as shown in Figure 1. Microphones 1 and 2 detect sounds emitted from multiple rotating machines 9A to 9E and generate sound signals indicating the sound pressure of the detected sounds.

[0015] The sound waveform processing unit 3 determines the characteristic features of the sound signal waveform generated by microphone 1. These characteristic features include, for example, the dominant frequency and amplitude of the sound waveform from microphone 1.

[0016] The sound waveform processing unit 4 determines the characteristic features of the sound signal waveform generated by microphone 2. These characteristic features include, for example, the dominant frequency and amplitude of the sound waveform from microphone 2.

[0017] The anomaly detection unit 5 detects abnormal sounds from the rotating machines 9A to 9E based on the results of comparing the characteristic quantities obtained by the sound waveform processing units 3 and 4 with the characteristic quantities of the rotating machines 9A to 9E under normal conditions.

[0018] The waveform time difference calculation unit 6 calculates the time difference between the sound waveforms between microphone 1 and microphone 2 when an abnormal sound is detected.

[0019] The abnormal sound source location unit 7 identifies the rotating machine that emitted the abnormal sound from among the rotating machines 9A to 9E based on the time difference calculated by the waveform time difference calculation unit 6 and the distance h between microphone 1 and microphone 2.

[0020] The following describes a method for estimating the location of an abnormal sound source using the inspection device 10 according to this embodiment. However, in this embodiment, it is assumed that when an abnormality occurs, the rotating machine emits sound having multiple frequency components. Here, as an example, let's assume that the rotating machine 9D shown in Figure 1 is emitting an abnormal sound.

[0021] Figure 2(a) shows an example of the frequency spectrum of a normal sound from the rotating machine 9D. Figure 2(b) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits a normal sound. Figure 2(c) shows an example of the frequency spectrum of an abnormal sound from the rotating machine 9D. Figure 2(d) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits an abnormal sound. In Figures 2(a) to 2(d), the horizontal axis represents frequency and the vertical axis represents sound pressure.

[0022] In the examples shown in Figures 2(a) and 2(c), the rotating machine 9D emits an abnormal sound with increased sound pressure at frequencies f1 [Hz] and f2 [Hz]. There are many types of vibrations in rotating machinery, including unbalanced vibrations of the rotating shaft, vibrations excited by the working fluid or bearing oil film, and resonance due to external forces. Furthermore, the frequency of the generated sound also differs depending on the type of vibration. Therefore, as shown in Figure 2(c), it is quite possible that the sound pressure of multiple frequency components will increase when the rotating machine 9D malfunctions.

[0023] Figure 3(a) shows another example of the frequency spectrum of the normal sound of the rotating machine 9D. Figure 3(b) shows another example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits a normal sound. Figure 3(c) shows another example of the frequency spectrum of the abnormal sound of the rotating machine 9D. Figure 3(d) shows another example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits an abnormal sound. In Figures 3(a) to 3(d), the horizontal axis represents frequency and the vertical axis represents sound pressure.

[0024] In the examples shown in Figures 3(a) and 3(c), the sound pressure of the normal sound of the rotating machine 9D is high only at frequency f1, while the sound pressure of the abnormal sound is high at both frequencies f1 and f2. Thus, in this embodiment, the normal sound of the rotating machine 9D may have multiple frequency components as shown in Figure 2(a), or it may have only one frequency component as shown in Figure 3(a).

[0025] Figure 4 is a flowchart showing an example of a method for estimating the location of an abnormal sound source according to the first embodiment.

[0026] First, the sound waveform processing unit 3 determines characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by microphone 1 (step S11). In step S11, the sound waveform processing unit 3 obtains the frequency spectrum by performing a Fourier transform on the sound signal from microphone 1, for example, as shown in Figures 2(b), 2(d), 3(b), and 3(d). In the frequency spectrum, frequencies at which the sound pressure exceeds a preset reference value are extracted as dominant frequencies. Note that microphone 1 measures not only the sound of the rotating machine 9D but also the sound of the remaining rotating machines. Therefore, as shown in Figures 2(b), 2(d), 3(b), and 3(d), the sound signal from microphone 1 contains other frequency components in addition to dominant frequency components such as frequencies f1 and f2.

[0027] In parallel with step S11, the sound waveform processing unit 4 determines characteristic quantities such as the dominant frequency and amplitude included in the waveform of the sound signal detected by microphone 2 (step S12). In step S12, the sound waveform processing unit 4 obtains the frequency spectrum by performing a Fourier transform on the sound signal from microphone 2. Like microphone 1, microphone 2 measures not only the sound from the rotating machine 9D but also the sound from the remaining rotating machines. Therefore, in addition to frequencies f1 and f2, other frequency components are also present in the sound signal from microphone 2.

[0028] Next, the anomaly detection unit 5 compares the feature quantities shown in the sound waveform of microphone 1 or microphone 2 with the feature quantities under normal conditions. At this time, the anomaly detection unit 5 calculates the difference between the sound pressures of frequencies f1 and f2 in the frequency spectrum shown in Figure 2(d), for example, and the sound pressures of frequencies f1 and f2 in the frequency spectrum shown in Figure 2(b), as the difference in feature quantities. Subsequently, the anomaly detection unit 5 determines whether the difference in feature quantities exceeds a preset threshold (step S13).

[0029] If the difference in feature quantities exceeds a threshold (Step S13: YES), the anomaly detection unit 5 determines that an anomaly has occurred in the rotating machinery. On the other hand, if the difference in feature quantities is less than or equal to the threshold (Step S13: NO), the anomaly detection unit 5 determines that each rotating machine is functioning normally. In this case, the process of estimating the location of the anomaly sound source is terminated.

[0030] If the anomaly detection unit 5 determines that an anomaly exists in step S13, the waveform time difference calculation unit 6 compares the sound waveform of microphone 1 with the sound waveform of microphone 2 and calculates the minimum time difference for each of the multiple frequency components that were determined to be an anomaly (step S14). Here, step S14 will be explained using Figure 5.

[0031] Figure 5 shows an example of the waveform of an abnormal sound from the rotating machine 9D. In Figure 5, the horizontal axis represents time [s], and the vertical axis represents the sound pressure of the abnormal sound. The upper part of Figure 5 shows an example of the waveform of an abnormal sound at frequency f1 measured by microphone 1. On the other hand, the lower part of Figure 5 shows an example of the waveform of an abnormal sound at frequency f2 measured by microphone 2.

[0032] The minimum time difference calculated in step S14 is Δt in Figure 5. o This corresponds to the minimum time difference Δt. o This corresponds to the minimum time difference between the time at which the sound pressure measured by microphone 1 reaches its peak value and the time at which the sound pressure measured by microphone 2 reaches its peak value. The waveform time difference calculation unit 6 calculates the phase difference for each frequency component of the abnormal sound, and then converts the phase difference into a time difference using a predetermined conversion formula. The waveform time difference calculation unit 6 calculates the converted time difference as the minimum time difference Δt o It is calculated as follows.

[0033] Next, the abnormal sound source location unit 7 calculates candidate distance differences between microphone 1 and microphone 2 from the abnormal sound source using the minimum time difference calculated for each frequency component of the abnormal sound (step S15).

[0034] Next, the abnormal sound source location identification unit 7 identifies the true distance difference d from the above candidates. t Calculate (Step S16). Now, refer to Figure 6 to explain Steps S15 and S16.

[0035] Figure 6 is a diagram illustrating the process for determining the distance difference between microphones from an abnormal sound source in the first embodiment. In Figure 6, the horizontal axis represents the wavenumber difference n, and the vertical axis represents the distance difference d between microphone 1 and microphone 2 from the abnormal sound source.

[0036] The distance difference d is calculated by the following equation (1). In equation (1), c is the speed of sound. Also, Δt is the time difference between the time at which the sound pressure measured by microphone 1 reaches its peak value and the time at which the sound pressure measured by microphone 2 reaches its peak value. d = cΔt (1)

[0037] When the sound waveform of each microphone is a sine wave with a frequency f [Hz], the time difference Δt is the minimum time difference Δt o and the wave number difference n are calculated by Equation (2). Here, the wave number difference 1 corresponds to the time difference of 1 / f shown in FIG. 5. Δt = Δt o + n / f (2)

[0038] Substituting Equation (2) into Equation (1), the distance difference d is expressed by the following Equation (3). d = c(Δt o + n / f) (3)

[0039] Here, if the minimum time difference of the frequency f1 component calculated in step S14 is Δt o1 and the minimum time difference of the frequency f2 component is Δt o2 then, based on the above Equation (3), the distance difference d1 of the frequency f1 component is expressed by the following Equation (4) and the distance difference d2 of the frequency f2 component is expressed by the following Equation (5). d1 = c(Δt o1 + n / f1) (4) d2 = c(Δt o2 + n / f2) (5)

[0040] In FIG. 6, straight lines represented by the functional expressions of Equations (4) and (5) are shown. Since the wave number difference n is an integer value, the points indicated by black circles and white circles are candidates for the distance difference d. The distances from the abnormal sound source to each microphone are independent of the frequency of the sound waveform. Therefore, among the candidates of the black circles on the straight line represented by the functional expression of Equation (5), there should be those whose distance difference d coincides with the candidates of the white circles on the straight line represented by Equation (4). In FIG. 6, the distance difference d t corresponds to this. In this way, the abnormal sound source position specifying unit 7 calculates candidates for the distance difference between the microphones from the abnormal sound source for each frequency component of the abnormal sound based on the minimum time difference, and calculates the distance difference that coincides among the plurality of frequency components as the true distance difference d t and calculates it.

[0041] Finally, the abnormal sound source location identification unit 7 calculates the location of the abnormal sound source and identifies the abnormal rotating machine (step S17). The location of the abnormal sound source is determined by the distance difference d calculated by the abnormal sound source location identification unit 7. t This can be calculated using a predetermined formula with known distance h between microphones and parallel distance L to the sound source as parameters.

[0042] For example, in this embodiment, as shown in Figure 1, the distance difference between microphones is d on a straight line with a parallel distance L from each microphone. t The point is, d t If the sign of is also taken into consideration, it can be determined to a single point, and the abnormal rotating machine can be identified as rotating machine 9D. In this embodiment, d t If the sign of is positive, that is, if the distance from the abnormal sound source is longer for microphone 2 than for microphone 1, then the rotating machine 9D is identified as abnormal. Conversely, d t If the sign of the signal is negative, that is, if the distance from the abnormal sound source is shorter for microphone 2 than for microphone 1, then, for example, the rotating machine 9B is identified as abnormal.

[0043] According to the embodiment described above, even if the sound waveform of the rotating machine has periodicity, it is possible to uniquely identify the location of the abnormal sound source.

[0044] (Second Embodiment) Figure 7 is a block diagram showing the schematic configuration of the inspection device according to the second embodiment. In Figure 7, the same reference numerals are used for components similar to those in the inspection device 10 according to the first embodiment described above, and detailed explanations are omitted.

[0045] The inspection device 12 shown in Figure 7 further includes a storage unit 8 in addition to the components of the inspection device 10 according to the first embodiment. The storage unit 8 stores abnormality cause estimation data for estimating the cause of abnormalities in rotating machinery.

[0046] Figure 8 shows an example of abnormality cause estimation data. The abnormality cause estimation data 100 shown in Figure 8 indicates the abnormality cause for each rotating machine for each dominant frequency. The abnormality detection unit 5 uses the abnormality cause estimation data 100 to estimate the abnormality cause of the rotating machine in which an abnormality has been detected.

[0047] The inspection device 12 according to this embodiment identifies the location of an abnormal sound source and identifies an abnormal rotating machine, similar to the inspection device 10 according to the first embodiment, according to the flowchart shown in Figure 4. However, in this embodiment, if a frequency component that is dominant during an abnormality is not present during normal operation, the abnormality detection unit 5 uses the abnormality cause estimation data 100 to estimate the cause of the abnormality in the rotating machine in which the abnormality was detected.

[0048] For example, as shown in Figures 3(a) and 3(b), the frequency f2 component is present during abnormal conditions but not during normal conditions. In this case, in step S17 of the flowchart shown in Figure 4, when the abnormal sound source location identification unit 7 calculates the location of the abnormal sound source and identifies the rotating machine 9D as an abnormal rotating machine, the abnormality detection unit 5 reads the cause of the abnormality of the dominant frequency f2 for the rotating machine 9D from the abnormality cause estimation data 100 stored in the memory unit 8. As a result, the abnormality detection unit 5 estimates that the cause of the abnormality is unbalanced vibration of the rotating shaft.

[0049] According to the embodiment described above, if a frequency component that is dominant during an abnormality is not present during normal operation, it is possible to predict what type of vibration is significantly occurring based on that frequency. Therefore, it is possible not only to identify the location of an abnormal sound source from the sound waveform of a periodic rotating machine, but also to estimate the cause of the abnormality in the rotating machine. In this embodiment, the inspection device 12 estimates the cause of the abnormality, but some or all of the operation of the inspection device 12 may be performed by the operator of the plant 11. For example, the inspection device 12 may display the abnormality cause estimation data 100, and the operator may estimate the cause of the abnormality.

[0050] (Third embodiment) The configuration of the inspection device according to the third embodiment is the same as that of the inspection device 10 according to the first embodiment described above, so a description will be omitted.

[0051] In this embodiment, it is assumed that an abnormal rotating machine emits sounds with different frequency components depending on the operating conditions, such as rotational speed and load. Furthermore, there is no particular limit to the number of dominant frequencies of sounds that can be emitted simultaneously from a single rotating machine. Here, as in the first embodiment, it is assumed that the rotating machine 9D shown in Figure 1 is emitting an abnormal sound.

[0052] Figure 9(a) shows an example of the frequency spectrum when the rotating machine 9D is rotating at a low speed. Figure 9(b) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D is rotating at a low speed. Figure 9(c) shows an example of the frequency spectrum when the rotating machine 9D is rotating at a high speed. Figure 9(d) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D is rotating at a high speed. In Figures 9(a) to 9(d), the horizontal axis represents frequency and the vertical axis represents sound pressure.

[0053] As shown in Figures 9(a) and 9(c), in this embodiment, the sound pressure of frequency f1 increases during low-speed rotation, and the sound pressure of frequency f2 increases during high-speed rotation. For example, if there is an imbalance in the rotating shaft of the rotating machine 9D, a centrifugal force synchronized with the rotational speed acts on the shaft, causing it to vibrate. As a result, as shown in Figures 9(a) and 9(c), it is quite possible that the sound pressure of different frequency components increases depending on the rotational speed.

[0054] Figure 10 is a flowchart showing an example of a method for estimating the location of an abnormal sound source according to the third embodiment.

[0055] First, the sound waveform processing unit 3 determines characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by the microphone 1 (step S21). In step S21, the sound waveform processing unit 3 obtains the frequency spectrum by performing a Fourier transform on the sound signal from the microphone 1 for multiple operating conditions, for example, as shown in Figures 9(b) and 9(d). In this frequency spectrum, frequencies in which the sound pressure exceeds a preset reference value are extracted as dominant frequencies.

[0056] As shown in Figure 9(b), when the abnormal rotating machine 9D is rotating at a low speed, frequency f1 is the dominant frequency. Also, as shown in Figure 9(d), when the abnormal rotating machine 9D is rotating at a high speed, frequency f2 is the dominant frequency.

[0057] In parallel with step S21, the sound waveform processing unit 4 obtains characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by microphone 2 (step S22). In step S22, the sound waveform processing unit 4 obtains the frequency spectrum by performing a Fourier transform on the sound signal from microphone 2 for multiple operating conditions. In this frequency spectrum, frequencies in which the sound pressure exceeds a preset reference value are extracted as dominant frequencies.

[0058] Next, the anomaly detection unit 5 compares the feature quantities shown in the sound waveform of microphone 1 or microphone 2 with the feature quantities under normal conditions. At this time, the anomaly detection unit 5 calculates the difference in feature quantities for each operating condition and determines whether this difference exceeds a preset threshold (step S23).

[0059] If the difference in feature quantities exceeds a threshold (Step S23: YES), the anomaly detection unit 5 determines that an anomaly has occurred in the rotating machinery. On the other hand, if the difference in feature quantities is less than or equal to the threshold (Step S23: NO), the anomaly detection unit 5 determines that each rotating machine is functioning normally. In this case, the process of estimating the location of the anomaly sound source is terminated.

[0060] If the abnormality detection unit 5 determines that there is an abnormality in step S23, the waveform time difference calculation unit 6 compares the sound waveform of microphone 1 and the sound waveform of microphone 2 under the same operating conditions and calculates the minimum time difference of the frequency component determined to be abnormal for each operating condition (step S24). In step S24, similar to step S14 of the first embodiment, the waveform time difference calculation unit 6 determines the phase difference between the sound waveform of microphone 1 and the sound waveform of microphone 2 for each operating condition, and then converts the phase difference into a time difference using a predetermined conversion formula. The waveform time difference calculation unit 6 calculates the minimum time difference Δt for each operating condition. o It is calculated as follows.

[0061] Next, the abnormal sound source location unit 7 calculates candidate distance differences between microphone 1 and microphone 2 from the abnormal sound source using the minimum time difference calculated for each operating condition (step S25). In step S25, candidate distance differences during low-speed rotation lie on the functional equation of distance difference d1 represented by equation (4) described in the first embodiment. Also, candidate distance differences during high-speed rotation lie on the functional equation of distance difference d2 represented by equation (5) described in the first embodiment.

[0062] Next, the abnormal sound source location identification unit 7 identifies the true distance difference d from the above candidates. t The true distance difference d is calculated (step S26). The distance of each microphone from the abnormal sound source is independent of the frequency of the sound waveform. Therefore, the abnormal sound source location unit 7 calculates the true distance difference d by finding the distance difference that matches the frequency components of multiple operating conditions. t It is calculated as follows.

[0063] Next, the abnormal sound source location identification unit 7 calculates the location of the abnormal sound source and identifies the abnormal rotating machine (step S27). The location of the abnormal sound source is determined by the distance difference d calculated by the abnormal sound source location identification unit 7, similar to the first embodiment. t This can be calculated using a predetermined formula with known distance h between microphones and parallel distance L to the sound source as parameters.

[0064] According to the embodiment described above, similar to the first embodiment, it is possible to uniquely identify the location of an abnormal sound source even when the sound waveform of the rotating machine has periodicity. Furthermore, in this embodiment, it is possible to identify the location of an abnormal sound source even if the dominant frequency component of the sound emitted by the abnormal rotating machine is single for each operating condition.

[0065] (Fourth Embodiment) The configuration of the inspection device according to the fourth embodiment is the same as that of the inspection device 12 according to the second embodiment described above, so its description will be omitted. That is, the inspection device according to this embodiment further includes a storage unit 8 in addition to the components of the inspection device 10 according to the first embodiment. However, in this embodiment, the storage unit 8 stores distance difference data instead of abnormality cause estimation data 100.

[0066] Figure 11 shows an example of distance difference data. The distance difference data 200 shown in Figure 11 shows the distance difference between microphone 1 and microphone 2 for each rotating machine. Once the positions of each rotating machine, microphone 1, and microphone 2 are determined, the distance difference to each rotating machine (d A d B d C d D d E ) is also uniquely determined. Therefore, in the inspection device according to this embodiment, distance difference data 200, which stores each rotating machine and the corresponding distance difference, is stored in the storage unit 8.

[0067] In this embodiment, as in the first embodiment, it is assumed that the rotating machine 9D shown in Figure 1 is emitting an abnormal noise.

[0068] Figure 12(a) shows an example of the frequency spectrum of a normal sound from the rotating machine 9D. Figure 12(b) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits a normal sound. Figure 12(c) shows an example of the frequency spectrum of an abnormal sound from the rotating machine 9D. Figure 12(d) shows an example of the frequency spectrum of the sound measured by microphones 1 and 2 when the rotating machine 9D emits an abnormal sound. In Figures 12(a) to 12(d), the horizontal axis represents frequency and the vertical axis represents sound pressure.

[0069] In this embodiment, the number of dominant frequencies for the sound emitted by the abnormally rotating machinery is not particularly limited. Therefore, as shown in Figures 2(b) and 2(d), it is acceptable to assume that the sound pressure of frequencies f1 and f2 increases during abnormal conditions.

[0070] Figure 13 is a flowchart showing an example of a method for estimating the location of an abnormal sound source according to the fourth embodiment.

[0071] First, the sound waveform processing unit 3 determines characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by the microphone 1 (step S31). In step S31, the sound waveform processing unit 3 obtains the frequency spectrum by performing a Fourier transform on the sound signal from the microphone 1, for example, as shown in Figures 12(b) and 12(d). In the frequency spectrum, frequencies in which the sound pressure exceeds a preset reference value are extracted as dominant frequencies.

[0072] In parallel with step S31, the sound waveform processing unit 4 obtains characteristic quantities such as the dominant frequency and amplitude included in the waveform of the sound signal detected by microphone 2 (step S32). In step S32, the sound waveform processing unit 4 obtains the frequency spectrum by performing a Fourier transform on the sound signal from microphone 2.

[0073] Next, the anomaly detection unit 5 compares the feature quantities shown in the sound waveform of microphone 1 or microphone 2 with the feature quantities under normal conditions. At this time, the anomaly detection unit 5 calculates the difference between the sound pressure at frequency f1 in the frequency spectrum shown in Figure 9(d), for example, and the sound pressure at frequency f1 in the frequency spectrum shown in Figure 2(b), as the difference in feature quantities. Subsequently, the anomaly detection unit 5 determines whether the difference in feature quantities exceeds a preset threshold (step S33).

[0074] If the difference in feature quantities exceeds a threshold (step S33: YES), the anomaly detection unit 5 determines that an anomaly has occurred in the rotating machinery. On the other hand, if the difference in feature quantities is less than or equal to the threshold (step S33: NO), the anomaly detection unit 5 determines that each rotating machine is functioning normally. In this case, the process of estimating the location of the anomaly sound source is terminated.

[0075] If the abnormality detection unit 5 determines that there is an abnormality in step S33, the waveform time difference calculation unit 6 compares the sound waveform of microphone 1 and the sound waveform of microphone 2 and calculates the minimum time difference for the dominant frequency component that was determined to be abnormal (step S34). In step S34, similar to step S14 of the first embodiment, the waveform time difference calculation unit 6 finds the phase difference between the sound waveform of microphone 1 at frequency f1 and the sound waveform of microphone 2 at frequency f1, and then converts the phase difference into a time difference using a predetermined conversion formula. The waveform time difference calculation unit 6 calculates the converted time difference as the minimum time difference Δt of the dominant frequency. o It is calculated as follows.

[0076] Next, the abnormal sound source location unit 7 uses the calculated minimum time difference to calculate candidate distance differences between microphone 1 and microphone 2 from the abnormal sound source (step S35). In step S35, candidate distance differences of frequency f1 lie on the functional equation of the distance difference d1 represented by equation (4) described in the first embodiment.

[0077] Next, the abnormal sound source location identification unit 7 uses the distance difference data 200 to select the true distance difference d from the above candidates. tCalculate (Step S36). Now, refer to Figure 14 to explain Step S36.

[0078] Figure 14 is an explanatory diagram of the process for determining the distance difference between microphones from an abnormal sound source in the fourth embodiment. In Figure 14, the horizontal axis represents the wavenumber difference n, and the vertical axis represents the distance difference d between microphone 1 and microphone 2 from the abnormal sound source.

[0079] In Figure 14, multiple candidates are drawn as black circles on the straight line represented by equation (4) described in the first embodiment. The distance of each microphone from the abnormal sound source is independent of the frequency of the sound waveform. Therefore, the distance difference d shown in distance difference data 200 A ~distance difference d E One of these should match one of the candidates above.

[0080] Therefore, the abnormal sound source location identification unit 7 compares the candidate distance difference with the distance difference data 200 and finds the distance difference that matches. In Figure 14, the distance difference d D The true distance difference d t It corresponds to this.

[0081] Finally, the abnormal sound source location unit 7 determines the location of the abnormal sound source and identifies the abnormal rotating machine (step S37). For example, the abnormal sound source location unit 7 determines the distance difference d in the distance difference data 200. D The rotating machine 9D associated with the sound source is identified as the source of the abnormal sound, and the position of the rotating machine 9D is determined as the position of the source of the abnormal sound.

[0082] According to this embodiment described above, as with the other embodiments described above, it is possible to uniquely identify the location of an abnormal sound source even when the sound waveform of the rotating machine has periodicity. Furthermore, in this embodiment, it is possible to identify the location of an abnormal sound source even if the dominant frequency component of the sound emitted by the abnormal rotating machine is a single frequency component.

[0083] (Fifth embodiment) The configuration of the inspection device according to the fifth embodiment is the same as that of the inspection device 10 according to the first embodiment described above, so a description will be omitted.

[0084] In this embodiment, it is assumed that when operating conditions such as rotational speed and load change in an abnormally functioning rotating machine, the sound waveform emitted from the machine also changes accordingly. Here, as in the first embodiment, it is assumed that the rotating machine 9D shown in Figure 1 is emitting an abnormal sound. Furthermore, it is assumed that when the abnormality of the rotating machine 9D occurs during low-speed rotation, the sound pressure at frequency f1 increases, as shown in Figure 12(c).

[0085] Figure 15 shows an example of sound waveforms measured by each microphone when the operating conditions of the abnormal rotating machine 9D are changed from "low-speed rotation" to "stopped". The upper part of Figure 15 shows the sound waveform measured by microphone 1. On the other hand, the lower part of Figure 15 shows the sound waveform measured by microphone 2. According to Figure 15, when the rotating machine 9D is rotating at low speed, a sine wave with frequency f1 is mainly measured, and when it stops, the sine wave disappears. In this embodiment, the change in operating conditions from "low-speed rotation" to "stopped" may be achieved by an abnormality detection function provided in the rotating machine 9D, or by remote operation by an operator of the plant 11.

[0086] Figure 16 is a flowchart showing an example of a method for estimating the location of an abnormal sound source according to the fifth embodiment.

[0087] First, the sound waveform processing unit 3 determines characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by the microphone 1 (step S41). In step S41, the sound waveform processing unit 3 performs a Fourier transform on the sound signal from the microphone 1, including changes in operating conditions, such as the change from "low-speed operation" to "stopped" as shown in the upper part of Figure 15, to obtain a frequency spectrum. In this frequency spectrum, frequencies in which the sound pressure exceeds a preset reference value are extracted as dominant frequencies.

[0088] In parallel with step S41, the sound waveform processing unit 4 determines characteristic quantities such as dominant frequencies and amplitudes included in the waveform of the sound signal detected by microphone 2 (step S42). In step S42, the sound waveform processing unit 4, similar to step S21, performs a Fourier transform on the sound signal from microphone 2, including changes in operating conditions, to obtain a frequency spectrum. In this frequency spectrum, frequencies in which the sound pressure exceeds a preset reference value are extracted as dominant frequencies.

[0089] Next, the anomaly detection unit 5 compares the feature quantities shown in the sound waveform of microphone 1 or microphone 2 with the feature quantities under normal conditions. At this time, the anomaly detection unit 5 calculates the difference between the sound pressure at frequency f1 of the frequency spectrum shown in Figure 12(d), for example, and the sound pressure at frequency f1 of the frequency spectrum shown in Figure 12(d), as the difference in feature quantities. Subsequently, the anomaly detection unit 5 determines whether the difference in feature quantities exceeds a preset threshold (step S43).

[0090] If the difference in feature quantities exceeds a threshold (Step S43: YES), the anomaly detection unit 5 determines that an anomaly has occurred in the rotating machinery. On the other hand, if the difference in feature quantities is less than or equal to the threshold (Step S43: NO), the anomaly detection unit 5 determines that each rotating machine is functioning normally. In this case, the process of estimating the location of the anomaly sound source is terminated.

[0091] If the abnormality detection unit 5 determines that there is an abnormality in step S23, the waveform time difference calculation unit 6 compares the sound waveform of microphone 1 and the sound waveform of microphone 2 and calculates the minimum time difference of the frequency component that was determined to be abnormal (step S44). In step S44, the abnormality detection unit 5 calculates the minimum time difference Δt as the time difference in which the sound waveform changes between the microphones due to changes in the operating conditions of the rotating machine 9D. o The minimum time difference Δt is calculated. For example, as shown in Figure 15, the sine wave of frequency f1 disappears when the operating conditions of the rotating machine 9D are changed from "low speed rotation" to "stopped". Therefore, the difference between the time detected by microphone 1 and the time detected by microphone 2 for this change in operating conditions is the minimum time difference Δt. oThis is the result. Furthermore, as a method for actually calculating the time difference, one could consider using the cross-correlation coefficient between waveforms, as shown in equation (4) of Japanese Patent Publication No. 2022-17037.

[0092] Next, the abnormal sound source location unit 7 determines the minimum time difference Δt o Based on the distance difference d from the abnormal sound source between microphones t The distance difference d is calculated (step S45). In step S45, the abnormal sound source location unit 7 uses the above-described formula (1) to calculate the distance difference d t Calculate.

[0093] Finally, the abnormal sound source location unit 7 determines the distance difference d t The location of the abnormal sound source is calculated using a predetermined formula with the distance h between the microphones and the parallel distance L to the abnormal sound source as parameters (step S46). This identifies the abnormal rotating machine D.

[0094] According to the embodiment described above, the minimum time difference Δt is achieved by utilizing the fact that the periodicity of the waveform is interrupted when the operating conditions change. o This is calculated. Therefore, even if the sound waveform of a rotating machine has periodicity, it is possible to uniquely identify the location of the abnormal sound source.

[0095] Although several embodiments have been described above, these embodiments are presented only as examples and are not intended to limit the scope of the invention. The novel system described herein can be implemented in a variety of other forms. Furthermore, various omissions, substitutions, and modifications can be made to the forms of the system described herein without departing from the spirit of the invention. The appended claims and equivalents are intended to include such forms and modifications that are included in the scope and spirit of the invention. [Explanation of Symbols]

[0096] 1: Microphone 2: Microphone 3: Sound Waveform Processing Unit 4: Sound Waveform Processing Unit 5: Anomaly detection unit 6: Waveform Time Difference Calculation Unit 7: Abnormal sound source location identification part 8: Storage section 9A~9E: Rotating machinery 10: Inspection device 11: Plant 12: Inspection device

Claims

1. Multiple microphones to detect sounds emitted from multiple rotating machines, A sound waveform processing unit that calculates the characteristic quantities of the sound waveform detected by the multiple microphones, An anomaly detection unit that detects abnormal sounds based on the aforementioned feature quantities, When the abnormal sound is detected, a waveform time difference calculation unit calculates the minimum time difference between the sound waveforms between the plurality of microphones, An abnormal sound source location identification unit identifies the rotating machine that emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference, Equipped with, The aforementioned abnormal sound has multiple frequency components, The waveform time difference calculation unit determines the phase difference of the sound waveform for each of the multiple frequency components, then converts the phase difference into a time difference using a predetermined conversion formula, and calculates the converted time difference as the minimum time difference. The abnormal sound source location identification unit is an inspection device that, for each of the multiple frequency components, determines candidate distance differences from the sound source between the multiple microphones based on the minimum time difference, and calculates the matching distance difference among the multiple frequency components from among the candidates.

2. If the plurality of frequency components have a dominant frequency that is not present in a normal sound, the abnormality detection unit estimates the cause of the abnormality in the rotating machine based on the dominant frequency, as described in claim 1.

3. Multiple microphones for detecting sounds emitted from multiple rotating machines, A sound waveform processing unit that calculates the characteristic quantities of the sound waveform detected by the multiple microphones, An anomaly detection unit that detects abnormal sounds based on the aforementioned feature quantities, When the abnormal sound is detected, a waveform time difference calculation unit calculates the minimum time difference between the sound waveforms between the plurality of microphones, An abnormal sound source location identification unit identifies the rotating machine that emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference, Equipped with, The sound waveform processing unit determines the dominant frequency of the abnormal sound for each of the multiple operating conditions, The waveform time difference calculation unit determines the phase difference of the sound waveform for each of the plurality of operating conditions, then converts the phase difference into a time difference using a predetermined conversion formula, and calculates the converted time difference as the minimum time difference. The abnormal sound source location identification unit is an inspection device that, for each of the multiple operating conditions, determines candidate distance differences from the sound source between the multiple microphones based on the minimum time difference, and calculates a matching distance difference among the candidates for each of the multiple operating conditions.

4. The system further includes a storage unit that stores distance difference data relating the distance difference between the multiple rotating machines and the multiple microphones. The inspection device according to claim 1, wherein the abnormal sound source location identification unit determines a distance difference that matches the distance difference data among the candidates.

5. Multiple microphones for detecting sounds emitted from multiple rotating machines, A sound waveform processing unit that calculates the characteristic quantities of the sound waveform detected by the multiple microphones, An anomaly detection unit that detects abnormal sounds based on the aforementioned feature quantities, When the abnormal sound is detected, a waveform time difference calculation unit calculates the minimum time difference between the sound waveforms between the plurality of microphones, An abnormal sound source location identification unit identifies the rotating machine that emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference, Equipped with, The waveform time difference calculation unit is an inspection device that determines the minimum time difference as the difference in the time at which the sound waveform changes in response to changes in the operating conditions of the plurality of rotating machines, measured between each of the plurality of microphones.

6. The sounds emitted from multiple rotating machines are detected by multiple microphones. The characteristic quantities of the sound waveforms detected by the aforementioned multiple microphones are obtained, Based on the aforementioned feature quantities, abnormal sounds are detected. When the abnormal sound is detected, the minimum time difference between the sound waveforms between the multiple microphones is calculated. This includes identifying the rotating machine that emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference, The phase difference of the sound waveform is determined for each of the multiple frequency components of the abnormal sound, and then the phase difference is converted into a time difference using a predetermined conversion formula, and the converted time difference is calculated as the minimum time difference. The further step includes, for each of the multiple frequency components, determining candidate distance differences from the sound source between the multiple microphones based on the minimum time difference, and calculating matching distance differences among the multiple frequency components from among the candidates. Method for estimating the location of a sound source.

7. Sounds emitted from multiple rotating machines are detected by multiple microphones, The characteristic quantities of the sound waveforms detected by the aforementioned multiple microphones are obtained, Based on the aforementioned feature quantities, abnormal sounds are detected. When the abnormal sound is detected, the minimum time difference between the sound waveforms between the multiple microphones is calculated. This includes identifying the rotating machine that emitted the abnormal sound from among the plurality of rotating machines based on the minimum time difference, This further includes calculating the difference in the time at which the sound waveform changes in response to changes in the operating conditions of the multiple rotating machines, as measured between each of the multiple microphones, as the minimum time difference. Method for estimating the location of a sound source.

Images