WO2015146082A1

WO2015146082A1 - Leak-detecting device, leak detection method, and program-containing recording medium

Info

Publication number: WO2015146082A1
Application number: PCT/JP2015/001519
Authority: WO
Inventors: 友督小野; 宝珠山　治
Original assignee: 日本電気株式会社
Priority date: 2014-03-26
Filing date: 2015-03-18
Publication date: 2015-10-01
Also published as: JPWO2015146082A1

Abstract

In this invention, leaks in pipes can be detected with a high degree of precision, even in noisy environments. This leak-detecting device comprises the following: a leak-presence determination unit that uses signals representing detection results from a plurality of sensors to compute the value of a cross-correlation function for said signals, and if the computed value is greater than or equal to a first threshold and the change over time exhibited by a value computed from measurements at a different point in time is less than or equal to a second threshold, determines that a leak is present; and a leak-location computation unit that computes leak-sound propagation-time differences on the basis of areas for which the value of the aforementioned cross-correlation function is greater than or equal to the aforementioned first threshold and the change over time exhibited by the value of said cross-correlation function from measurements at a different point in time is less than or equal to the aforementioned second threshold and identifies the location of the leak on the basis of said propagation-time differences.

Description

Leak detection device, leak detection method, and recording medium for storing program

The present invention relates to a technique for detecting that a delivery target such as liquid or gas leaks from a delivery structure such as a pipe.

In the work of investigating the leakage of liquids and gases to be delivered from pipes such as water pipes, sewage pipes and gas pipes buried in the ground, in order to reduce the cost of pipe repair, It is important to accurately identify As a method for specifying the leak position, there is a method in which an inspector listens to a sound generated due to the leak, and a position having the largest leak sound is set as the leak position. However, this method requires skill and takes time because it is performed by sharpening the nerve.

An example of a method for automating the specification of a leakage position is described in Patent Document 1. In the pipe leak detection method described in Patent Document 1, sensors are installed at two points on both ends of a survey target section on a pipeline to measure the leak sound, and the leak point is determined based on the propagation time difference of the leak sound to the two sensors. Identify the location. Specifically, first, a cross-correlation function is calculated from leaked sound signals acquired by two sensors, and a propagation time difference of the leaked sound is calculated based on the calculated cross-correlation function value. Based on the calculated propagation time difference, the distance from the leakage point to the sensor installation position is obtained.

Japanese Patent Laid-Open No. 11-201859

As shown in FIG. 11A, the pipe leakage detection method (correlation method) described in Patent Literature 1 is based on the point (peak) where the value of the cross-correlation function φ (τ) is the maximum value. A propagation time difference _Tm is calculated. In this method, the accuracy of specifying the leakage position depends on how accurately the propagation time difference _Tm is calculated.

In the pipe leak detection method described in Patent Document 1, when the influence of ambient noise on the leaked sound is small, a clear peak appears in the cross-correlation function as shown in FIG. We calculate the time difference _{T m.} However For example, when containing a large noise in one of the two sensor installation point may not appear distinct peak as shown in (b) of FIG. 11, the accuracy of the transit time T _m is reduced. Further, when noise enters both of the two points, peaks appear as many as the number of noise sources as shown in FIG. 11C, and the peak caused by the noise is erroneously recognized as the peak of the leaked sound. Accordingly, the accuracy of the propagation velocity difference T _m is reduced significantly.

環境 There are many noisy environments, such as roads near busy roads and busy streets. The piping leak detection method described in Patent Document 1 has a problem that it is difficult to accurately specify the leak position in an environment where such noise is generated.

The main object of the present invention is to provide a technique for solving the above-described problems.

A leak detection device according to an aspect of the present invention that achieves the above-described object,
Using signals representing the detection results of a plurality of sensors, the value of the cross-correlation function of these signals is calculated, the calculated value is equal to or greater than the first threshold value, and the time change of the calculated value at different measurement times is the first. A leakage presence / absence determination means for determining that there is a leakage when the threshold is equal to or less than 2,
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. And leakage position calculation means for specifying the leakage position based on the propagation time difference.

A leakage detection method according to an aspect of the present invention that achieves the above object is as follows.
Using signals representing the detection results of multiple sensors, calculate the value of the cross-correlation function of those signals,
It is determined that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at different measurement times is equal to or less than a second threshold;
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. ,
A leak position is specified based on the propagation time difference.

A program according to an aspect of the present invention that achieves the above object is as follows:
On the computer,
Using signals representing the detection results of multiple sensors, calculate the value of the cross-correlation function of those signals,
It is determined that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at different measurement times is equal to or less than a second threshold;
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. ,
The leakage position is specified based on the propagation time difference.

Furthermore, a further aspect of the present invention for achieving the above object is a computer-readable storage medium storing such a program.

According to the present invention, it is possible to accurately detect leakage from a pipe even in an environment where noise is generated.

It is a block diagram showing the structure of the leak detection apparatus which concerns on 1st embodiment of this invention. It is a figure for demonstrating the determination method of the leak sound and noise in a leak presence-and-absence determination part. It is a block diagram showing the structure of the leak detection apparatus which concerns on 2nd embodiment of this invention. It is a figure explaining the example of the frequency distribution (histogram) of a cross correlation function. It is a figure showing the histogram of a cross correlation function, and an identification boundary. It is a block diagram showing the structure of the leak detection apparatus which concerns on 3rd embodiment of this invention. It is a block diagram showing the structure of the leak detection apparatus which concerns on 3rd embodiment of this invention. It is a block diagram showing the structure of the leak detection apparatus which concerns on 4th embodiment of this invention. It is a block diagram showing the structure of the leak detection apparatus which concerns on 5th embodiment of this invention. It is a block diagram showing the structure of the leak detection apparatus which concerns on 7th embodiment of this invention. It is a figure explaining the structural example regarding embodiment of this invention. It is explanatory drawing which shows the example of the time change of the cross-correlation function (phi) ((tau)) calculated using the piping leak detection method of patent document 1. FIG. It is a block diagram showing the structure of the leak detection apparatus which concerns on 6th embodiment of this invention. It is the figure which represented typically the time change of the cross correlation function (phi) ((tau)). It is a flowchart explaining operation | movement of the leak detection apparatus A1 in 1st embodiment. It is a flowchart explaining operation | movement of the leak detection apparatus A2 in 2nd embodiment. FIG. 6 is a diagram illustrating an identification boundary learned by an identification boundary learning unit 84. It is a block diagram showing the structure of the leak detection apparatus which concerns on 8th embodiment of this invention. And FIG. 11 is a block diagram illustrating an example of elements constituting a computer.

Hereinafter, an embodiment in which the present invention is applied to two sensors spaced apart from a pipe will be described in detail with reference to the drawings.

FIG. 10 is a diagram illustrating a configuration example relating to the embodiment of the present invention. As shown in FIG. 10, in the embodiment of the present invention, sensors 100 </ b> A and 100 </ b> B for measuring leakage sound are installed at two points on both ends of the survey target section L on the pipeline. And the position of the leak point (leakage location) 110 is specified based on the propagation time difference of the leaked sound to the sensor 100A and the sensor 100B. In the following embodiments, the medium flowing through the pipe is gas, powder, liquid, or the like, and the present invention is not limited to a specific medium.

In the embodiment of the present invention, the leakage position is specified by utilizing the feature that leakage sound is generated steadily or substantially steadily and the correlation between a plurality of sensors installed in the pipe is large. Specifically, first, a leak detection device (not shown in FIG. 10) according to each of the following embodiments has a time τ based on leaked sound signals measured by a plurality of sensors installed on a pipeline. A cross-correlation function φ (τ), which is a function, is calculated from Equation 1. Here, x _A (t) represents an input signal input from the sensor 100A at time t, and x _B (t) represents an input signal input from the sensor 100B at time t. T represents the total measurement time.
[Equation 1]

Then, when the value of the calculated cross-correlation function φ (τ) is large and the time change is small, the leak detection device regards it as the peak (maximum value) of the cross-correlation function caused by the leaked sound, Based on the peak, the leak position is specified by the correlation method. The time change (small) of the cross-correlation function φ (τ) represents the amount (small) that the value of φ (τ) in a certain value of τ fluctuates in different measurement times (measurement periods). . FIG. 13 is a diagram schematically showing a time change of the cross-correlation function φ (τ). 13 (a), 13 (b), and 13 (c) represent cross-correlation functions calculated from input signals at measurement times t1 to t2, t2 to t3, and t3 to t4, respectively. In the embodiment of the present invention, as shown in FIG. 13 (a), FIG. 13 (b), and FIG. 13 (c), a peak that always appears at the same time τ and has a large value becomes a leaky sound. Considered as a peak due. This is because the value of “leakage sound” in FIGS. 13A, 13B, and 13C is large, and the amount that the value fluctuates at different measurement times is small. It is. In the embodiment of the present invention, other sudden peaks are regarded as peaks caused by noise. This is because the “noise” portion in FIGS. 13A, 13B, and 13C has a large value at a certain measurement time, but a small value at a different measurement time. This is because the amount of fluctuation is large. Note that whether or not the value is large and whether or not the time change of the value is small may be determined by comparing with a predetermined threshold value.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following, in order to simplify the description, a case where two sensors are used as shown in FIG. 10 will be described. However, if there are a plurality of sensors, three or more sensors may be used. Moreover, the kind of sensor to be used is not ask | required. The sensor used may be, for example, a vibration sensor or a microphone.

[First embodiment]
FIG. 1 is a block diagram showing a configuration of a leak detection apparatus according to the first embodiment of the present invention. The leak detection apparatus A1 according to the first embodiment of the present invention includes a signal input unit 10, a time division unit 11, a cross-correlation function calculation unit 12, an average / variance calculation unit 13, and a leak presence / absence determination unit 15. , A propagation time difference calculation unit 16 and a leakage position calculation unit 17 are provided. The signal input unit 10 inputs time-synchronized signals obtained by sensors installed at two points across the inspection target section. The longer the input signal measurement time, the higher the accuracy. The measurement time of the input signal may be 24 hours, for example. The time division unit 11 outputs a signal obtained by dividing the total measurement time of the two input signals into fixed time intervals (periods) T, respectively. The cross-correlation function calculating unit 12 calculates a cross-correlation function for each signal obtained by the time dividing unit 11 from two input signals at each divided measurement time. When the leaked sound signals detected by the two sensors 100A and 100B are x _A (t) and x _B (t), respectively, the cross-correlation function in a certain divided measurement time (t _n to (t _n + T)) φ _n (τ) is given as a function of time τ as shown in Equation 2.
[Equation 2]

Or

When

The number 2-2 normalized by may be used as the cross-correlation function φ _n (τ).
[Equation 2-2]

The average / dispersion calculation unit 13 calculates a function M (τ) that gives an average value of φ _n (τ) for each τ and a function V (τ) that gives a dispersion value by Equations 3 and 4, respectively. . As in Equation 3, the function M (τ) is an average value divided through each of the different measurement times. Further, as shown in Equation 4, the function V (τ) is a dispersion value divided through different measurement times. The variance value refers to a value that represents the degree to which the sample values are scattered from the average value. The variance value is obtained by, for example, averaging the square of the difference between the value of each sample and the average value.
[Equation 3]

[Equation 4]

N is the number of measurement signals divided by the time division unit 11. The leakage presence / absence determination unit 15 compares M (τ) and V (τ) with preset threshold values M _th and V _th , respectively, and at least one τ = τ ′ satisfying both of the two conditions of Formula 5 If it exists, it is determined that there is a leak. And after outputting a determination result, it progresses to the process by the propagation time difference calculation part 16. FIG.
[Equation 5]
M (τ ′)> M _th
V (τ ′) <V _th
Note that the thresholds M _th and V _th are determined based on the actually acquired leakage sound. If there is no τ ′ that satisfies both of the two conditions of Equation 5, the leakage presence / absence determination unit 15 determines that there is no leakage, and the process ends.

FIG. 2 is a conceptual diagram for explaining the determination method in the leakage presence / absence determination unit 15 in the case of leakage sound and noise. 2A, 2B, and 2C show the cross-correlation function φ _n (τ) in the case of leakage sound and noise. 2 (A), FIG. 2 (B), and FIG. 2 (C) are plots of the average value and dispersion value of φ _n (τ) for each τ in FIG. D), (E) in FIG. 2, and (F) in FIG. In the case of (A) in FIG. 2 and (D) in FIG. 2, since there exists τ in which the average value of φ _n (τ) is larger than the threshold value M _th and the variance value is smaller than the threshold value V _th , the leakage presence / absence determining unit 15 Is regarded as a peak of φ _n (τ) due to leaked sound, and it is determined that there is a leak. On the other hand, in the case of FIGS. 2B and 2E, the average value is large, but the variance value is large, so the leakage presence / absence determination unit 15 determines noise (not leakage). 2C and 2F, since the average value is small although the variance value is small, the leakage presence / absence determination unit 15 determines noise (not leakage). By presenting the result of judgment on the presence or absence of leakage in a form that can be seen by the investigator, it can be used to determine the necessity of pipe repair.

Propagation time difference calculating portion 16 calculates the 2 satisfies τ'number 5 as the propagation time difference T _m of a leakage sound. When there are a plurality of τ satisfying the two conditions of Equation 5, it is considered that there are a plurality of leakage points, and the propagation time difference _Tm is calculated for each. Leakage position calculating unit 17, the number 6, and calculates a distance _{L a} from leaking point to the sensor 100A, a distance _{L b} from the leakage point to the sensor 100B.
[Equation 6]
L _a = (L−T _m C) / 2
L _b = L−L _a
However, C represents the leakage sound propagation speed.

FIG. 14 is a flowchart for explaining the operation of the leak detection apparatus A1 in the first embodiment. Next, with reference to FIG. 14, operation | movement of the leak detection apparatus A1 in 1st embodiment is demonstrated. First, the signal input unit 10 inputs a signal representing a detection result obtained from two sensors synchronized in time (step S1). The time division unit 11 outputs a signal obtained by dividing the total measurement time of the two obtained input signals into fixed time intervals (step S2). The cross-correlation function calculation unit 12 calculates a cross-correlation function of two input signals at each time τ for each of the divided signals (step S3). The average / variance calculation unit 13 calculates the average value and the variance value of the cross-correlation function for each time τ (step S4). The leakage presence / absence determination unit 15 compares the average value and the variance value of the cross-correlation function with a predetermined threshold value (step S5). Then, the leakage presence / absence determination unit 15 determines that there is leakage when there is one or more time τ in which the average value is greater than or equal to the threshold value and the variance value is less than or equal to the threshold value (YES in step S5) (step S6). In other cases (NO in step S5), it is determined that there is no leakage (step S7). If it is determined that the leakage is present, the propagation time difference calculating portion 16, by regarding the propagation time difference of the leakage sounds time tau, calculates the propagation time difference T _m (step S8). The leakage position calculation unit 17 calculates the distance from the leakage point to each sensor based on _Tm (step S9).

As described above, the leak detection device A1 according to the first embodiment of the present invention uses the continuity of leaked sound and the correlation between multiple sensors to increase the value of the cross-correlation function and change with time. If it is small, the position is identified as a peak due to leaked sound. Specifically, the leakage detection apparatus A1 determines whether the cross-correlation function value is large and the time change is small by determining whether the average value of the cross-correlation function is equal to or greater than the threshold value and the variance value is equal to or less than the threshold value. Determine. For this reason, the leak detection apparatus A1 has an effect that the presence / absence of the leak and the leak position can be specified even in an environment where noise is generated. In particular, even when there is noise that overlaps the leaked sound and the frequency band, such as a running sound of an automobile, the leak detection device A1 can accurately identify the presence / absence of the leak and the leak position.

[Second Embodiment]
Next, a second embodiment based on the leak detection device according to the first embodiment described above will be described. In the following description, the characteristic part according to the second embodiment will be mainly described. At this time, the same reference numerals are assigned to the same configurations as those in the first embodiment described above, and redundant description is omitted.

FIG. 3 is a block diagram showing the configuration of the leak detection apparatus according to the second embodiment of the present invention. The leak detection device A2 according to the second embodiment of the present invention is replaced with the average / dispersion calculation unit 13, the leak presence / absence determination unit 15, and the propagation time difference calculation unit 16 in the leak detection device A1 of the first embodiment. The mode value calculation unit 23, the leakage presence / absence determination unit 25, and the propagation time difference calculation unit 26 are included. The mode value calculation unit 23 is a cross-correlation function φ _n (τ) (0 ≦ n ≦ N) in the divided measurement times (t _n to (t _n + T)) obtained by the cross-correlation function calculation unit 12. ), The number of times (frequency) C (τ, i) where φ _n (τ) satisfies the inequality of Equation 7 is counted for each τ and each i. That is, the mode value calculation unit 23 converts the value of φ _n (τ) to the section values (b ₀ to b ₁ ,..., B _i to b _{i + 1} ,..., B _M−1 to b for each predetermined section width S. _M ). Then, the mode value calculation unit 23 counts the number of times (appearance frequency) corresponding to the value of φ _n (τ) for each section value. Thereafter, as shown in Equation 8, the mode value calculation unit 23 calculates the maximum value (mode) C _m (τ) of C (τ, i) and the cross-correlation function φ ( τ) is calculated.
[Equation 7]

[Equation 8]

However, as shown in Equation 7, the minimum value b ₀ of b _{i is} the minimum value of φ _n (τ), and the maximum value b _M is the maximum value of φ _n (τ). The section width S = b _{i + 1} −b _i is determined to be sufficiently small in a range where C (τ, i) has a certain variation or more. An example of the frequency distribution (histogram) of the cross-correlation function obtained by the mode value calculation unit 23 is shown in FIG. However, the vertical axis is a value obtained by normalizing the frequency C (τ, i) by the section width S.

As shown in FIG. 4A, the leakage presence / absence determination unit 25 normalizes the mode value C _m (τ) by the section width S and the cross-correlation function φ (τ ) Each determine that there is a leak when there is one or more τ = τ ′ satisfying both of the two conditions of Equation 9. And after outputting a determination result, it progresses to the process by the propagation time difference calculation part 26. FIG.
[Equation 9]
C _m (τ ′) / S> C _th
φ (τ ′)> Φ _th
The threshold _values C _th and Φ _th indicated by dotted lines in FIG. 4 are determined from the actually acquired leaked sound. As shown in (B) of FIG. 4 and (C) of FIG. 4, if there is no τ satisfying Equation 9, it is output that there is no leakage, and the process ends. The propagation time difference calculation unit 26 calculates τ ′ when the condition of Equation 9 is satisfied as the propagation time difference T _m caused by the leaked sound.

FIG. 15 is a flowchart for explaining the operation of the leak detection apparatus A2 in the second embodiment. Next, the operation of the leakage detection apparatus A2 including the mode value calculation unit 23, the leakage presence / absence determination unit 25, the propagation time difference calculation unit 26, and the leakage position calculation unit 17 will be described with reference to the flowchart of FIG. . In addition, about the process similar to the flowchart of FIG. 14, the same number is provided and description is abbreviate | omitted. The cross-correlation function calculation unit 12 calculates a cross-correlation function between two input signals at each time for the divided signals (step S3). The mode value calculation unit 23 calculates the frequency of the cross-correlation function at each time τ (step S14), and calculates the maximum value (mode value) of the frequency and the cross-correlation function when the mode value is taken ( Step S15). The leakage presence / absence determination unit 25 compares the mode value and the cross-correlation function when taking the mode value with a threshold value (step S16). Then, the leakage presence / absence determination unit 25 determines that there is leakage when there is a time τ in which both exceed the threshold (YES in step S16), and outputs a determination result (step S17), otherwise (step S17). If NO in S16, it is determined that there is no leakage, and the process ends (step S18). When it is determined that there is leakage, the propagation time difference calculation unit 26 calculates the time τ when the cross-correlation function when taking the mode value and the mode value exceeds the threshold as the propagation time difference T _m (step S19). . Finally, the leakage position calculating section 17 calculates the leakage position from the propagation time difference _{T m} (step S20).

As described above, the leak detection device A2 according to the second embodiment of the present invention constantly leaks sound but does not temporarily generate noise while measuring a long-time input signal. The feature that there is a time zone is used. Therefore, the leak detection device A2 can determine whether there is a leak and specify the leak position even in an environment where noise frequently occurs, such as a road with a lot of traffic.

[Third embodiment]
Next, a third embodiment based on the leakage detection device according to the first embodiment and the second embodiment described above will be described. In the following description, the characteristic part according to the third embodiment will be mainly described. In this case, the same reference numerals are assigned to the same configurations as those in the first embodiment and the second embodiment described above, and redundant description is omitted.

In the third embodiment of the present invention, in order to extract a section in which noise is not generated as much as possible, the sensor input signal is divided into fixed time intervals as in the first embodiment or the second embodiment. Before, a time interval in which the levels of the input signals of a plurality of sensors are low is extracted, and each interval is divided into fixed time intervals. For this reason, the leak detection devices A3 and A3 ′ according to the third embodiment of the present invention have leak detection devices A1 and A1 of the first embodiment and the second embodiment, respectively, as shown in FIGS. 6A and 6B. A signal level calculation unit 30 and an evaluation interval calculation unit 31 are provided after the signal input unit 10 in A2. In this way, by removing a section in which the input signal level is particularly large, it is possible to remove the influence of the temporarily generated noise, as opposed to the leaked sound that is always generated. For this reason, in addition to the effects in the first embodiment and the second embodiment, the leak detection devices A3 and A3 ′ according to the third embodiment of the present invention determine whether there is a leak in a noise environment and specify the leak position. This has the effect of increasing the accuracy.

[Fourth embodiment]
Next, a fourth embodiment based on the leakage detection device according to the second embodiment described above will be described. In the following description, the characteristic part according to the fourth embodiment will be mainly described. In that case, about the structure similar to 2nd embodiment mentioned above, the overlapping description is abbreviate | omitted by attaching | subjecting the same reference number.

As shown in FIG. 7, the leak detection device A4 according to the fourth embodiment of the present invention has a signal input unit 40 instead of the signal input unit 10 in the leak detection device A2 of the second embodiment. For example, the signal input unit 40 does not input a signal continuously for 24 hours, but measures one hour every three hours, for example, to measure for a certain period at a certain time interval. To do. Thereby, the leak detection device A4 according to the fourth embodiment of the present invention can efficiently acquire various data of day and night even when the amount of data and power that can be processed are limited. For this reason, in addition to the effect in 2nd embodiment, the leak detection apparatus A4 which concerns on 4th embodiment of this invention has the effect that a leak presence determination and leak position specification can be performed without reducing accuracy.

[Fifth embodiment]
Next, a fifth embodiment based on the leakage detection apparatus according to the first embodiment described above will be described. In the following description, the characteristic part according to the fifth embodiment will be mainly described. At this time, the same reference numerals are assigned to the same configurations as those in the first embodiment described above, and redundant description is omitted.

In the fifth embodiment of the present invention, when the average value M (τ) of the cross-correlation function φ _n (τ) is calculated as in the first embodiment, the value of the cross-correlation function is particularly high. When the value is low, the average value is calculated from only the remaining value without using the value at that time as the abnormal value. For example, when the number of measurement signals divided by the time division unit 11 is N = 100 and the cross correlation function values are arranged in descending order, the remaining 80 other than the values of the top 10 and the bottom 10 When the average value is calculated using only the value, the average value M (τ) is given by Equation 10.
[Equation 10]

Alternatively, median μ (τ) may be calculated from Equation 11 instead of the average value.
[Equation 11]

For this reason, as shown in FIG. 8, the leak detection device A5 according to the fifth embodiment of the present invention uses an average / variance instead of the average / variance calculation unit 13 in the leak detection device A1 of the first embodiment. A calculation unit 53 is included. Thereby, in addition to the effect of 1st embodiment, the said leak detection apparatus A5 can suppress the precision fall of leak presence determination and leak position specification by an abnormal value.

[Sixth embodiment]
Next, a sixth embodiment based on the leakage detection apparatus according to the first embodiment described above will be described. In the following description, the characteristic part according to the sixth embodiment will be mainly described. At this time, the same reference numerals are assigned to the same configurations as those in the first embodiment described above, and redundant description is omitted.

The leakage presence / absence determination unit 15 of the first embodiment determines the presence / absence of leakage by comparing the average value and the variance value of the cross-correlation function with a threshold value. In the sixth embodiment of the present invention, instead of this, based on a plurality of leaked sound signals (leakage sound information) and noise signals (noise information) prepared in advance, an identification boundary on the average / dispersion graph is set. Learning is performed, and the presence or absence of leakage is determined depending on which side the evaluation target signal is located with respect to the identification boundary. Here, the identification boundary is a boundary line that defines each region when there is leakage and when there is no leakage. That is, the identification boundary is a set of combinations of average values and variance values (points on the average / variance graph) that serve as a boundary for determining whether there is leakage.

As shown in FIG. 12, the leak detection device A6 according to the sixth embodiment of the present invention is a leak sound / noise signal input unit instead of the leak presence / absence determination unit 15 in the leak detection device A1 of the first embodiment. 80, a time division unit 81, a cross-correlation function calculation unit 82, an average / variance calculation unit 83, an identification boundary learning unit 84, and a leakage presence / absence determination unit 95. The leaked sound / noise signal input unit 80 inputs a plurality of leaked sounds and noises different from the evaluation target signal input by the signal input unit 10. Leakage sound is measured by installing sensors at two points on the pipeline across the leak point. Noise is measured by sensors installed at two points on the pipeline where there is no leakage in the vicinity. The time division unit 81 and the cross-correlation function calculation unit 82 are respectively connected to the time division unit 11 and the cross-correlation function calculation unit 12 of the first embodiment with respect to the signal input from the leaked sound / noise signal input unit 80. A similar operation is performed.

The average / dispersion calculation unit 83 calculates a function M (τ) that gives an average value of a cross-correlation function and a function V (τ) that gives a dispersion value from Equations 3 and 4, respectively, for each leaked sound and noise. The discrimination boundary learning unit 84 calculates a support vector machine (SVM), a neural network, and a k-neighbor discriminator from values of M (τ) and V (τ) of the leaked sound and noise obtained by the average / variance calculation unit 83. A discriminating boundary on a two-dimensional space (average / dispersion graph) is learned using a discriminator such as. The leakage presence / absence determination unit 95 determines that there is leakage when the average value and the variance value of the evaluation target signal calculated by the average / variance calculation unit 13 are on the leakage sound side with respect to the identification boundary, and the noise side It is determined that there is no leakage.

FIG. 16 is a conceptual diagram showing the identification boundary learned by the identification boundary learning unit 84. The horizontal axis in FIG. 16 is the average value, and the vertical axis is the variance value. The leaked sound and noise prepared in advance are indicated by black circles and white circles, respectively. The dotted line is the identification boundary learned from leaked sound and noise. If the evaluation signal is in the shaded area, it is determined that there is a leak, and if it is in any other area, it is determined that there is no leak.

As described above, the leak detection device A6 according to the sixth embodiment of the present invention can perform determination with higher accuracy in addition to the effects of the first embodiment.

[Seventh embodiment]
Next, a seventh embodiment based on the leakage detection device according to the second embodiment described above will be described. In the following description, the characteristic part according to the seventh embodiment will be mainly described. In that case, about the structure similar to 2nd embodiment mentioned above, the overlapping description is abbreviate | omitted by attaching | subjecting the same reference number.

The leakage presence / absence determination unit 25 of the second embodiment determines the presence / absence of leakage by comparing the mode value and the cross-correlation function with a threshold value. In the seventh embodiment of the present invention, instead of this, an identification boundary on the histogram of the cross-correlation function based on a plurality of leaked sound signals (leakage sound information) and noise signals (noise information) acquired in advance. And determining whether or not there is a leak depending on which side the mode value of the signal to be evaluated is located with respect to the identification boundary. Here, the identification boundary is a boundary line that defines each region when there is leakage and when there is no leakage. That is, the identification boundary is a set of combinations of the interval value and the appearance frequency (points on the histogram of the cross-correlation function) that serve as a determination boundary for the presence or absence of leakage.

As shown in FIG. 9, the leak detection device A7 according to the seventh embodiment of the present invention is a leak sound / noise signal input unit instead of the leak presence / absence determination unit 25 in the leak detection device A2 of the second embodiment. 80, a time division unit 81, a cross-correlation function calculation unit 82, a mode value calculation unit 63, an identification boundary learning unit 64, and a leakage presence / absence determination unit 85.

As described in the sixth embodiment, the leaked sound / noise signal input unit 80, the time division unit 81, and the cross-correlation function calculation unit 82 respectively calculate a cross-correlation function for a plurality of leaked sounds and noises. calculate.

The mode value calculation unit 63 counts the number of times (frequency) C (τ, i) that the cross-correlation function φ _n (τ) satisfies the inequality of Equation 7 for each τ and each i, and then, as shown in Equation 12. , C (τ, i) maximum value (mode) C _m ′ is calculated.
[Equation 12]

At the same time, the mode value calculation unit 63 also calculates the cross-correlation function φ _n (τ ′, i ′) from τ = τ ′, i = i ′ when the mode value C _m ′ is taken.

The discriminating boundary learning unit 64 uses the mode C _m ′ of the noise and leaked sound obtained by the mode calculation unit 63 and the value of the cross-correlation function φ _n (τ ′, i ′) at that time. Using a classifier such as a support vector machine, a neural network, or a k-neighbor classifier, a classification boundary on a two-dimensional space (cross correlation function histogram) is learned. The leakage presence / absence determination unit 85 determines that there is leakage when the mode value of the evaluation target signal calculated by the mode value calculation unit 23 is on the leakage sound side with respect to the identification boundary, and is on the noise side. It is determined that there is no leakage.

FIG. 5 is a conceptual diagram showing the histogram of the cross-correlation function calculated from the evaluation target signal and the identification boundary learned in advance. The horizontal axis in FIG. 5 is the value of the cross-correlation function φ (τ), and the vertical axis is the value obtained by normalizing the frequency C (τ, i) by the section width S. In the example shown in FIG. 5, since the histogram of the signal to be evaluated enters the hatched area surrounded by the identification boundary, it is determined that there is a leak.

As described above, the leak detection device A7 according to the seventh embodiment of the present invention can perform the determination with higher accuracy in addition to the effects of the second embodiment.

[Eighth embodiment]
Next, an eighth embodiment based on the leakage detection apparatus according to the first embodiment described above will be described. In the following description, the characteristic part according to the eighth embodiment will be mainly described. At this time, the same reference numerals are assigned to the same configurations as those in the first embodiment described above, and redundant description is omitted.

FIG. 17 is a block diagram showing the configuration of the leak detection apparatus according to the eighth embodiment of the present invention. The leak detection device A8 according to the eighth embodiment of the present invention includes a leak presence / absence determination unit 15 and a leak position calculation unit 17.

The leakage presence / absence determination unit 15 determines that there is leakage when the value of the cross-correlation function between input signals input from a plurality of sensors is equal to or greater than a threshold value and the time change is equal to or less than the threshold value.

The leakage position calculation unit 17 calculates the propagation time difference of the leaked sound based on the location where the value of the cross-correlation function is equal to or greater than the threshold value and the time change is equal to or less than the threshold value, and specifies the leakage position based on the propagation time difference.

As described above, the leak detection device A8 according to the eighth embodiment of the present invention has a large value of the cross-correlation function and a change over time by utilizing the steadiness of the leaked sound and the correlation between the plurality of sensors. If small, it is determined that there is a leak. Then, the leak detection device A8 calculates the propagation time difference of the leaked sound based on the location where the value of the cross-correlation function is large and the time change is small, and specifies the leak position based on the propagation time difference. For this reason, the leak detection apparatus A8 has an effect that the presence / absence of the leak and the leak position can be specified even in an environment where noise is generated.

[Hardware configuration of leak detection device]
Hereinafter, a hardware configuration example capable of realizing the leakage detection device according to each embodiment described above will be described.

FIG. 18 is a block diagram showing an example of elements constituting the computer. A computer 900 in FIG. 18 includes a CPU (Central Processing Unit) 910, a RAM (Random Access Memory) 920, a ROM (Read Only Memory) 930, a hard disk drive 940, and a communication interface 950.

The components of the leak detection devices A1, A2, A3, A3 ′, A4, A5, A6, A7, and A8 described above are realized by executing a program (software program, computer program) in the CPU 910 of the computer 900. Also good. Specifically, the

time division units

11 and 81 that are the components described in FIGS. 1, 3, 6A, 6B, 7, 8, 9, 12, and 17 described above, Correlation

function calculation units

12, 82, average /

variance calculation units

13, 53, 83, leakage presence /

absence determination units

15, 25, 85, 95, propagation time

difference calculation units

16, 26, leakage position calculation unit 17, The mode

value calculation units

23 and 63, the signal level calculation unit 30, the evaluation section extraction unit 31, and the identification

boundary learning units

64 and 84 read the program read by the CPU 910 from the ROM 930 or the hard disk drive 940. For example, it may be realized by execution by the CPU 910 as in the procedures of the flowcharts shown in FIGS. In such a case, the present invention described by using the above-described embodiment as an example can be applied to a computer-readable storage medium (for example, hard disk drive 940 or the like) that stores a code representing the computer program or a code representing the computer program. It can be understood that it is constituted by a removable magnetic disk medium, optical disk medium, memory card, etc. (not shown).

Alternatively, the

time division units

11, 81, the cross-correlation

function calculation units

12, 82, the average /

variance calculation units

13, 53, 83, the leakage presence /

absence determination units

15, 25, 85, 95, and the propagation time

difference calculation unit

16 , 26, leakage position calculation unit 17, mode

value calculation units

23 and 63, signal level calculation unit 30, evaluation interval extraction unit 31, and identification

boundary learning units

64 and 84 are realized by dedicated hardware. May be. Further, the leakage detection devices A1, A2, A3, A3 ′, A4, A5, A6, A7, and A8 may be dedicated hardware including these components.

Some or all of the above embodiments may be described as in the following supplementary notes, but are not limited to the following.

(Appendix 1)
Using signals representing the detection results of a plurality of sensors, the value of the cross-correlation function of these signals is calculated, the calculated value is equal to or greater than the first threshold value, and the time change of the calculated value at different measurement times is the first. A leakage presence / absence determination means for determining that there is a leakage when the threshold is equal to or less than 2,
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. A leak detection device comprising: a leak position calculation means for specifying a leak position based on the propagation time difference.

(Appendix 2)
Supplementary note 1 wherein the leakage presence / absence determining means determines that there is leakage when an average value of the cross-correlation function values over different measurement times is equal to or greater than a third threshold value and a variance value is equal to or less than a fourth threshold value. Leak detection device.

(Appendix 3)
The leak presence / absence determining means has a leak when a section value of the cross-correlation function assigned for each predetermined section width is equal to or greater than a fifth threshold and the frequency of appearance of the section value is equal to or greater than a sixth threshold. The leakage detection device according to supplementary note 1, wherein

(Appendix 4)
The leakage detection device according to supplementary note 2, wherein the leakage presence / absence determination means calculates an average value from remaining values obtained by removing a predetermined number of values in order from the maximum value of the cross-correlation function and a predetermined number of values in order from the minimum value.

(Appendix 5)
Based on a plurality of leaked sound information and noise information acquired in advance, an identification boundary learning means for learning an identification boundary that is a set of combinations of the average value and the variance value, which is a boundary for determining whether there is leakage In addition,
The leakage detection apparatus according to appendix 2 or 4, wherein the leakage presence / absence determination means determines whether there is leakage based on the identification boundary.

(Appendix 6)
An identification boundary learning unit that learns an identification boundary that is a set of combinations of the section value and the appearance frequency, which is a boundary for determining whether there is leakage, based on a plurality of leaked sound information and noise information acquired in advance. In addition,
The leakage detection device according to supplementary note 3, wherein the leakage presence / absence determination means determines whether there is leakage based on the identification boundary.

(Appendix 7)
The discrimination boundary learning means learns the discrimination boundary using a discriminator that is one of a support vector machine, a neural network, and a k-neighbor discriminator,
The leakage presence / absence determining means determines that there is leakage when the mode value of the input signal to be evaluated is on the leakage sound side with respect to the identification boundary, and determines that there is no leakage when on the noise side. 5. The leak detection device according to 5 or 6.

(Appendix 8)
A cross-correlation function calculating unit that extracts a section whose level is equal to or less than a threshold value from the input signal, divides the extracted section into fixed time intervals, and calculates a cross-correlation function between the divided input signals; The leak detection device according to any one of appendices 1 to 7.

(Appendix 9)
9. The leak detection device according to appendix 8, further comprising signal input means for inputting the input signal for each time interval M during the period N (where time interval M> period N).

(Appendix 10)
Using signals representing the detection results of multiple sensors, calculate the value of the cross-correlation function of those signals,
It is determined that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at different measurement times is equal to or less than a second threshold;
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. ,
A leak detection method for specifying a leak position based on the propagation time difference.

(Appendix 11)
The leak detection method according to appendix 10, wherein it is determined that there is a leak when an average value over different measurement times of the cross-correlation function value is equal to or greater than a third threshold value and a variance value is equal to or less than a fourth threshold value.

(Appendix 12)
Item 11. The supplementary note 10, wherein it is determined that there is a leak when the interval value of the cross-correlation function assigned for each predetermined interval width is equal to or greater than a fifth threshold value and the frequency of occurrence of the interval value is equal to or greater than the sixth threshold value. Leak detection method.

(Appendix 13)
12. The leakage detection method according to claim 11, wherein an average value is calculated from a remaining value obtained by excluding a predetermined number of values in order from the maximum value of the cross-correlation function and from a minimum value.

(Appendix 14)
Based on a plurality of leaked sound information and noise information acquired in advance, learn an identification boundary that is a set of a combination of the average value and the variance value, which is a boundary for determining whether there is leakage,
14. The leak detection method according to

appendix

11 or 13, wherein the presence / absence of a leak is determined based on the identification boundary.

(Appendix 15)
Based on a plurality of leaked sound information and noise information acquired in advance, learning an identification boundary that is a set of combinations of the section value and the appearance frequency, which is a boundary for the presence or absence of leakage,
The leakage detection method according to appendix 12, wherein the presence or absence of leakage is determined based on the identification boundary.

(Appendix 16)
Learning the discrimination boundary using a discriminator that is one of a support vector machine, a neural network, or a k-neighbor discriminator;
The leak detection according to appendix 14 or 15, wherein when the mode value of the input signal to be evaluated is on the leaky sound side with respect to the identification boundary, it is determined that there is a leak, and when it is on the noise side, it is determined that there is no leak Method.

(Appendix 17)
Extracting a section whose level is below a threshold from the input signal;
Divide the extracted interval into fixed time intervals,
The leakage detection method according to any one of supplementary notes 10 to 16, wherein a cross-correlation function between the divided input signals is calculated.

(Appendix 18)
18. The leakage detection method according to appendix 17, wherein the input signal is input for each time interval M during a period N (where time interval M> period N).

(Appendix 19)
On the computer,
Using a signal representing the detection results of a plurality of sensors, calculating a value of a cross-correlation function of these signals,
A process of determining that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at a different measurement time is equal to or less than a second threshold;
A leakage sound propagation time difference is calculated based on a location where the cross-correlation function value is equal to or greater than the first threshold value and the time change of the cross-correlation function value at different measurement times is equal to or less than the second threshold value. Processing,
The program which performs the process which specifies a leak position based on the said propagation time difference.

(Appendix 20)
On the computer,
The program according to appendix 19, which executes a process of determining that there is a leak when an average value of the cross-correlation function values during a predetermined period is equal to or greater than a third threshold value and a variance value is equal to or less than a fourth threshold value.

(Appendix 21)
On the computer,
A process of determining that there is a leak when the section value of the cross-correlation function assigned for each predetermined section width is equal to or greater than a fifth threshold and the frequency of appearance of the section value is equal to or greater than a sixth threshold is executed. The program according to appendix 19.

(Appendix 22)
On the computer,
21. The program according to appendix 20, which executes a process of calculating an average value from a remaining value obtained by excluding a predetermined number of values in order from a maximum value of the cross-correlation function in order from a minimum value.

(Appendix 23)
On the computer,
Based on a plurality of leaked sound information and noise information acquired in advance, a process for learning an identification boundary that is a set of combinations of the average value and the variance value, which is a boundary for determining whether there is leakage,
The program according to appendix 20 or 22, which executes a process of determining the presence or absence of leakage based on the identification boundary.

(Appendix 24)
On the computer,
Based on a plurality of leaked sound information and noise information acquired in advance, a process for learning an identification boundary that is a set of combinations of the section value and the appearance frequency, which is a boundary for determining whether there is leakage,
The program according to appendix 21, which executes a process of determining the presence or absence of leakage based on the identification boundary.

(Appendix 25)
On the computer,
Learning the discrimination boundary using a discriminator that is one of a support vector machine, a neural network, and a k-neighbor discriminator;

Appendix

23 or 23 for executing a process of determining that there is a leak when the mode value of the input signal to be evaluated is on the leaky sound side with respect to the identification boundary and determining that there is no leak when on the noise side 24. The program according to 24.

(Appendix 26)
On the computer,
A process of extracting a section whose level is equal to or lower than a threshold value from the input signal;
A process of dividing the extracted section into a certain time interval;
The program according to any one of appendices 19 to 25, wherein the program executes a process of calculating a cross-correlation function between the divided input signals.

(Appendix 27)
On the computer,
27. The program according to appendix 26, wherein a process for inputting the input signal is executed for each time interval M during a period N (where time interval M> period N).
While the present invention has been described with reference to the embodiments (and examples), the present invention is not limited to the above embodiments (and examples). Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.
This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2014-063543 for which it applied on March 26, 2014, and takes in those the indications of all here.

A1, A2, A3, A3 ', A4, A5, A6, A7, A8

Leakage detection device

10, 40

Signal input unit

11, 81

Time division unit

12, 82 Cross correlation

function calculation unit

13, 53, 83 Average /

variance calculation

15, 25, 85, 95 Leakage presence /

absence determination unit

16, 26 Propagation time difference calculation unit 17 Leakage

position calculation unit

23, 63 Mode value calculation unit 30 Signal level calculation unit 31 Evaluation

interval extraction unit

64, 84 Discrimination boundary learning unit 80 Leaked sound / noise signal input unit 100A, 100B Sensor 110 Leakage point 900 Computer 910 CPU
920 RAM
930 ROM
940 Hard disk drive 950 Communication interface

Claims

Using signals representing the detection results of a plurality of sensors, the value of the cross-correlation function of these signals is calculated, the calculated value is equal to or greater than the first threshold value, and the time change of the calculated value at different measurement times is the first. A leakage presence / absence determination means for determining that there is a leakage when the threshold is equal to or less than 2,
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. A leak detection device comprising: a leak position calculation means for specifying a leak position based on the propagation time difference.
The leak presence / absence determining means determines that there is a leak when an average value over a different measurement time of the cross-correlation function value is not less than a third threshold value and a variance value is not more than a fourth threshold value. The leakage detection device described.
The leak presence / absence determining means has a leak when a section value of the cross-correlation function assigned for each predetermined section width is equal to or greater than a fifth threshold and the frequency of appearance of the section value is equal to or greater than a sixth threshold. The leak detection device according to claim 1, which is determined as follows.
3. The leak detection apparatus according to claim 2, wherein the leak presence / absence determining means calculates an average value from the remaining values obtained by removing a predetermined number of values in order from the maximum value of the cross-correlation function and a predetermined number of values in order from the minimum value. .
Based on a plurality of leaked sound information and noise information acquired in advance, an identification boundary learning means for learning an identification boundary that is a set of combinations of the average value and the variance value, which is a boundary for determining whether there is leakage In addition,
The leak detection device according to claim 2, wherein the leak presence / absence determination unit determines whether there is a leak based on the identification boundary.
An identification boundary learning unit that learns an identification boundary that is a set of combinations of the section value and the appearance frequency, which is a boundary for determining whether there is leakage, based on a plurality of leaked sound information and noise information acquired in advance. In addition,
The leakage detection device according to claim 3, wherein the leakage presence / absence determination unit determines whether there is leakage based on the identification boundary.
The discrimination boundary learning means learns the discrimination boundary using a discriminator that is one of a support vector machine, a neural network, and a k-neighbor discriminator,
The leakage presence / absence determination means determines that there is leakage when the mode value of the input signal to be evaluated is on the leakage sound side with respect to the identification boundary, and determines that there is no leakage when on the noise side. Item 7. The leakage detection device according to Item 5 or 6.
A cross-correlation function calculating unit that extracts a section whose level is equal to or less than a threshold value from the input signal, divides the extracted section into fixed time intervals, and calculates a cross-correlation function between the divided input signals; The leak detection apparatus according to any one of claims 1 to 7.
9. The leak detection apparatus according to claim 8, further comprising a signal input means for inputting the input signal for each time interval M during a period N (however, time interval M> period N).
Using signals representing the detection results of multiple sensors, calculate the value of the cross-correlation function of those signals,
It is determined that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at different measurement times is equal to or less than a second threshold;
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. ,
A leak detection method for specifying a leak position based on the propagation time difference.
11. The leak detection method according to claim 10, wherein it is determined that there is a leak when an average value through measurement times having different values of the cross-correlation function is not less than a third threshold value and a variance value is not more than a fourth threshold value.
The determination is made that there is a leak when an interval value of the cross-correlation function allocated for each predetermined interval width is equal to or greater than a fifth threshold value and an appearance frequency of the interval value is equal to or greater than a sixth threshold value. Leak detection method.
Based on a plurality of leaked sound information and noise information acquired in advance, learn an identification boundary that is a set of a combination of the average value and the variance value, which is a boundary for determining whether there is leakage,
The leakage detection method according to claim 11, wherein presence / absence of leakage is determined based on the identification boundary.
On the computer,
Using signals representing the detection results of multiple sensors, calculate the value of the cross-correlation function of those signals,
It is determined that there is a leak when the value of the cross-correlation function is equal to or greater than a first threshold and the time change of the calculated value at different measurement times is equal to or less than a second threshold;
A difference in propagation time of leaked sound is calculated based on a location where the value of the cross-correlation function is not less than the first threshold and the time change of the value of the cross-correlation function at different measurement times is not more than the second threshold. ,
Identifying a leak location based on the propagation time difference;
A recording medium for storing a program for executing the above.