CN112525437A

CN112525437A - Underwater identification method for leakage noise of large-scale water delivery building

Info

Publication number: CN112525437A
Application number: CN202011372320.1A
Authority: CN
Inventors: 刘毅; 商峰; 黄涛; 冯少孔; 宋文杰
Original assignee: China Institute of Water Resources and Hydropower Research
Current assignee: China Institute of Water Resources and Hydropower Research
Priority date: 2020-11-30
Filing date: 2020-11-30
Publication date: 2021-03-19
Anticipated expiration: 2040-11-30
Also published as: CN112525437B

Abstract

The invention relates to an underwater identification method for leakage noise of a large-scale water delivery building, which is characterized by comprising the following steps of: 1) continuously collecting leakage noise generated when a large-scale water delivery building leaks to obtain an underwater sound signal; 2) extracting a continuous spectrum from the collected underwater sound signal by adopting an iteration method to obtain a continuous spectrum of the leakage noise; 3) and identifying and analyzing the duration and the on-way distribution mode of the extracted continuous spectrum of the leakage noise, and positioning the leakage point of the large-scale water delivery building according to the identification and analysis result. The invention identifies the leakage signal by extracting the continuous spectrum in the underwater mobile detection data, and can position the leakage source according to the on-way distribution result of the actually measured leakage noise signal, so that the positioning result is more accurate. The invention can be widely applied to the field of leakage noise detection of large-scale water delivery buildings.

Description

Underwater identification method for leakage noise of large-scale water delivery building

Technical Field

The invention relates to an underwater identification method for leakage noise of a large-scale water delivery building, and belongs to the field of engineering detection of water delivery buildings.

Background

The large-scale water delivery building comprises aqueducts, inverted siphons, hidden culverts (canals), tunnels, large-scale pipelines and other structural types, is a key node of water transfer engineering, and has important significance for guaranteeing water supply safety and public safety in operation safety. In the long-term operation process of large-scale water delivery buildings, water stop damage, concrete cracking and other reasons at joints generate leakage, the safe operation of the structure is influenced, and the cases of causing the structure damage are frequently rare. Therefore, it is necessary to periodically detect leakage defects of large water transport buildings in service and to repair and reinforce the large water transport buildings on a selective basis.

For a long time, the service state of most water delivery buildings needs to be emptied periodically for service state detection. On the one hand, however, with the development of economy, the social requirements on the water supply guarantee rate are increased, the time for emptying detection is shorter and shorter, the time interval between two adjacent emptying detections is longer and longer, and the requirements on water filling or detection in an operating state are stronger and stronger. On the other hand, the detection and judgment difficulty of key factors influencing leakage, such as the reliability of water-stopping and water-proof sealing, whether cracks penetrate and the like, is high under the emptying condition, and the leakage defect is easier to find under the water-filling or running state. Therefore, it is a necessary choice to develop a leak defect detection method in a water-filled or operating state.

Electrical geophysical prospecting is a common leakage defect detection method and is commonly used in reservoir slope, dike and other geotechnical engineering. However, most of large water delivery buildings are thin-wall concrete structures, and the leakage defects of the structures are detected by using an electrical method and geophysical prospecting in few cases. When a large water delivery building leaks in the operation process, a local flow field near a leakage point can be changed, and the leakage point can be positioned by detecting the flow velocity perpendicular to the wall surface to be detected. The Du nationality's level develops a three-dimensional flow velocity vector sonar measuring system, and positions the leakage points by detecting and analyzing the local change of the flow field; the method is derived from a logging method in hydrogeological geophysical prospecting, and is particularly suitable for seepage detection with small flow velocity. The visual-acoustic integrated comprehensive detection method for the face rockfill dam is developed by combining a sonar method, underwater robot high-definition camera shooting, ink-jet tracing, poplar nobility and the like; the method needs to enable the equipment to be close to the panel to be detected, so that the method is suitable for fine detection under the condition that the leakage area is approximately known, and is not high in efficiency for general leakage defect investigation of long-distance water delivery buildings.

Unlike rock fill dam panel damage, the leakage point area of large water transport buildings is much smaller, and the leakage flow rate can be much larger, also referred to as leakage. Leakage not only changes the local flow field, but such local changes propagate in the form of wave motion all around and excite stress waves in the structure carrying the water flow. Thus, the leak point can be identified and located based on acoustic detection principles. Similar research works develop earlier in oil gas pipeline, municipal water supply pipe-line engineering, and the achievement of obtaining is more. In general, existing acoustic detection methods include: negative pressure wave method, stress wave method, transient detection method, and leakage noise detection method.

Compared with oil and gas conveying pipelines and municipal water supply pipelines, the large-scale water conveying building has the advantages that the structural size and the self weight are much larger, the conveying pressure is much smaller, and flexible waterproof connection is usually formed between structural sections; therefore, the energy of the negative pressure wave and the stress wave caused by local leakage of a large water delivery building is small, the attenuation is fast in the propagation process, and the detection and the positioning are difficult to use. The negative pressure wave method and the stress wave method belong to a passive acoustic detection technology, the transient detection method belongs to an active acoustic detection technology, a water hammer pressure wave is manufactured through a control valve, and water pressure fluctuation passing through a leakage point is analyzed so as to identify and position the leakage point. The transient detection method is mainly used for medium and small caliber water pipeline systems at present.

The fluid acoustic signal frequency domain scope that the leakage arouses is wider, and the negative pressure wave belongs to the infrasonic wave category essentially, detects the fluid acoustic signal of higher frequency and also can fix a position minute leak point. A smart ball developed by Pure Technologies, canada (which was purchased by Xylem, inc. in 2018) may be implemented to identify and locate leaks by detecting acoustic signals propagating within the fluid as it moves within the pipe. Similar spherical internal detectors are also developed by Tianjin university in China, and internal sensors comprise hydrophones, accelerometers, magnetometers, gyroscopes and the like, wherein the hydrophones are used for recording fluid acoustic signals; the intensity and frequency distribution of the fluid acoustic signal may exhibit different characteristics as the detector passes through the leak. At present, relevant research at home and abroad is only limited to oil and gas pipeline leakage detection, and the method and the precision research for detecting the leakage of a large-scale water delivery building are blank. Obviously, the frequency spectrum characteristics of the leakage noise of a large-scale water delivery structure are greatly different from those of an oil-gas pipeline, a municipal water supply pipeline and the like, and the oil-gas pipeline leakage detection method cannot be directly used.

Disclosure of Invention

In view of the above problems, the present invention aims to provide an underwater identification method for leakage noise of a large water transport building, which takes a large water filling aqueduct as an example, and develops a principle experiment of underwater movement detection in a non-pressure runner leakage state, researches a method for identifying and positioning a leakage point, and lays a foundation for developing an underwater leakage detection system of a large water transport building.

In order to achieve the purpose, the invention adopts the following technical scheme: an underwater identification method for leakage noise of a large water delivery building comprises the following steps: 1) continuously collecting leakage noise generated when a large-scale water delivery building leaks to obtain an underwater sound signal; 2) carrying out continuous spectrum extraction on the collected underwater sound signals by adopting an iteration method to obtain a continuous spectrum of leakage noise; 3) and identifying and analyzing the duration and the on-way distribution mode of the extracted continuous spectrum of the leakage noise, and positioning the leakage point of the large-scale water delivery building according to the identification and analysis result.

Further, in the step 1), the method for continuously collecting leakage noise generated when a large water delivery building leaks comprises the following steps:

1.1) determining the position of a leakage point which is possibly leaked by a large-scale water delivery building according to actual experience, and determining a data acquisition track according to the position of the leakage point;

1.2) uniformly arranging a plurality of measuring points along the data acquisition track, and continuously acquiring the underwater sound signals at the positions of the measuring points by adopting a hydrophone array according to a preset speed to obtain the underwater sound signals.

Further, in the step 1.2), the hydrophone arrays are arranged at preset intervals Δ d, and the preset intervals Δ d are 0.5m or 1 m.

Further, in the step 1.2), the hydrophone array includes not less than 3 hydrophones.

Further, in step 1.2), the hydrophone array includes 12 or 24 hydrophones.

Further, in the step 1.2), the acquisition frequency of the hydrophone array is 0.1Hz to 15MHz when the hydrophone array acquires at each measurement point.

Further, in the step 1.2), the measuring point interval is 1m, 3m or 5m according to the size of the test field.

Further, in the step 2), the method for extracting the continuous spectrum signal comprises the following steps:

2.1) averaging the measurement value results of each hydrophone in the hydrophone array, performing truncation sampling on the obtained average value data by adopting a rectangular frame, and performing short-time fast Fourier analysis on the truncation sampling data;

2.2) adopting an iteration method to extract a continuous spectrum of the short-time fast Fourier analysis result obtained in the step 2.1) to obtain a continuous spectrum of the leakage noise.

Further, in the step 2.2), the formula for performing the continuous spectrum extraction is as follows:

in the formula, H_i(f_j) The ith iteration result is shown, and a subscript j shows a jth frequency point obtained by short-time fast discrete Fourier analysis; a. the_i-1,jRepresenting the follow-up mean value of the frequency band signal intensity with the center length of 2n by the jth frequency point of i-1 iteration; m represents a multiple, and m is more than or equal to 1; k is a radical of_dDisplay folderCoefficient of subtraction, and 0 < k_d≤1。

Further, in the step 3), the duration and the distribution of the leakage noise signal along the way are:

in the formula, H (s, t) represents the time-duration curve of the measured leakage noise, P(s) represents the distribution of the measured leakage noise signal power along the way, s represents the position coordinate of the sensor, and sigma, mu and p₀And k is a parameter to be determined, wherein mu indicates the position coordinate of the leakage source, and delta t represents the window duration for calculating the power.

Due to the adoption of the technical scheme, the invention has the following advantages: (1) theoretical analysis shows that the frequency spectrum of the leakage noise of the large-scale water delivery structure comprises a continuous spectrum and reflects the randomness of turbulence and cavitation; the method also comprises a line spectrum with higher energy, and reflects the friction effect of the leakage flow and the pipe wall structure, so that the leakage signal is identified by extracting the continuous spectrum in the underwater movement detection data, and the leakage source positioning method is simple and reliable. (2) According to the invention, the leakage source can be positioned according to the actual measurement along-the-way distribution result of the leakage noise signal, and the positioning result is more accurate. (3) The method extracts the continuous spectrum in the underwater movement detection data and analyzes the distribution of the intensity along the way, so that the positioning precision of the simulated leakage source with the leakage flow rate of more than 2L/s is 0.9 m; the simulation leakage source with the leakage flow rate less than 0.5L/s cannot be identified and positioned; thereby verifying the feasibility and reliability of the method of the invention. The invention can be widely applied to the field of safety detection of large-scale water delivery buildings.

Drawings

FIG. 1 is the extraction of leakage noise of the present invention, and FIG. 1(a) is the original signal spectrum; FIG. 1(b) is a leakage noise spectrum;

FIG. 2 is a water-filled flume test case of the present invention, wherein FIG. 2(a) is a schematic plan view and FIG. 2(b) is a schematic elevation view; fig. 2(c) is a schematic illustration of the encoding of each hydrophone on the hydrophone array.

FIG. 3 is a time-frequency map of a motion-detected underwater acoustic signal of the present invention, wherein FIG. 3(a) is after band-pass filtering of raw data and FIG. 3(b) is after extraction of a leakage noise continuum;

FIG. 4 is a schematic view of the location of a leak;

fig. 5 is an underwater identification roadmap of large water-filled aqueduct leakage noise.

Detailed Description

The invention is described in detail below with reference to the figures and examples.

The invention provides an underwater identification method for leakage noise of a large-scale water delivery building, which comprises the following steps:

1) and continuously collecting leakage noise generated when the large water delivery building leaks to obtain an underwater sound signal.

Research shows that the acoustic signals generated when the water pipeline leaks are derived from the following sources: the change of the flow field near the leakage point causes turbulence, leakage flow and pipe wall friction and cavitation phenomena. The intensity and frequency distribution of acoustic signals caused by the three phenomena are related to factors such as flow velocity in a pipeline, conveying pressure, gas content in water flow, shape, size and roughness of a pipe wall structure and a discharge opening. Wherein most of the energy of low frequency will propagate along the water body, and the energy of high frequency will propagate along the solid and the water body simultaneously. Comparing the occurrence mechanisms of the three phenomena, it can be known that turbulence and cavitation noise have randomness, but still belong to a stable process within a period of time; the friction between the leakage flow and the pipe wall is related to the natural vibration characteristic of the structure, and the regularity is obvious. Therefore, the frequency spectrum of the leakage noise comprises a continuous spectrum and reflects the randomness of turbulence and cavitation; and also contains a discrete spectral line with higher energy, reflecting the effect of the leakage flow rubbing against the pipe wall.

For large water transport structures, the factors that induce structural vibration include not only the leakage process, but also a number of other environmental factors. In general, the acoustic signals generated by the leakage flow and the structural friction are difficult to separate from the environmental noise. Therefore, identifying the continuum in the leakage noise is of greater significance.

For a water pipeline, when a leakage noise identification method is verified through experiments, a leakage process is simulated by opening and closing a valve on a branch pipe. For large water transport buildings, the condition for connecting branch pipes is not generally provided. Research shows that when the water pump works, water pressure pulsation signals near the near field of a water inlet of the water pump also comprise two types, wherein one type is related to near-field turbulence and cavitation, has randomness and is represented as a continuous spectrum, and the other type is related to mechanical motion of blades and a transmission shaft in the water pump, has strong regularity and is represented as a line spectrum with high energy. Because the near-field noise of the water inlet of the water pump has certain similarity with the leakage noise, the invention adopts the water pumping process of the water pump to simulate the leakage source.

Specifically, when continuously collecting leakage noise generated when a large-scale water delivery building leaks, the method comprises the following steps:

2) And (4) extracting a continuous spectrum from the collected underwater sound signal by adopting an iteration method to obtain a continuous spectrum of the leakage noise.

Specifically, the method for extracting the continuous spectrum signal comprises the following steps:

2.1) averaging the measurement value results of each hydrophone in the hydrophone array, performing truncation sampling on the obtained average value data by adopting a rectangular frame, and performing short-time fast Fourier analysis on the truncation sampling data.

2.2) adopting an iteration method to extract a continuous spectrum from the short-time fast Fourier analysis result obtained in the step 2.1), wherein the calculation formula is as follows:

in the formula, H_i(f_j) The ith iteration result is shown, and a subscript j shows a jth frequency point obtained by short-time fast Fourier analysis; a. the_i-1,jRepresenting the follow-up mean value of the signal intensity of the frequency band with the center length of 2n by the jth frequency point; m represents a multiple, and m is more than or equal to 1; k is a radical of_dRepresents a reduction coefficient, and 0 < k_dLess than or equal to 1; m and k in the invention_dAll taken as 1.0.

As shown in fig. 1(a) and 1(b), fig. 1(a) is a frequency spectrum of an original underwater sound signal, fig. 1(b) is a frequency distribution diagram of the intensity of a leakage noise signal obtained through 100 iterations, and fig. 1(a) and 1(b) are respectively subjected to normalization processing. It can be seen that the smoothness of the spectral distribution curve of the leakage noise in fig. 1(b) is significantly improved.

3) And identifying and analyzing the duration and the on-way distribution mode of the extracted continuous spectrum of the leakage noise signal, and positioning the leakage point of the large-scale water delivery building according to the identification and analysis result.

The leakage noise has a sharp temporal distribution, and when the leakage amount does not change much, the intensity of the leakage noise should be kept in a stable continuous distribution on the time axis. In this way, leakage noise and other short-term and sporadic interfering vibrations can be distinguished.

As shown in fig. 1(a), for a single leakage source, the distribution of the intensity of the leakage noise along the way should be attenuated towards the two sides of the flow channel by taking the leakage source as the center, and then the leakage source can be positioned according to the measured distribution result of the leakage noise along the way. Assuming that the leakage source signal intensity distribution along the way is gaussian, fitting the signal intensity distribution along the way measured when the hydrophone array passes near the leakage source as follows:

wherein H (s, t) represents the extracted leakage noise duration curve, P(s) represents the on-way distribution of the measured signal power, s represents the position coordinate of the hydrophone, and sigma, mu and p₀K is a parameter to be determined, wherein mu indicates the position coordinate of the leakage source; Δ t represents the window duration for calculating power.

Example one

1 method of experiment

In the embodiment, a double river aqueduct with a central line adjusted in north and south water is selected to carry out an on-site detection experiment. The total length 810m of the aqueduct comprises an inlet and outlet transition section, an inlet and outlet lock chamber section, an inlet and outlet connecting section and a channel body section. Wherein, the length of the groove body section is 600m, the groove body section is a beam type aqueduct, and 20 spans are formed in every 30 m. The groove body is in two-connection with four holes, the net width of a single hole is 7.0m, the net height is 7.45m, and 2 holes are in 1-connection and 2-connection. The groove body is a prestressed concrete rectangular groove, and the section of the groove body is shown in fig. 2(a) and 2 (b). And selecting the east-side 2 nd hole as an experimental flow channel. During the experiment, at the maintenance period, the gates at two ends of the hole flow passage are closed, still water is in the groove, the water depth is 6.1m, and the rest hole flow passages are still filled with water to operate according to the normal water conveying requirement.

As shown in fig. 5, the method for continuously collecting leakage noise generated by an analog leakage source includes the following steps:

first, a simulated leakage source (in the present invention, a water pump) is vertically sunk at each simulated leakage point of the floor of a large water transport building.

Secondly, the flow of each leakage source is set according to the preset leakage strength, and the continuous leakage process is continuously simulated.

And (3) vertically sinking a simulated leakage source into the bottom of the aqueduct, selecting a water pump as the simulated leakage source, and simulating a continuous leakage process when the leakage source flows into the 1 st hole on the left side as shown in fig. 2(a) and shown in fig. 2 (b). In the experiment, two simulated leakage sources 1 and 2 are axially arranged along an aqueduct, and 5m is selected as 2 leakage sources arranged at intervals; through field measurement, the leakage rate of the leakage source 1 is about 0.5L/s, and the leakage rate of the leakage source 2 is 2.0L/s. The leakage sources 1 and 2 are powered by a diesel generator 3.

And thirdly, acquiring the underwater sound signals of the measuring points by adopting the hydrophone array 3 according to a preset speed, wherein the acquired data comprises the underwater sound signals of the measuring points when the leakage is not started and after the leakage is started. Preferably, the hydrophone arrays comprise 12 or 24 hydrophone arrays, and the hydrophone interval can be set to 0.5m or 1m, and the like.

A hydrophone array 3 is adopted for detecting simulated leakage noise, the hydrophone array 3 is customized by Tianjin ocean acoustics, and comprises 12 hydrophones with the spacing of 0.5m, the hydrophone array is suspended at the position of a spool with the water depth of about 3m (half depth of a aqueduct) in the aqueduct, see figure 2(b), and after a simulated leakage source starts to work, the hydrophone array is moved axially along a tank body through manual pulling of the cable to realize the underwater movement detection process. Moving 3m each time, standing for 5 minutes after reaching a preset position, acquiring signals in the state, and encoding each hydrophone in the hydrophone array as shown in figure 2 (c). A Geode24 seismograph 5 is adopted to collect signals along the way, the sampling frequency is 1KHz, and the acquisition instrument is connected with a hydrophone array through a cable to carry out continuous signal acquisition. And (3) starting to collect signals at the position shown in fig. 2(a), before the time D0, starting the simulated leakage source for signal collection without opening the simulated leakage source, starting the simulated leakage source for signal collection at the time D0, moving the hydrophone array by 3m after 5 minutes, collecting signal data at the position D3, and so on, and finishing the signal data collection when the position D60 is detected.

2 results of detection

2.1 identification of leakage noise

First, the leakage signal is identified from the measured data along the way. And averaging the measured values of 12 hydrophones in the hydrophone array, and intercepting the sampling data for short-time fast Fourier analysis. To identify the leakage signal from the frequency domain, the frequency resolution needs to be increased. This embodiment is cut out in rectangular boxes. And removing infrasound signals with lower frequency, and extracting continuous spectrums by applying the formulas (1) and (2) to obtain the time-frequency analysis result of the underwater sound signals measured by the hydrophone array statically arranged near each measuring point.

As shown in fig. 3, the time-frequency analysis results near each measurement point are mapped two-dimensionally when the leakage source is not opened and after the leakage source is opened. For comparison, the image result of band-pass filtering of 10-400 Hz of the original data is also shown in the figure. The tone scale in fig. 3(a) and 3(b) indicates the magnitude of the sound intensity. Since the discrete spectral lines in the raw data have much higher energy than the surrounding continuum. It can be seen that there are a large number of line spectra with more concentrated energy in the raw data, and the time distribution (along-the-way distribution) of the intensity of these line spectra has no obvious regularity, so that it is difficult to deduce the corresponding frequency and position of the leakage signal. This is because the structural vibration caused by the environmental factors corresponds to a line spectrum having high energy, and is mixed with the leakage signal, which is difficult to distinguish.

After extracting the continuous spectrum, it can be found that there is a certain regularity in the distribution of random noise intensity along the course and the distribution of random noise intensity along the duration. In general, in signals measured at D42 and D45, the random noise intensity of the frequency bands of 100-150 Hz and 200-270 Hz is obviously higher than that of other surrounding measuring points, and the signals are continuously distributed on a time axis, so that leakage sources are near D42 and D45, and leakage noise is mainly distributed at 100-150 Hz and 200-270 Hz.

In signals measured at each measuring point D6-D60, the intensity of random noise in a frequency range of 100-150 Hz gradually increases along the way, which shows that a strong interference source exists outside the measuring point D60, and the random noise mainly comes from the mechanical vibration of a field power supply source (a diesel generator); random noise in the frequency band of 100-150 Hz is mixed with leakage signals and interference signals.

In signals measured after a leakage source (pump) is opened at a measuring point D0, the intensity of random noise in frequency ranges of 100-150 Hz and 200-270 Hz is obviously higher than that of signals measured before the leakage source is arranged at the measuring point D0 and at the measuring point D3, which shows that other interference factors exist near the measuring point D0. Upon field inspection, the water flow from point D0 downstream into the left 1 st hole (see FIG. 2b) due to the long length of the simulated source of leakage water pipe. This noise therefore comes from the noise generated when water drawn by the simulated leakage source flows into the left 1 st hole.

2.2 location of leak

And (3) selecting signals measured by each sensor when measuring points D39, D42, D45 and D48 with typical leakage duration intensity, extracting leakage noise in a frequency band of 200-270 Hz, calculating the along-the-way distribution of signal power according to the formula (4), and fitting according to the formula (3), wherein the signals are shown in figure 4. It can be seen that, 12 hydrophones are uniformly distributed in the hydrophone array, the interval between the hydrophones is 0.5m, the hydrophone moves forward 3m each time, the positions of the hydrophones 7 to 12 at the previous measuring point are the same as the positions of the hydrophones 1 to 6 at the next measuring point, and therefore the power measurement value of the hydrophones 7 to 12 at the previous measuring point is the same as that of the hydrophones 1 to 6 at the next measuring point. Similar phenomena that the measured values are connected end to end can be observed at points D36 and D39, points D39 and D42 and points D45 and D48, and the power of the No. 7-12 hydrophone at a point D42 is lower than that of the No. 1-6 hydrophone at a point D45. The transverse position of the hydrophone array on the tank body can be changed under the manual traction state because the traction cable is difficult to keep in a straight state in a still water state and has certain natural bending; when the hydrophone array is far away from the leakage source, the error caused by the transverse position change of the tank body is small, and when the hydrophone array is close to the leakage source, the error caused by the transverse position change of the tank body is large.

Fig. 4 shows the power of the leakage noise measured by each sensor of the hydrophone array at different measuring point positions, and the position of the leakage point can be found to be D36+9.9m, i.e. measuring point D45+0.9m is the position for identifying the leakage, and the difference between the position and the leakage position 2 is 0.9 m. The locations of the two deployed leakage sources and the identified leakage locations are identified in FIG. 4. It can be seen that, because the leakage flow of the leakage source 2 is large, the intensity of the generated noise signal is large, the method can accurately position, and the actual position of the leakage source 2 is 45m, so that the positioning error is 0.9 m; the leakage amount of the leakage source 1 is small, and the generated noise signal intensity is covered by background noise, so that the leakage source cannot be identified and positioned.

The above embodiments are only used for illustrating the present invention, and the structure, connection mode, manufacturing process, etc. of the components may be changed, and all equivalent changes and modifications performed on the basis of the technical solution of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. An underwater identification method for leakage noise of a large-scale water delivery building is characterized by comprising the following steps:

1) continuously collecting leakage noise generated when a large-scale water delivery building leaks to obtain an underwater sound signal;

2) carrying out continuous spectrum extraction on the collected underwater sound signals by adopting an iteration method to obtain a continuous spectrum of leakage noise;

3) and identifying and analyzing the duration and the on-way distribution mode of the extracted continuous spectrum of the leakage noise, and positioning the leakage point of the large-scale water delivery building according to the identification and analysis result.

2. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 1, wherein: in the step 1), the method for continuously collecting the leakage noise generated when the large water delivery building leaks comprises the following steps:

3. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 2, wherein: in the step 1.2), the hydrophone arrays are arranged at preset intervals Δ d, and the preset intervals Δ d are 0.5m or 1 m.

4. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 2, wherein: in the step 1.2), the hydrophone array comprises not less than 3 hydrophones.

5. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 2, wherein: in the step 1.2), the hydrophone array comprises 12 or 24 hydrophones.

6. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 2, wherein: in the step 1.2), the acquisition frequency of the hydrophone is 0.1Hz-15MHz when the hydrophone array is acquired at each measuring point.

7. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 2, wherein: in the step 1.2), the measuring point interval is 1m, 3m or 5m according to the size of a test field.

8. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 1, wherein: in the step 2), the method for extracting the continuous spectrum signal comprises the following steps:

9. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 1, wherein: in the step 2.2), the formula for continuous spectrum extraction is as follows:

in the formula, H_i(f_j) To representThe ith iteration result is obtained, and the subscript j represents the jth frequency point obtained by short-time fast discrete Fourier analysis; a. the_i-1,jRepresenting the follow-up mean value of the frequency band signal intensity with the center length of 2n by the jth frequency point of i-1 iteration; m represents a multiple, and m is more than or equal to 1; k is a radical of_dRepresents a reduction coefficient, and 0 < k_d≤1。

10. The underwater recognition method of the leakage noise of the large water transport building as claimed in claim 1, wherein: in the step 3), the duration and the distribution of the leakage noise signal along the way are as follows: