Water supply network leakage identification and positioning method and device based on earth surface array sampling

Abstract

The invention discloses a water supply network leakage identification and positioning method based on earth surface array sampling, which comprises the following steps: acquiring working condition parameters of an area where a water supply network is located, and arranging sensor arrays on the earth surface at intervals along the axial direction of a pipeline of the water supply network based on the working condition parameters; collecting a vibration signal in the pipeline through the sensor array, and performing spatial domain Fourier transform on the collected vibration signal at a preset time step length to construct a corresponding wave number-frequency spectrum; and performing data analysis based on the waveform image of the wave number-frequency spectrum to obtain whether the pipeline leaks or not and the relative position of a leakage point and the sensing array. The invention further provides a water supply network leakage identifying and positioning device. When the method provided by the invention is applied to the ground surface, the technical bottleneck that the leakage of the buried pipeline cannot be detected and cannot be accurately determined is broken through, and an efficient solution which can be popularized in an engineering manner is provided for reducing the leakage rate of a pipe network.

Description

Technical Field

[0001] This invention belongs to the technical field of municipal water supply systems, and particularly relates to a method and device for identifying and locating leaks in water supply networks based on surface array sampling. Background Technology

[0002] During water supply, approximately 20-30% of treated water is lost due to pipeline leaks. This not only results in a huge waste of water resources but also triggers secondary disasters such as drinking water pollution and ground subsidence, threatening public safety and social stability. Among numerous leak detection methods, acoustic detection has become the mainstream method worldwide due to its non-invasiveness, high sensitivity, and low cost.

[0003] Acoustic-based leak detection methods aim to identify and locate leaks. Regarding leak identification: Traditional acoustic leak identification involves analyzing leak signals collected by single-point sensors in the time, frequency, or time-frequency domains to extract features (such as entropy, standard deviation, and root mean square) to determine if a leak has occurred. However, this approach faces significant challenges: the intensity of the leak vibration source is close to the background noise, greatly reducing the accuracy of leak identification. Regarding leak location: Correlation analysis methods are widely used for locating leaks in water supply pipelines. This method calculates the time delay between the leak signal and sensors on both sides of the leak point to pinpoint the location. However, correlation analysis requires wave velocity estimation for leak location. In buried pipeline systems, the propagation speed of vibration is related to the pipeline and soil properties, and the wave velocity of leak noise in a specific system is not singular, exhibiting multimodal wave aliasing, making wave velocity difficult to calculate. Current methods rely on empirical wave velocity measurements of fluid wave or shell wave modes, resulting in insufficient location accuracy. In addition, for non-metallic pipes, due to the damping of the pipe wall, the attenuation rate experienced by the wave as it propagates along the pipe is relatively high. The propagation distance of leakage noise on the pipe is generally less than 50m, and at this time it is impossible to use fire hydrants to perform correlation calculations to locate the source.

[0004] Patent CN120488150A discloses a method and system for locating leaks in water network pipelines. The method includes: collecting pressure and flow data of the water network pipelines; analyzing the pressure and flow data to determine whether a leak has occurred in the pipeline being monitored; discretizing the pipeline system in a pre-constructed mathematical model of the leaking pipeline to determine the calculation area and selecting pressure detection points in the water network pipeline; calculating the pressure values ​​at each pressure detection point under different leakage conditions; performing sensitivity analysis on each node of the main and branch pipelines of the water network using the correlation function method and comparing the correlation coefficients of each node; and determining the location of the pipeline leak point based on the correlation coefficients.

[0005] Patent document CN117628421A discloses a leak location method in a water supply network environment, comprising: acquiring detection signals Sa and Sb respectively, wherein the detection signals Sa and Sb are respectively placed on both sides of the leak point; performing ensemble empirical mode decomposition on the detection signals Sa and Sb respectively to obtain several component signals A and several component signals B; selecting component signals containing the leak source from the time-frequency domain information of the several component signals A and several component signals B respectively to obtain leak component signals A and leak component signals B respectively; reconstructing the signals based on the leak component signals A and leak component signals B respectively to obtain leak source signals A and leak source signals B respectively; determining the time delay information based on the time-domain waveforms of the leak source signals A and leak source signals B respectively; and determining the leak location on the water supply pipe based on the time delay information. Summary of the Invention

[0006] The purpose of this invention is to provide a method and device for identifying and locating leaks in water supply networks based on surface array sampling. When applied on the surface, this method breaks through the technical bottleneck of being unable to detect or accurately locate leaks in buried pipelines, and provides an efficient solution that can be engineered and promoted to reduce the leakage rate of pipeline networks.

[0007] To achieve the first objective of this invention, the following technical solution is provided: a method for identifying and locating leaks in water supply networks based on surface array sampling, comprising: The operating parameters of the water supply network area are obtained, and sensor arrays are arranged at intervals on the ground along the pipeline axis of the water supply network based on the operating parameters. Vibration signals in the pipeline are collected by the sensor array, and the collected vibration signals are subjected to a two-dimensional Fourier transform in the time-space domain at a preset time step to construct the corresponding wavenumber-frequency spectrum. The wavenumber-frequency spectrum image is labeled according to whether it is a leakage signal and the relative position of the leakage point to the sensor array, and the vibration signal, signal wavenumber-frequency spectrum and labels are combined to form a dataset; Construct a leakage detection network, which includes a feature extraction module and an analysis module; The feature extraction module is used to extract wavenumber-frequency spectrum features from the input vibration signal; The analysis module performs data analysis based on wavenumber-frequency spectrum characteristics to obtain identification results. The leakage detection network was trained using the dataset to obtain a water supply pipeline leakage detection model for pipeline leakage detection and location. The vibration signal to be identified is input into the water supply pipeline leakage identification model to output whether the pipeline is leaking and the relative position of the leakage point to the sensor array.

[0008] This invention utilizes a sensor array to capture vibration information in both spatial and temporal dimensions, thereby inferring whether a leak has occurred and the location of the leak.

[0009] Specifically, the leak location of the pipeline is determined by repeatedly moving the position of the sensor array and combining the relative position of the leak point with that of the sensor array.

[0010] Specifically, the dataset covers all water supply network leakage scenarios under different operating conditions, including different pipe materials, burial depth, soil, surface conditions, pipe diameter, and leak size.

[0011] Specifically, the operating parameters include pipeline fluid wave velocity, pipeline shell wave velocity, soil shear wave velocity, and soil compression wave velocity.

[0012] Specifically, the expression for the spacing distance of the sensor array is as follows: Where L represents the distance between adjacent sensors, vF Indicates the wave velocity of the fluid in the pipe. vW Indicates the wave velocity of the pipe shell. vS Indicates soil shear wave velocity, vP Indicates soil compression wave velocity, f max This indicates the maximum observation frequency.

[0013] Specifically, the mounting aperture expression for each sensor in the sensor array is as follows: in, f min This indicates the minimum observation frequency. When the axial distance between the sensor array and the leakage point is >2m, c 1 and c 2 for vF and vW The axial distance between the sensor array and the leakage point is ≤2m. c 1 and c 2 for vS , vP .

[0014] Specifically, the data analysis includes leak identification and leak location.

[0015] Specifically, the leakage identification network is constructed based on the pre-trained neural network model SqueezeNet framework and the convolutional neural network model framework.

[0016] Specifically, the expression for the two-dimensional Fourier transform in the time-space domain is as follows: ; in, yes Fourier transforms in the x and t directions, with x pointing towards the pipe axis, The axial wave number of the pipe is represented by i, where i represents the imaginary unit and f represents the frequency.

[0017] Specifically, the sensor array includes wireless sensors or wired sensors.

[0018] Specifically, when the sensor arrays are all positioned on one side of the leak point, the presence of the aforementioned four modes of acoustic wave wave numbers in the wavenumber-frequency spectrum can determine whether there is a leak point near the sensor arrays. When a leak point exists and the sensor arrays are all located on one side of the leak point, the intensity of positive and negative wavenumbers in the wavenumber-frequency spectrum is severely imbalanced. Based on the comparison of positive and negative wavenumber intensities, a simple judgment can be made regarding the location of the leak point. Based on this preliminary judgment, the sensor array positions are gradually moved until the intensity of positive and negative wavenumbers in the wavenumber-frequency spectrum reaches a balanced level. At this point, the sensor arrays will pass directly above the leak point and collect leakage signals on both sides of the leak point. It can then be determined that the sensor arrays are essentially located directly above the leak point.

[0019] To achieve the second objective of the present invention, the following solution is provided: a water supply network leakage identification and location device, used to implement the steps of the above-mentioned water supply network leakage identification and location method based on surface array sampling.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: The sensor array captures vibration information in the spatial and temporal dimensions, and with the help of a specific judgment strategy, it can infer whether there is a leak and the location of the leak. This overcomes the dependence of existing leak detection products on professional leak listeners and realizes intelligent leak detection. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the water supply network leakage identification and location method based on surface array sampling provided in this embodiment; Figure 2 The wavenumber frequency spectrum is provided for different sampling intervals and apertures in this embodiment; Figure 3 This is a schematic diagram of the water supply network leak identification and location device based on surface array sampling provided in this embodiment. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0023] like Figure 1 As shown, the water supply network leakage identification and location method based on surface array sampling provided in this embodiment includes: Obtaining the training dataset for the machine learning model: Under different sampling strategies, leakage system characteristics, and relative orientations of the leak and the sensor array, the two-dimensional Fourier transform (kf) spectrum characteristics of the sensor array's sampling signals will differ. The kf spectrum contains information on whether a leak has occurred, as well as the orientation information of the leak sound source relative to the sensor array. Data was sampled from experimental and actual urban pipe networks, with no fewer than 1000 sampling conditions. Variables included sampling aperture, sampling interval, soil cover height, soil properties, pipe material, pipe diameter, leak size, leak orientation, and internal pipe pressure. The leakage conditions involved in the sampling were verified through on-site excavation to ensure the quality of the training data. The dataset includes both non-leaking and leaking conditions in a ratio of approximately 4:6, forming the dataset required to construct the model.

[0024] The signal data collected by the surface array sensor group is signal amplitude information for a specific duration, which may contain various environmental interference noises. The method of this invention relies on Kf spectrum features when extracting features, which has strong anti-noise interference capabilities. The established Kf spectrum dataset provides high-quality data support for the identification of leakage signals and azimuth in water supply networks.

[0025] The system acquires operating parameters for the water supply network area, including pipe fluid wave velocity, pipe shell wave velocity, soil shear wave velocity, and soil compression wave velocity. Based on these operating parameters, a sensor array is then spaced along the pipe axis of the water supply network on the ground surface. Vibration signals in the pipeline are collected by the sensor array, and the collected vibration signals are subjected to spatial and temporal Fourier transforms at a preset time step to construct the corresponding wavenumber-frequency spectrum. More specifically, step 1 involves collecting information on the soil type (sand or clay, etc.) and pipe type (PE pipe, steel pipe, cast iron pipe, etc.) for the tested conditions. Based on the soil and pipe type, the empirical wave velocities of the pipe fluid and shell waves in the low-frequency band (usually not exceeding 2000Hz) and the free-field shear and compression wave velocities of the soil in the 20-2000Hz frequency band are output. These acoustic waves represent four high-energy modes that can be monitored on the surface under leak conditions.

[0026] Step 2: According to the Nyquist sampling theorem, determine that the sampling frequency is greater than 4000Hz. Adjust the sampling interval and aperture according to the acoustic mode to be distinguished.

[0027] Step 3: Perform Discrete Fourier Transform (DFT) processing on the vibration signals acquired by the sensor array. The result is: In the above formula express The (th) of the matrix p,q ) elements, Represents the imaginary unit. P and Q These represent the number of discrete Fourier transforms in space and time, respectively, in this embodiment. .

[0028] For the same discrete data, wavenumber frequency spectra with different sampling intervals and apertures are as follows: Figure 2 .

[0029] The machine learning model trained on waveform images based on wavenumber-frequency spectrum is analyzed to obtain information on whether there is a leakage event and the location of the leakage.

[0030] Training a water supply pipeline leakage identification and guidance model based on surface array sampling: A dataset composed of various leakage and non-leakage signals collected from experimental devices and actual pipe networks was used to train the machine learning model and verify its identification performance. A pre-trained neural network model, SqueezeNet, was used to classify the k-f images of the sampling signals from different sensor arrays. Classification information included leakage, non-leakage, and whether the leakage point was in the direction of the head sensor or the tail sensor of the sampling array. The dataset was defined as four classes: non-leakage signals (labeled "0"), leakage signals (labeled "1"), leakage signals with the leakage point in the direction of the head sensor of the sampling array (labeled "2"), and leakage signals with the leakage point in the direction of the tail sensor of the sampling array (labeled "3"). The dataset, composed of both leakage and non-leakage sampling signals, was used to train the machine learning model and verify its identification performance. The training process was supervised learning, meaning that the signal labels were also used as input information to the model.

[0031] The specific training process is as follows: (1) The SqueezeNet model is trained using the signal kf spectrum images in the training set; (2) Use the trained models to identify the validation set and modify the hyperparameters of the corresponding models according to the application effect to obtain better performance on the validation set. (3) Input the test set into the trained model to determine whether the signal is lost and the direction of leakage, and obtain the recognition result.

[0032] This embodiment also provides a water supply network leak identification and location device, used to perform the steps of the water supply network leak identification and location method based on surface array sampling provided in the above embodiments, such as... Figure 3 The diagram shown is a schematic of the device based on a wireless sensor array provided in this embodiment.

[0033] To systematically evaluate the engineering applicability of the proposed positioning method under real-world site conditions, testing and verification were conducted at a leakage training base in a certain city. The test pipeline was a 100mm PE pipe, buried at a depth of 0.5 meters. A total of 52 sensors were used during the test, with an array aperture of 3.57 meters.

[0034] The initial sampling results were input into the water supply pipeline leakage identification model and guidance model based on surface array sampling. The result showed a leakage signal with the leakage point located at the sensor direction at the tail of the sampling array. Therefore, the sensor array was shifted 10 meters towards the tail.

[0035] The second sampling result was input into the water supply pipeline leakage identification model and guidance model based on surface array sampling. The result showed a leakage signal and the leakage point was in the direction of the head sensor of the sampling array. Therefore, the sensor array was shifted 5 meters towards the head.

[0036] The location of the leak sound source can be approximated by repeatedly translating the sensor array. The final results show that the deviation between the leak point predicted by this method and the actual location is 1.2m, demonstrating the feasibility of the water supply network leak identification and location method and device based on surface array sampling in this invention for engineering applications.

Claims

1. A method for identifying and locating leaks in water supply networks based on surface array sampling, characterized in that, include: The operating parameters of the water supply network area are obtained, and sensor arrays are arranged at intervals on the ground along the pipeline axis of the water supply network based on the operating parameters. Vibration signals in the pipeline are collected by the sensor array, and the collected vibration signals are subjected to a two-dimensional Fourier transform in the time-space domain at a preset time step to construct the corresponding wavenumber-frequency spectrum. The wavenumber-frequency spectrum image is labeled according to whether it is a leakage signal and the relative position of the leakage signal to the sensor array, and the vibration signal, signal wavenumber-frequency spectrum and labels are combined to form a dataset. Construct a leakage detection network, which includes a feature extraction module and an analysis module; The feature extraction module is used to extract wavenumber-frequency spectrum features from the input vibration signal; The analysis module performs data analysis based on wavenumber-frequency spectrum characteristics to obtain identification results. The leakage detection network was trained using the dataset to obtain a water supply pipeline leakage detection model for pipeline leakage detection and location. The vibration signal to be identified is input into the water supply pipeline leakage identification model to output whether the pipeline is leaking and the relative position of the leakage point to the sensor array.

2. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The location of the leak in the pipeline is determined by repeatedly moving the sensor array and combining the relative position of the leak point with that of the sensor array.

3. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The expression for the spacing distance of the sensor array is as follows: ； Where L represents the distance between adjacent sensors, vF Indicates the wave velocity of the fluid in the pipe. vW Indicates the wave velocity of the pipe shell. vS Indicates soil shear wave velocity, vP Indicates soil compression wave velocity, f max This indicates the maximum observation frequency.

4. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The expression for the mounting aperture of each sensor in the sensor array is as follows: ； in, f min This indicates the minimum observation frequency; when the axial distance between the sensor array and the leakage point is >2m, c 1 and c 2 for vF and vW The axial distance between the sensor array and the leakage point is ≤2m. c 1 and c 2 for vS , vP .

5. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The expression for the two-dimensional Fourier transform in the time-space domain is as follows: ； in, yes Fourier transforms in the x and t directions, with x pointing towards the pipe axis, The axial wave number of the pipe is represented by i, where i represents the imaginary unit and f represents the frequency.

6. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The data analysis includes leak identification and leak location.

7. The method for identifying and locating leaks in water supply networks based on surface array sampling according to claim 1, characterized in that, The leakage identification network is constructed based on the pre-trained neural network model SqueezeNet framework and the convolutional neural network model framework.

8. A device for identifying and locating leaks in a water supply network, characterized in that, The steps are for performing the method for identifying and locating leaks in water supply networks based on surface array sampling as described in any one of claims 1 to 7.

Images