Water supply network leakage detection method and system combined with time-frequency domain analysis and medium

Abstract

The invention discloses a water supply pipe network leakage detection method and system combined with time-frequency domain analysis and a medium, and relates to the technical field of pipe network leakage detection, and the method comprises the steps: obtaining the normal water supply historical pressure data of each target pressure measurement point; performing normalization processing and framing processing to obtain a plurality of frame window water supply pressure data sequences and obtain window pressure time domain signals; learning window local pressure features; obtaining data key information under each frequency; reconstructing frequency domain signal features, reconstructing time domain signal features and obtaining a fitting window pressure water supply data sequence; obtaining a reconstruction loss set; judging a pipe network leakage event by using the reconstruction loss set; in the scheme of the invention, the influence of the time domain and frequency domain signals on the water supply pressure is considered, and the possible important characteristics of the water supply network in leakage are retained to the greatest extent, so that the problem of low leakage detection accuracy in the water supply network can be solved, and the method is suitable for different pipe network topological structures.

Description

Technical Field

[0001] The present invention relates to the technical field of pipe network leakage detection, and in particular to a water supply pipe network leakage detection method, system and medium combined with time-frequency domain analysis. Background Art

[0002] During the normal operation of the pipeline system, pipeline cracks caused by various reasons will lead to changes in pressure and flow. When the pipeline leaks, the fluid can no longer flow from one end of the crack to the other end at the agreed pressure. At the same time, the flow through the node will be reduced from the agreed water volume. The reduced flow is the amount of water that is leaked. Pipeline leakage usually refers to the gradual leakage of the pipeline due to some reason (such as pipeline aging, valve damage or other defects). The leakage volume gradually increases over time until it is finally discovered and measures are taken to stop it. Pipeline leakage causes significant damage to the operation of the city.

[0003] The normal operation of the pipe network system is an important infrastructure pillar for the rapid development of cities. The pipe network system is usually laid underground along the buildings. The pipe network system in old urban areas has a history of several decades. Its pipe network condition may change at any time due to various external forces, thus affecting the normal operation of life. Therefore, it is very important to effectively predict the operating pressure of the pipe network. If the pressure data changes of the pipe network can be detected in time and artificial intervention can be made in advance for emergencies, various economic losses caused by abnormal pipe network pressure can be avoided to a large extent. Effective detection of abnormal leakage in the water distribution network (WDN) can ensure the safety of municipal water supply and improve urban operation services.

[0004] In the prior art, the leakage detection method for water distribution network (WDN) mainly relies on human experience. On the one hand, the demand for professional talents is very high. On the other hand, even very experienced professionals cannot accurately detect various emergencies. The leakage detection method for water supply network mainly adopts historical data analogy. This method often uses historical data features too roughly and cannot detect subtle changes between data, resulting in low accuracy of water network leakage detection and prediction. In view of this, it is necessary to propose a water supply network leakage detection method, system and medium combined with time-frequency domain analysis to solve or at least partially solve the above technical problems. Summary of the invention

[0005] The main purpose of the present invention is to provide a water supply network leakage detection method, system and medium combined with time-frequency domain analysis, aiming to solve the technical problem of inaccurate detection results of existing water supply network leakage.

[0006] To achieve the above object, the present invention provides a water supply network leakage detection method combined with time-frequency domain analysis, comprising the steps of: S100, obtaining historical pressure data of normal water supply at each target pressure measuring point within a target time period in each operation scenario; S200, normalizing the historical pressure data of normal water supply at the measuring point, dividing the normalized historical pressure data of normal water supply at the measuring point into frames according to a preset frame length, acquiring multiple frame window water supply pressure data sequences and obtaining a window pressure time domain signal, wherein a frame window water supply pressure data sequence includes multiple node pressure data arranged in time sequence, one preset time step corresponds to one node pressure data, and adjacent frame window water supply pressure data sequences have overlapping lengths; S300, using the attention model to learn the window local pressure features based on the window pressure time domain signal; S400, transforming each window pressure time domain signal to obtain a window pressure frequency domain signal; The window pressure frequency domain signal is divided into high-frequency data, medium-frequency data and low-frequency data according to the frequency, wherein the medium-frequency data is the frequency data within the range of 40 to 90 Hz; The attention mechanism is used to learn the features of high-frequency data, medium-frequency data, and low-frequency data to obtain the key information of the data at each frequency; The key information of the data at each frequency is combined to obtain the complete reconstructed frequency domain signal characteristics; S500, using inverse Fourier transform to convert the reconstructed frequency domain signal features into reconstructed time domain signal features; S600, combining the local pressure characteristics of the window and reconstructing the time domain signal characteristics by weight allocation, and generating a new fitting window pressure water supply data sequence; S700, obtaining the window reconstruction loss of each fitting window pressure water supply data sequence and the corresponding frame window water supply pressure data sequence, and all window reconstruction losses constitute a reconstruction loss set; S800: Using the reconstructed loss set to identify leakage events in the pipe network.

[0007] Furthermore, “using the reconstructed loss set to identify pipeline network leakage events” specifically includes: Obtain the historical water supply pressure data to be tested of the pipe network to be tested; process the historical water supply pressure data to be tested to obtain the corresponding original water supply pressure data sequence to be tested and the reconstructed water supply pressure data sequence to be tested; if the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is greater than the loss peak value in the reconstruction loss concentration, it is determined that the pipe network to be tested has a leakage; if the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is greater than the loss peak value in the reconstruction loss concentration, it is determined that the pipe network to be tested has a leakage.

[0008] Furthermore, the low-frequency data is frequency data with a frequency between 0 and 40 Hz; and the high-frequency data is frequency data with a frequency between 90 and 128 Hz.

[0009] Furthermore, the value of the weight assignment is 0.5.

[0010] Furthermore, the target duration is one year, the preset frame length is one week, and the preset time step is thirty minutes.

[0011] Furthermore, the WNTR tool is used to generate pressure data under different operation scenarios. A pressure data node is generated every half hour, and a pressure data set with a duration of one year is generated in total. The pressure data set under each operation scenario is normalized to the interval [0,1], and the pressure data nodes in the pressure data set are divided into frames according to a fixed time length of one week. There are overlapping parts between frames. A frame window water supply pressure data sequence includes 336 node pressure data arranged in time sequence.

[0012] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned water supply network leakage detection method combined with time-frequency domain analysis are implemented.

[0013] The present invention also provides a water supply network leakage detection system combined with time-frequency domain analysis, including a processing device, and the processing device is used to implement the steps of the above-mentioned water supply network leakage detection method combined with time-frequency domain analysis.

[0014] Compared with the prior art, the water supply network leakage detection method combined with time-frequency domain analysis provided by the present invention has the following beneficial effects: The present invention provides a water supply network leakage detection method combined with time-frequency domain analysis. Based on the time local dependence of multiple frame window water supply pressure data sequences, the time domain characteristics and frequency domain characteristics of the normal water supply historical pressure data of the measuring point are respectively obtained, wherein the leakage detection performance of the water supply network is improved by respectively learning the high-frequency, medium-frequency and low-frequency signal characteristics, and the local pressure characteristics of the combined window are weighted and the time domain signal characteristics are reconstructed to generate a new fitting window pressure water supply data sequence. Considering the influence of time domain and frequency domain signals on the water supply pressure, the possible important characteristics of the water supply network in the leakage are retained to the greatest extent, thereby solving the problem of low leakage detection accuracy in the water supply network, effectively distinguishing leakage worlds, and adapting to different network topologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of the water supply network leakage detection method combined with time-frequency domain analysis of the present invention; Figure 2 It is a schematic diagram of the principle of the water supply network leakage detection method combined with time-frequency domain analysis of the present invention.

[0017] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0021] Please refer to Figure 1 and Figure 2The present invention provides a water supply network leakage detection method combined with time-frequency domain analysis, comprising the steps of: S100, obtaining the normal water supply historical pressure data of each target pressure measuring point within the target time length in each pipe operation scenario; S200, normalizing the normal water supply historical pressure data of the measuring point, dividing the normalized normal water supply historical pressure data of the measuring point into frames according to a preset frame length, obtaining multiple frame window water supply pressure data sequences and obtaining a window pressure time domain signal, wherein one frame window water supply pressure data sequence includes multiple node pressure data arranged in time sequence, one preset time step corresponds to one node pressure data, and adjacent frame window water supply pressure data sequences have an overlapping length; S300, using an attention model to learn the window local pressure features based on the window pressure time domain signal; S400, transforming each window pressure time domain signal to obtain a window pressure frequency domain signal; according to The window pressure frequency domain signal is divided into high-frequency data, medium-frequency data and low-frequency data according to the frequency, wherein the medium-frequency data is the frequency data in the range of 40 to 90 Hz; the attention mechanism is used to perform feature learning on the high-frequency data, medium-frequency data and low-frequency data to obtain the key information of the data at each frequency; the key information of the data at each frequency is combined to obtain the complete reconstructed frequency domain signal characteristics; S500, the reconstructed frequency domain signal characteristics are converted into reconstructed time domain signal characteristics by inverse Fourier transform; S600, the window local pressure characteristics and the reconstructed time domain signal characteristics are combined by weight distribution, and a new fitting window pressure water supply data sequence is generated; S700, the window reconstruction loss of each fitting window pressure water supply data sequence and the corresponding frame window water supply pressure data sequence is obtained, and all window reconstruction losses constitute a reconstruction loss set; S800, the reconstruction loss set is used to distinguish the leakage events of the pipeline network.

[0022] The present invention provides a water supply network leakage detection method combined with time-frequency domain analysis. Based on the time local dependence of multiple frame window water supply pressure data sequences, the time domain characteristics and frequency domain characteristics of the normal water supply historical pressure data of the measuring point are respectively obtained, wherein the leakage detection performance of the water supply network is improved by respectively learning the high-frequency, medium-frequency and low-frequency signal characteristics, and the local pressure characteristics of the combined window are weighted and the time domain signal characteristics are reconstructed to generate a new fitting window pressure water supply data sequence. Considering the influence of time domain and frequency domain signals on the water supply pressure, the possible important characteristics of the water supply network in the leakage are retained to the greatest extent, thereby solving the problem of low leakage detection accuracy in the water supply network, effectively distinguishing leakage worlds, and adapting to different network topologies.

[0023] It is understandable that the target duration can be a historical one-year duration, or a historical half-year duration, or other durations; the preset frame length is a one-week duration, a two-week duration, or other durations, and the preset frame length is less than the target duration; the preset time step duration is a half-hour duration, an hour duration, or other durations. The solution of the present invention is selected according to actual conditions. Optionally, the overlap length is a preset time step.

[0024] It can be understood that in the scheme of the present invention, the historical pressure data of normal water supply at each target pressure measuring point is normalized and then framed to form multiple frame window water supply pressure data sequences. The local pressure characteristics of the windows can be learned by performing model training on the multiple frame window water supply pressure data sequences and after the model converges. At the same time, each frame window water supply pressure data sequence forms a corresponding window pressure frequency domain signal after Fourier transform. The key information of the data at each frequency can be obtained by performing model training and learning on the multiple window pressure frequency domain signals and after the model converges.

[0025] It can be understood that the water supply pressure data refers to the data obtained by the pressure sensor installed on the detection node on the pipeline. The pressure sensor is used to monitor the pressure signal in the pipe and send it to the data platform through the Internet of Things terminal.

[0026] Furthermore, step S800 specifically includes: obtaining historical water supply pressure data to be tested at the pressure measuring point to be tested of the pipe network to be tested; processing the historical water supply pressure data to be tested to obtain the corresponding original water supply pressure data sequence to be tested and the reconstructed water supply pressure data sequence to be tested; if the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is greater than the loss peak value in the reconstruction loss set, it is judged that the pipe network to be tested has leakage; if the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is not greater than the loss peak value in the reconstruction loss set, it is judged that the pipe network to be tested has no leakage. It can be understood that in the scheme of the present invention, the historical water supply pressure data of the tested pipe network can be obtained with a time window of one week, or with a time window of two weeks or other time lengths, and processed with a preset frame length and a preset time step to obtain the corresponding reconstructed water supply pressure data sequence to be tested, and then the original water supply pressure data sequence to be tested and the reconstructed water supply pressure data sequence to be tested are obtained based on the attention model, frequency division processing, attention mechanism, etc.; finally, the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is obtained; if the reconstruction loss value is greater than the loss peak value in the reconstruction loss concentration, it is judged that the tested pipe network has a leak, otherwise it is judged that the tested pipe network has no leak.

[0027] Furthermore, low-frequency data is frequency data with a frequency between 0 and 40 Hz; high-frequency data is frequency data with a frequency between 90 and 128 Hz. In the scheme of the present invention, medium-frequency data includes frequency data of 40 Hz and 90 Hz, low-frequency data is frequency data less than medium-frequency data, and high-frequency data is frequency data greater than medium-frequency data. The historical pressure data of normal water supply at the measuring point of each pipe operation scenario is converted into a frequency domain signal through short-time Fourier transform. The data length of the frequency domain signal is consistent with the original pressure length. At the same time, the frequency domain signal is divided into three parts: high, medium and low. The attention mechanism is used to perform feature learning on these three frequency domain segments respectively, and the learned frequency domain signals are recombined into a complete frequency domain signal. Finally, the frequency domain signal is converted back to the time domain signal through inverse Fourier transform, which solves the problem that the signals in the high-frequency segment and the low-frequency segment are greatly affected by environmental interference factors. If the data in these frequency segments are directly input into the learning model, the learning performance of the model will be reduced, thereby affecting the accuracy of pressure data prediction.

[0028] Furthermore, a learnable weight parameter is used to weight the local features and frequency domain features respectively, and input into a linear activation function to generate a new sequence. Optionally, the local feature is multiplied by a, and the time domain converted from the frequency domain is multiplied by (1-a) to generate a new sequence. The new sequence is compared with the original sequence to see the similarity. Since the training is performed with non-leakage data, the similarity will be very large when no leakage occurs (the loss value is very small). Preferably, the weight value of the weight allocation is 0.5.

[0029] In a specific embodiment of the present invention, the target duration is one year, the preset frame length is one week, and the duration of a preset time step is thirty minutes; a frame window water supply pressure data sequence includes 336 node pressure data arranged in time sequence.

[0030] Furthermore, the WNTR tool was used to generate pressure data under different pipe network operation scenarios. A pressure data node was generated every half an hour, and a total of one year of pressure data was generated. The pressure data under each scenario was normalized to the [0,1] interval, and the pressure data was divided into frames according to a fixed frame length. There was an overlap between frames. The fixed frame length was set to one week. A frame window water supply pressure data sequence includes 336 (24 2 7=336) node pressure data arranged in time sequence, with the overlap length being a preset time step.

[0031] Please refer again Figure 2 , Figure 2The single curve in the middle represents the data of a pressure station on the pipeline (multiple pressure stations will be installed on the pipeline), that is, a multivariate time series; the concept of the scheme of the present invention mainly includes: the data of the pressure station after normalization processing; the frame window water supply pressure data sequence (window pressure time domain signal) uses the attention mechanism to learn the temporal changes of pressure data; obtain the local pressure characteristics of the window; after obtaining the window pressure frequency domain signal, it is divided into high-frequency data, medium-frequency data and low-frequency data for feature learning to obtain the key information of the data at each frequency; the key information of the data at each frequency is combined to obtain the complete reconstructed frequency domain signal characteristics; the reconstructed time domain signal characteristics are obtained; and finally the local pressure characteristics of the window and the reconstructed time domain signal characteristics are combined through weight distribution β.

[0032] The present invention provides a method for detecting water supply network leakage by combining time-frequency domain analysis, and the implementation method is as follows: The WNTR tool is used to collect the historical pressure data of normal water supply at the measuring points under various pipe operation scenarios (urban pipeline operation scenarios, such as A city pipe operation scenario, B city pipe operation scenario, etc.), and the historical pressure data of normal water supply at the measuring points are preprocessed and normalized to the [0,1] interval; specifically, the wntr tool is used to generate the historical pressure data of normal water supply at the measuring points under different pipe operation scenarios, with one pressure data node every half an hour, and a total of one year of pressure data nodes. The pipeline network used in the historical data set is provided in the form of an EPANET INP file. The model parameters (pipeline length, diameter, roughness) of each pipeline network are different to increase the uncertainty of the model data. The present invention uses a pressure-driven simulation mode for each leakage scenario, in which the amount of water delivered depends on the pressure. At the same time, the basic periodic component of the demand is approximated based on the Fourier series (FS) of the real historical water demand of the water supply enterprise.

[0033] Extract the characteristics of the preprocessed pressure signal; normalize the pressure data in each scene to the interval [0,1], divide the frames into frames according to the fixed frame length, and there is an overlap between frames. The fixed frame length is set to one week, that is, 24 2 7 = 336 time points, with an overlap length of one time step; when normalizing, L is the frame window size, is the pressure data under each frame window (the window data sequence at the tth moment of the nth pressure station), and the expression of data normalization is min represents the minimum pressure value in the sequence, and max represents the maximum pressure value in the sequence. It is the minimum constant. When the maximum value is the same as the minimum value, division by zero can be avoided.

[0034] The attention model is used to reconstruct a new fitting window pressure water supply data sequence; specifically, the reconstruction expression using the attention model is z'= T(z), where z is the pressure sequence of the original time domain features input into the attention model, T is the attention model of the encoder-decoder structure, and z' is the pressure sequence of the new time domain features generated by the attention reconstruction. In the solution of the present invention, the pressure data is first connected with the data encoding, instead of adding and summing them as in most methods. pe is the position code corresponding to the pressure value of x1(t), specifically, pe1(t) is the position code corresponding to x1(t); The attention mechanism is mainly used to calculate the degree of dependence of a certain value on other pressure values, and the final output carries the attention weight.

[0035] Extract the frequency domain features of the preprocessed pressure signal; process the historical pressure data of normal water supply at the measuring point of each operation scenario to obtain the window pressure time domain signal, and then convert it into a frequency domain signal through short-time Fourier transform. The data length of the frequency domain signal is consistent with the original pressure length. At the same time, the frequency domain signal is divided into three parts: high, medium and low. The attention mechanism is used to perform feature learning on these three frequency domain segments respectively, and finally the learned frequency domain signals are recombined into a complete frequency domain signal. Finally, the frequency domain signal is converted back to the time domain signal through inverse Fourier transform. The present invention believes that the highest frequency signal and the low frequency band signal are generally more affected by environmental interference factors. If the data of these frequency bands are directly input into the learning model, the learning performance of the model will be reduced, thereby affecting the accuracy of pressure data prediction.

[0036] Use learnable weight allocation to combine local features (window local pressure features) and periodic features (reconstructed time domain signal features) and generate a new fitting window pressure water supply data sequence; optionally, use a learnable weight parameter to weight the local features and frequency domain features respectively, and input them into a linear activation function to generate a new fitting window pressure water supply data sequence. Specifically, when combining, the weight parameter is 0.5.

[0037] Calculate the reconstruction loss between the reconstructed pressure and the original pressure; the expression for calculating the reconstruction loss is: loss = ||z' - z||2, which is expressed as the 2-norm distance between the reconstructed pressure and the pressure data.

[0038] Reconstruction loss is used to identify pipeline leakage events. In the training phase, the model is trained using data without leakage to obtain a series of loss value sets, and the peak value of the loss value set is calculated. In the testing phase, the test loss value is compared with the peak value. If it is smaller than the peak value, the output indicates no leakage, otherwise the output indicates leakage.

[0039] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the water supply network leakage detection method combined with time-frequency domain analysis are performed.

[0040] The present invention also provides a water supply network leakage detection system combined with time-frequency domain analysis, including a processing device, wherein the processing device is used to implement the steps of the above-mentioned water supply network leakage detection method combined with time-frequency domain analysis.

[0041] The water supply network leakage detection method, system and computer storage medium combined with time-frequency domain analysis of the present invention have the following effects: since only the pressure data of the pressure station is considered for abnormal judgment, and no additional information about the network is required, different network structures can be adapted; the network operation pressure in each scenario is inconsistent, so the pressure data in each scenario is normalized separately, which can effectively capture the changes in pressure data; while considering the influence of time domain characteristics and frequency domain characteristics on water supply pressure, the possible characteristics of the network in leakage are retained to the greatest extent; the water supply network can be leaked, and the leakage scenarios can be effectively distinguished through the learned time domain and frequency domain pressure data characteristics, thereby improving the accuracy of water supply network leakage detection.

[0042] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0043] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A water supply network leakage detection method combined with time-frequency domain analysis, characterized in that: Includes steps: Obtain the historical pressure data of normal water supply at each target pressure measuring point within the target time period in each operation scenario; Normalizing the historical pressure data of normal water supply at the measuring point, dividing the normalized historical pressure data of normal water supply at the measuring point into frames according to a preset frame length, acquiring multiple frame window water supply pressure data sequences and obtaining a window pressure time domain signal, wherein one frame window water supply pressure data sequence includes multiple node pressure data arranged in time sequence, one preset time step corresponds to one node pressure data, and adjacent frame window water supply pressure data sequences have overlapping lengths; Based on the window pressure time domain signal, the attentiion model is used to learn the window local pressure characteristics; Transform each window pressure time domain signal to obtain a window pressure frequency domain signal; divide the window pressure frequency domain signal into high frequency data, medium frequency data and low frequency data according to the frequency, wherein the medium frequency data is the frequency data in the range of 40 to 90 Hz; use the attention mechanism to perform feature learning on the high frequency data, medium frequency data and low frequency data respectively to obtain the key information of the data at each frequency; combine the key information of the data at each frequency to obtain the complete reconstructed frequency domain signal feature; Using inverse Fourier transform to transform the reconstructed frequency domain signal features into reconstructed time domain signal features; The weight distribution is used to combine the window local pressure characteristics and reconstruct the time domain signal characteristics, and a new fitting window pressure water supply data sequence is generated; Obtaining the window reconstruction loss of each of the fitting window pressure water supply data sequences and the corresponding frame window water supply pressure data sequence, all window reconstruction losses constitute a reconstruction loss set; The reconstructed loss set is used to identify leakage events in the pipeline network.

2. The water supply network leakage detection method combined with time-frequency domain analysis according to claim 1 is characterized in that: “Using the reconstructed loss set to identify leakage events in the pipeline network” specifically includes: Obtain the historical water supply pressure data of the pressure measuring point to be tested in the pipe network to be tested; Processing the historical water supply pressure data to be tested to obtain a corresponding original water supply pressure data sequence to be tested and a reconstructed water supply pressure data sequence to be tested; If the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is greater than the loss peak value in the reconstruction loss set, it is judged that the pipeline network to be tested has leakage; if the reconstruction loss value of the reconstructed water supply pressure data sequence to be tested and the original water supply pressure data sequence to be tested is not greater than the loss peak value in the reconstruction loss set, it is judged that the pipeline network to be tested has no leakage.

3. The water supply network leakage detection method combined with time-frequency domain analysis according to claim 1 is characterized in that: The low-frequency data is the frequency data between 0 and 40 Hz; the high-frequency data is the frequency data between 90 and 128 Hz.

4. The water supply network leakage detection method combined with time-frequency domain analysis according to claim 1 is characterized in that: The value of the weight allocation is 0.

5.

5. The water supply network leakage detection method combined with time-frequency domain analysis according to any one of claims 1 to 4, characterized in that: The target duration is one year, the preset frame length is one week, and the preset time step is thirty minutes.

6. The water supply network leakage detection method combined with time-frequency domain analysis according to claim 5 is characterized in that: Use the WNTR tool to generate pressure data in different pipeline operation scenarios, generate a pressure data node every half an hour, and generate a pressure data set for a year in total; The pressure data set in each operation scenario is normalized to the interval [0,1], and the pressure data nodes in the pressure data set are divided into frames according to a fixed time length of one week, with overlapping parts between frames. A frame window water supply pressure data sequence includes 336 node pressure data arranged in time sequence.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the water supply network leakage detection method combined with time-frequency domain analysis as described in any one of claims 1 to 6 are implemented.

8. A water supply network leakage detection system combined with time-frequency domain analysis, characterized in that: It comprises a processing device, which is used to implement the steps of the water supply network leakage detection method combined with time-frequency domain analysis as described in any one of claims 1 to 6.

Images