A method, device, equipment and medium for monitoring leakage in urban underground pipe networks.

Abstract

This application discloses a method, device, equipment, and medium for monitoring leakage in urban underground pipe networks, relating to the field of intelligent maintenance technology for urban infrastructure. The method includes: acquiring data collected by multiple target devices installed at various locations within a target area of ​​the urban underground pipe network over a preset time period to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors; the distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing; preprocessing the initial pipe network status data, and inputting the processed status data into a pipe network leakage detection model based on a self-attention mechanism to perform leakage detection on each urban underground pipe network and obtain leakage detection results. This application can simultaneously detect leakage in multiple urban underground pipe networks, improving the accuracy and efficiency of detection while saving labor and time costs.

Description

Technical Field

[0001] This application relates to the field of urban infrastructure operation and maintenance technology, and in particular to a method, device, equipment and medium for monitoring leakage in urban underground pipe networks. Background Technology

[0002] With the acceleration of urbanization and the aging of pipeline networks, frequent pipeline leaks, ruptures, blockages, and corrosion not only cause resource waste and environmental pollution, but also pose a serious threat to public safety, property, and urban operations. Therefore, achieving 24 / 7, high-precision, and intelligent leakage monitoring and fault early warning for urban underground pipeline networks (covering water supply, drainage, heating, and gas) has become a core technological requirement for smart city construction and urban lifeline safety projects.

[0003] However, underground pipe networks are complex and highly concealed, and the operating conditions of different types of pipe networks vary significantly. Traditional leakage monitoring methods based on single-type sensors suffer from information silos, making it difficult to achieve collaborative utilization of multi-source data, resulting in low detection accuracy and limited positioning accuracy of a single sensor.

[0004] Furthermore, current leakage monitoring methods based on multi-sensor fusion still have significant shortcomings in terms of positioning accuracy, real-time performance, inherent safety, and emergency decision support. For example, the monitoring methods are singular, relying excessively on manual inspections and discrete sensors, making it difficult to achieve real-time and holistic perception of the pipeline network's operating status, resulting in a serious lag in the detection of accidents such as leaks and spills; different types of pipeline networks are independent of each other, and the multi-source monitoring data across systems is isolated and cannot be integrated and shared, resulting in low accuracy of detection results and the existence of missed and false alarms; drainage detection relies on manual interpretation, which is inefficient and highly subjective; heating pipeline networks lack dynamic stress life assessment; and in the face of emergencies such as pipe bursts and gas leaks, the lack of real-time data-driven intelligent decision-making often leads to passive and delayed emergency responses. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a method, device, equipment, and medium for monitoring leakage in urban underground pipe networks, which can improve the accuracy of leakage detection in urban underground pipe networks, save labor and time costs, and simultaneously detect leakage in multiple urban underground pipe networks within a target area, thereby improving the efficiency of leakage detection. The specific solution is as follows: Firstly, this application discloses a method for monitoring leakage in urban underground pipe networks, including: Data collected by multiple target devices installed at various urban underground pipe networks in the target area within a preset time period is obtained to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors; the distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. The initial pipeline network status data is preprocessed to obtain processed status data; The processed state data is input into the trained pipeline leakage detection model to perform leakage detection on each of the city's underground pipelines based on the processed state data, and the leakage detection results are obtained. The pipeline leakage detection model is a model obtained by training a neural network model based on a self-attention mechanism using historical pipeline status data.

[0006] Optionally, the preprocessing of the initial pipeline network status data to obtain processed status data includes: The initial pipeline status data collected by each target device is sampled according to a preset sampling interval to obtain the sampled pipeline status data. Identify and remove outliers from the sampled pipeline network status data to obtain the first removed status data; The number of all outliers in each of the sampled pipeline network status data is counted to obtain a first outlier count, and the ratio of the first outlier count to the total number of status values ​​is calculated to obtain a first ratio; the total number of status values ​​is the number of all status values ​​in the sampled pipeline network status data. Determine whether the first ratio is less than the first threshold; If the first ratio is less than the first threshold and the missing points at different outliers are isolated from each other, then the missing points in the first removed state data are filled by interpolation to obtain filled state data. Identify and remove outliers from the filled state data to obtain the second removed state data; The second removed state data is preprocessed to obtain processed state data.

[0007] Optionally, the step of identifying and removing outliers from the sampled pipeline network status data to obtain the first removed status data includes: Identify state values ​​in the sampled pipeline status data that exceed a preset state threshold, and treat state values ​​that exceed the preset state threshold as outliers; Calculate the difference between any two adjacent state values ​​in the sampled pipeline network state data in sequence, and determine whether the difference exceeds the second threshold. If the difference exceeds the second threshold, the corresponding state value will be regarded as an outlier. The outliers in the sampled pipeline status data are removed to obtain the first set of removed status data.

[0008] Optionally, when the urban underground pipe network includes water supply network, drainage network, heating network and gas network simultaneously, the preprocessing operation on the second rejected state data to obtain processed state data includes: The second rejected state data corresponding to the pressure sensor is denoised by low-pass filtering to obtain the first denoised state data, and it is determined whether there is baseline drift in the pressure signal in the first denoised state data. If the first denoised state data has a baseline drift, then remove the trend term from the first denoised state data to obtain the first processed state data; The second denoised state data corresponding to the flow sensor is denoised by median filtering to obtain the second denoised state data, and the node flow balance is verified on the second denoised state data to obtain the verification result. If the verification results show that the inflow and outflow at each node of the urban underground pipeline network do not satisfy the conservation relationship, then the second denoised state data is corrected using the least squares method to obtain the second processed state data. Temperature compensation is applied to the second rejected state data corresponding to the water quality sensor to obtain compensated state data. Piecewise linear interpolation is then used to interpolate the compensated state data to obtain the third processed state data. The acoustic sensing data in the second denoised state data corresponding to the distributed optical fiber sensor is denoised to obtain the third denoised state data. The leakage acoustic wave in the third denoised state data is separated from other interference sources by independent component analysis to obtain the fourth processed state data. The temperature sensing data in the second denoised state data corresponding to the distributed optical fiber sensor is denoised to obtain the fourth denoised state data, and the fourth denoised state data is trend corrected to obtain the fifth processed state data.

[0009] Optionally, the step of performing leakage detection on each of the city's underground pipe networks based on the processed state data to obtain leakage detection results includes: The location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply network is determined to obtain the target optical fiber location. Based on the first processed state data and the second processed state data corresponding to the target optical fiber location, the pressure drop rate and flow imbalance of the corresponding pipeline segment are calculated. If the pressure drop rate exceeds a preset rate threshold and the flow imbalance is greater than a third threshold, then the water supply network is determined to have leakage abnormalities. Based on the fifth processed state data corresponding to the drainage pipe network, the temperature change of each pipe network segment of the drainage pipe network is calculated to obtain the first temperature change. If the first temperature change exceeds the fourth threshold, the liquid level change trend and liquid level change amount of the corresponding pipeline segment are identified based on the processed status data corresponding to the liquid level sensor upstream of the corresponding pipeline segment. If the liquid level change trend shows a downward trend and the liquid level change exceeds the fifth threshold, then it is determined that there is an abnormal leakage in the drainage network. Based on the fifth processed state data corresponding to the heating network, the temperature change of each section of the heating network is calculated to obtain the second temperature change. If the second temperature change exceeds the fifth threshold, it is determined whether there is a target high-frequency jet sound wave in the fourth processed state data corresponding to the corresponding pipeline segment; if so, it is determined that there is a leakage abnormality in the thermal pipeline. Based on the fourth processed state data corresponding to the gas pipeline network, it is determined whether there are ultrasonic waves of a specific frequency band in each pipeline segment of the gas pipeline network; If a section of the gas pipeline network contains ultrasonic waves of a specific frequency, then the gas pipeline network is determined to have a leakage anomaly.

[0010] Optionally, determining the location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply network to obtain the target optical fiber location includes: The acoustic wave features are extracted from the fourth processed state data corresponding to the water supply network to obtain broadband acoustic wave features; the broadband acoustic wave features include any one or more of kurtosis features, short-time energy features, and spectral centroid features. Detect whether the broadband acoustic wave characteristics meet preset requirements; the preset requirements are that the kurtosis characteristic is less than a preset kurtosis threshold, and / or the short-time energy characteristic is less than a preset energy threshold, and / or the spectral centroid characteristic is located within a preset frequency domain distribution centroid range; If the broadband acoustic wave characteristics do not meet the preset requirements, the position of the distributed optical fiber sensor corresponding to the fourth processed state data is determined to obtain the target optical fiber position.

[0011] Optionally, determining whether a target high-frequency jet sound wave exists in the fourth processed state data corresponding to the corresponding pipeline segment, and if so, determining that the thermal pipeline has a leakage anomaly, includes: The fourth processed state data corresponding to the corresponding pipeline segment is obtained, and the acoustic wave features of the fourth processed state data corresponding to the heating pipeline are extracted to obtain high-frequency acoustic wave features; the high-frequency acoustic wave features refer to acoustic wave features with frequencies exceeding a preset frequency threshold. Based on the high-frequency acoustic wave characteristics, it is determined whether there is a target high-frequency jet acoustic wave in the fourth processed state data. If so, it is determined that there is a leakage abnormality in the thermal pipeline network. Accordingly, determining that the gas pipeline network has leakage anomalies includes: Obtain the corresponding fifth-processed state data for the corresponding pipeline segment, and identify the temperature change trend of the corresponding pipeline segment based on the fifth-processed state data; If the temperature change trend shows a downward trend, it is determined that there is an abnormal leakage in the gas pipeline network.

[0012] Secondly, this application discloses a leakage monitoring device for urban underground pipe networks, comprising: The acquisition module is used to acquire data collected by multiple target devices installed at various urban underground pipe networks located in the target area within a preset time period to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors; the distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. The processing module is used to preprocess the initial pipeline network status data to obtain processed status data; The input module is used to input the processed state data into the trained pipeline leakage detection model, so as to perform leakage detection on each of the urban underground pipelines based on the processed state data and obtain leakage detection results. The pipeline leakage detection model is a model obtained by training a neural network model based on a self-attention mechanism using historical pipeline status data.

[0013] Thirdly, this application discloses an electronic device, including a processor and a memory; wherein, when the processor executes a computer program stored in the memory, it implements the aforementioned method for monitoring leakage in urban underground pipe networks.

[0014] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for monitoring leakage in urban underground pipe networks.

[0015] As can be seen, this application first acquires data collected by multiple target devices installed at various urban underground pipe networks located in the target area within a preset time period to obtain initial pipe network status data. The target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors. The distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. Next, the initial pipe network status data is preprocessed to obtain processed status data, and the processed status data is input into the trained pipe network leakage detection model to perform leakage detection on each of the urban underground pipe networks based on the processed status data, thereby obtaining leakage detection results. This application pre-installs multiple different types of sensor acquisition devices, such as pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors, at various urban underground pipe networks. The data monitored by these multiple sensor acquisition devices is pre-processed, and the processed multimodal data is input into a neural network model based on a self-attention mechanism for leakage detection. Compared to single-modal data leakage detection, fusing data from multiple sensor acquisition devices significantly improves the accuracy of urban underground pipe network leakage detection. Since the entire leakage detection process requires no manual intervention, it saves labor and time costs. Furthermore, it allows for simultaneous leakage detection of multiple different types of urban underground pipe networks within the target area, thereby improving the efficiency of leakage detection. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] Figure 1 This is a flowchart of a method for monitoring leakage in urban underground pipe networks disclosed in this application; Figure 2 This application discloses a specific method for monitoring leakage in urban underground pipe networks; Figure 3 This is a schematic diagram of the structure of a leakage monitoring device for urban underground pipe networks disclosed in this application; Figure 4 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] This application discloses a method for monitoring leakage in urban underground pipe networks. (See also...) Figure 1 As shown, the method includes: Step S11: Obtain data collected by multiple target devices installed at various urban underground pipe networks in the target area within a preset time period to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed optical fiber sensors, and level gauge sensors; the distributed optical fiber sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing.

[0020] Understandably, the urban underground pipeline system (which includes various types of urban underground pipelines) is the core lifeline for maintaining the normal operation of a city, covering various pipeline types such as water supply, drainage, heating, and gas. In this application, in order to simultaneously detect leakage in multiple different types of underground pipe networks within an urban underground pipe network system, the target area for leakage monitoring can be pre-divided into several relatively closed and independent monitoring blocks. Based on the monitoring requirements of each block (such as pressure, flow rate, temperature, vibration, etc.), multiple data acquisition devices (such as any combination of pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, level gauge sensors, and pipe inspection robots (such as drainage pipe inspection robots used to collect video streams from drainage pipes)) and key nodes in different blocks are determined. Then, corresponding data acquisition devices are deployed at the key nodes so that the data collected by the data acquisition devices installed at different pipe network locations can be used to perceive the operating status of the pipe network in real time, thereby accurately diagnosing whether there is leakage or abnormality, and achieving efficient operation and maintenance of the urban underground pipe network system.

[0021] Specifically, the process begins by collecting GIS pipeline maps of water supply, drainage, heating, and gas pipelines, including data on pipe diameter, material, burial depth, and node coordinates. The location and type of ancillary facilities such as valve wells, inspection wells, pressure regulating boxes, and heat exchange stations are also confirmed. Next, data on road networks, land parcel boundaries, and terrain elevation are collected. All data is then converted to a geodetic coordinate system and imported into a Geographic Information System (GIS) layer by layer. Then, rectangular grids covering the entire target area are generated at intervals. This can be done by first dividing the target area into smaller grids, and then further subdividing the water supply, drainage, heating, and gas pipelines into even smaller grids to obtain the final grid.

[0022] Furthermore, the grid division can be adjusted and optimized by first checking the integrity, sealing, and thermal independence of the water supply network, drainage network, heating network, and gas network within the grid, and then adjusting and optimizing the pipe sections that do not meet the monitoring requirements, and clarifying the boundary points of the sections.

[0023] Then, the boundary nodes of the water supply network are determined as the inlet and outlet of valve wells, as well as the installation locations of flow sensors, water quality sensors, and pressure sensors; the boundary nodes of the drainage network are the inspection wells at the lowest points; the boundary nodes of the heating network are the shut-off valve wells for supply and return water; and the boundary nodes of the gas network are valve wells or pressure regulating boxes. Finally, each block is uniformly coded to achieve independent monitoring within the block and decoupling between blocks, thus laying the foundation for subsequent sensor deployment and multi-source data fusion, and ensuring the accuracy of monitoring different areas of different networks.

[0024] In one specific implementation, pressure sensors, flow sensors, water quality sensors, and distributed fiber optic sensors can be deployed in the water supply network to collect acoustic and temperature sensing data. The water quality sensors are used to collect water quality parameters such as turbidity, pH value, and residual chlorine. Flow sensors and level gauges are deployed in the inspection wells of the drainage network, and robot deployment ports are reserved in key network sections to deploy pipeline inspection robots and distributed fiber optic sensors to collect acoustic sensing data. Distributed fiber optic sensors (for collecting temperature, strain, and acoustic sensing data), flow sensors, and pressure sensors are deployed in the water supply and drainage pipelines of the heating network. Low-power gas concentration sensors are deployed in the valve wells and pressure regulating boxes of the gas pipeline network as auxiliary verification for fiber optic monitoring, and pressure sensors, flow sensors, and distributed fiber optic sensors (for collecting temperature and acoustic sensing data) are also deployed therein.

[0025] Among them, distributed fiber optic sensors can be deployed for multiple purposes with a single cable. A multi-core special temperature-sensing, vibration-sensing, and strain-sensing optical fiber is laid along the same trench. This optical fiber uses different core bundles and optical time-domain reflectometry to simultaneously realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. For example, by capturing acoustic vibration disturbances along the pipeline, acoustic wave sensing can be achieved to determine leakage jets, mechanical impacts, and negative pressure waves; by measuring changes in the pipeline temperature field distribution, abnormal temperature at leak points and damage to the heating insulation layer can be determined; and by monitoring the circumferential or axial strain of the pipeline, ground settlement, pipe deformation, and thermal expansion and contraction stress can be determined.

[0026] In this embodiment, data collected by multiple target devices (such as pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors) installed in various types of urban underground pipe networks located in a target area within a preset period or in real time can be received to obtain initial pipe network status data in sequence form.

[0027] Step S12: Preprocess the initial pipeline network status data to obtain processed status data.

[0028] In this embodiment, after obtaining the initial pipeline network status data in multiple modes, the initial pipeline network status data can be preprocessed (such as data cleaning, format conversion, etc.) and spatiotemporally aligned to form status data that can accurately reflect the operation of the pipeline network.

[0029] In one specific implementation, the preprocessing of the initial pipeline network status data to obtain processed status data may specifically include: sampling the initial pipeline network status data collected by each target device according to a preset sampling interval to obtain sampled pipeline network status data; identifying and removing outliers in the sampled pipeline network status data to obtain first removed status data; counting the number of all outliers in each sampled pipeline network status data to obtain a first outlier count, and calculating the ratio of the first outlier count to the total number of status values ​​to obtain a first ratio; the total number of status values ​​is the number of all status values ​​in the sampled pipeline network status data; determining whether the first ratio is less than a first threshold; if the first ratio is less than the first threshold and the missing points at different outliers are isolated from each other, then using interpolation to fill the missing points in the first removed status data to obtain filled status data; identifying and removing outliers in the filled status data to obtain second removed status data; and performing preprocessing operations on the second removed status data to obtain processed status data. Specifically, the timestamps of the initial pipeline status data corresponding to all target devices are first converted to a unified reference. The converted pipeline status data is then resampled at equal intervals. Next, the units of the sampled pipeline status data are normalized; for example, all units are converted to standard physical units, where pressure is in MPa, flow rate is in m³ / h, water quality is in mg / L, temperature is in ℃, and acoustic sensor data is in dimensionless relative amplitude. Additionally, a data storage structure can be set to organize the normalized data into a tabular format, for example, setting rows as time points and columns as sensor IDs and physical quantities for batch processing.

[0030] Furthermore, outliers in the normalized pipeline network status data are identified and removed to obtain the first set of outlier status data. Then, statistical analysis is performed on newly generated missing points. For example, the number of all outliers in the sampled pipeline network status data corresponding to each target device is counted to obtain the first number of outliers. The ratio of the first number of outliers to the total number of status values ​​in the sampled pipeline network status data (i.e., the entire data window) is calculated to obtain the missing proportion. If this missing proportion is less than 5% and the missing points at different outlier locations are isolated missing points, linear interpolation or cubic spline interpolation is used to fill the missing points in the first set of outlier status data to obtain filled status data. For other data points in the filled status data that were not removed (i.e., non-outliers) and those filled by interpolation, the 3σ criterion or interquartile range method can be used to identify and remove univariate outliers, thus completing all cleaning and outlier removal operations to obtain the second set of outlier status data. Finally, the second set of outlier status data is preprocessed to obtain processed status data. The above data processing operations ensure the continuity and reliability of the pipeline network status data.

[0031] In addition, if the length of consecutive missing data accounts for 10% to 30% of the entire window (i.e., the total number of all state values ​​in the sampled pipeline state data), extrapolation interpolation can be performed by utilizing the changing trend of valid data before and after the missing segment, or historical data segments under similar working conditions can be found for matching and filling. If the proportion of consecutive missing data exceeds 50% of the entire window, the data segment is considered to have lost its analytical value and is directly marked as invalid. No further interpolation processing is performed. At this time, the data acquisition in step S11 and the preprocessing operation in step S12 can be performed again.

[0032] Specifically, identifying and removing outliers from the sampled pipeline network status data to obtain the first removed status data may include: identifying status values ​​in the sampled pipeline network status data that exceed a preset status threshold, and treating these outliers as outliers; sequentially calculating the difference between any two adjacent status values ​​in the sampled pipeline network status data, and determining whether the difference exceeds a second threshold; if the difference exceeds the second threshold, then treating the corresponding status value as an outlier; and removing the outliers from the sampled pipeline network status data to obtain the first removed status data. For example, based on the range and operating limits of each sensor (such as the maximum allowable value of a pressure sensor, the physical upper limit of a flow sensor, etc.), data (i.e., status values) in the sampled pipeline network status data that exceed a reasonable range (such as a preset status threshold) are directly identified as outliers and removed, and the specific location of the removal can be recorded. Further, a secondary screening of outliers is performed. Specifically, the difference between the status values ​​at any two adjacent moments in the sampled pipeline network status data is calculated, and then it is determined whether the difference exceeds a second threshold; if it does, the corresponding status value is treated as an outlier and removed.

[0033] Alternatively, secondary filtering can be performed based on the rate of change of the time series. The ratio of the difference to the corresponding sampling interval can be calculated to obtain the rate of change. If this rate of change exceeds the maximum rate of change for the corresponding sampling interval, the corresponding state value is considered an outlier and removed.

[0034] Specifically, when the urban underground pipe network includes water supply, drainage, heating, and gas pipelines, the preprocessing operation on the second excluded state data to obtain processed state data may include: denoising the second excluded state data corresponding to the pressure sensor using low-pass filtering to obtain first denoised state data, and determining whether there is baseline drift in the pressure signal of the first denoised state data; if there is baseline drift in the first denoised state data, removing the trend term from the first denoised state data to obtain the first processed state data; denoising the second excluded state data corresponding to the flow sensor using median filtering to obtain second denoised state data, and performing node flow balance verification on the second denoised state data to obtain a verification result; if the verification result indicates that the inflow flow at each node of the urban underground pipe network is equal to the flow rate at each node, the preprocessing operation may be performed on the second denoised state data to obtain the processed state data. If the outflow rate does not satisfy the conservation relationship, the second denoised state data is corrected using the least squares method to obtain the second processed state data. Temperature compensation is applied to the second rejected state data corresponding to the water quality sensor to obtain compensated state data, and piecewise linear interpolation is used to interpolate the compensated state data to obtain the third processed state data. The acoustic sensing data in the second rejected state data corresponding to the distributed optical fiber sensor is denoised to obtain the third denoised state data, and independent component analysis is used to separate the leakage acoustic waves in the third denoised state data from other interference sources to obtain the fourth processed state data. The temperature sensing data in the second rejected state data corresponding to the distributed optical fiber sensor is denoised to obtain the fourth denoised state data, and trend correction is applied to the fourth denoised state data to obtain the fifth processed state data. In this embodiment, for scenarios where water supply networks, drainage networks, heating networks, and gas networks exist simultaneously in the target area, corresponding preprocessing operations can be performed on the sensor data at different network locations. For example, for the second rejected state data corresponding to the pressure sensor, a low-pass filter is first used to denoise it to obtain the first denoised state data. The cutoff frequency is set to 1.2 times the effective frequency of the signal (which can be a preset frequency) to retain the main pressure change characteristics and suppress high-frequency noise. Then, it is determined whether there is baseline drift in the pressure signal in the first denoised state data. If there is, that is, the overall trend of the pressure signal shifts with time, a high-pass filter is used or the trend term in the first denoised state data is subtracted by polynomial fitting to obtain a stable pressure signal (i.e., the first processed state data).

[0035] For the second denoised state data corresponding to the flow sensor (which is usually mixed with impulse noise), the impulse noise in the second denoised state data is first suppressed by a median filter with a window length of 3 to 5, resulting in the second denoised state data. This can remove isolated noise points while retaining the edge information of flow changes. Subsequently, the node flow balance is checked on the above second denoised state data, that is, whether the inflow flow and outflow flow (including leakage flow) at each node satisfy the conservation relationship. If not, the least squares method is used to correct each state value in the second denoised state data, so that the overall flow data is physically closed, thereby improving the reliability of subsequent leakage detection.

[0036] For the second set of rejected state data (including water quality parameters such as turbidity, pH, and residual chlorine) corresponding to the water quality sensor, firstly, all state values ​​in the second set of rejected state data are corrected to the standard temperature (e.g., 25℃) according to the preset temperature compensation formula to eliminate the influence of temperature on the sensor's sensitive element and obtain the compensated state data. Then, using the reference values ​​obtained from periodic calibration, the compensated state data is interpolated using a piecewise linear interpolation method to deduct instrument drift, thereby obtaining a stable and reliable water quality time series (i.e., the third set of processed state data).

[0037] For the second rejection state data (including temperature sensing data, strain sensing data, and acoustic wave sensing data) corresponding to the distributed optical fiber sensor, different data processing methods can be used to preprocess the different sensing data. For example, for acoustic wave sensing data, wavelet denoising is first performed, with a decomposition layer of 5 to 8 layers. Each layer uses soft threshold denoising to effectively remove broadband noise, resulting in the third denoised state data. Next, adaptive filtering is applied to the third denoised state data to suppress environmental noise. Specifically, acoustic wave signals during periods without leakage can be selected as background noise templates, and a Normalized Least Mean Squares (NLMS) adaptive filter is used to offset deterministic environmental noise generated by vehicle traffic, construction activities, etc., in real time. Finally, leakage acoustic wave separation is performed on the filtered state data. Specifically, frequency characteristic differences can be utilized (leakage signal energy is mainly concentrated in the 50–200 Hz range, vehicle noise is mostly distributed in the low-frequency band below 20 Hz, and construction noise is intermittent broadband). First, the filtered state data is bandpass filtered in the 50–200 Hz range and the energy envelope is extracted. If multiple sound sources are mixed, independent component analysis can be used to further separate the leakage acoustic waves from other interference sources, resulting in the fourth processed state data.

[0038] For example, for temperature sensing data, firstly, median filtering with a window length of 3 to 5 is used to denoise the data to remove isolated outliers caused by scattering noise. Then, Gaussian smoothing is used to further suppress residual fluctuations, resulting in the fourth denoised state data. Next, spatial domain anomaly removal is performed on the fourth denoised state data. Specifically, the neighborhood mean and standard deviation can be calculated point by point along the distance dimension of the fiber optic sensor. Data points that deviate from the neighborhood mean by more than 4 times the standard deviation are identified as anomalies caused by fiber breakage or connector loss and removed. Finally, if the monitored ambient temperature is constant but the data shows slow drift, the time domain trend correction is performed on the removed state data by using an exponentially weighted moving average or by directly subtracting the linear trend term, thereby obtaining an accurate temperature field distributed in the spatiotemporal along the fiber, resulting in the fifth processed state data.

[0039] Step S13: Input the processed state data into the trained pipeline network leakage detection model to perform leakage detection on each of the urban underground pipeline networks based on the processed state data, and obtain leakage detection results; wherein, the pipeline network leakage detection model is a model obtained by training a neural network model based on a self-attention mechanism using historical pipeline network state data.

[0040] In this embodiment, the processed state data can be fused / combined according to different pipe networks. For example, for a water supply pipe network, the processed state data corresponding to pressure sensors, flow sensors, water quality sensors, and acoustic and temperature sensor data collected by distributed optical fibers can be fused. Then, the fused data is input into the pipe network leakage detection model in the form of feature vectors. This model is obtained by training a neural network model based on a self-attention mechanism using historical pipe network state data, thereby detecting leakage in the water supply pipe network and obtaining corresponding leakage detection results. The leakage detection results may include, but are not limited to, the probability of the existence of a leak, its precise location, type, and severity level.

[0041] For drainage pipe networks, the processed status data corresponding to the video streams collected by level gauge sensors, flow sensors, pipeline robots, and acoustic sensing data collected by distributed fiber optic sensors can be fused.

[0042] For thermal pipeline networks, the processed status data corresponding to temperature sensing data, strain sensing data, and acoustic wave sensing data collected by supply and return water temperature sensors, pressure sensors, flow sensors, and distributed optical fiber sensors can be fused.

[0043] For gas pipeline networks, the processed status data from pressure sensors, flow sensors, distributed fiber optic sensors, acoustic and temperature sensors, and low-power gas concentration sensors can be fused together.

[0044] In this embodiment, see Figure 2 As shown, the leakage detection of each of the city's underground pipe networks based on the processed state data, to obtain leakage detection results, may specifically include: Step S21: Determine the location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply network, and obtain the target optical fiber location.

[0045] In this embodiment, for the water supply network, the processed state data corresponding to the acoustic wave sensing data and temperature sensing data in the distributed optical fiber sensor can be used as the dominant factor for leakage detection. Specifically, the fourth processed state data corresponding to the water supply network is first analyzed. If abnormal acoustic waves are found, the location of the corresponding distributed optical fiber sensor is determined, and the target optical fiber location is obtained.

[0046] Step S22: Based on the first processed state data and the second processed state data corresponding to the target optical fiber location, calculate the pressure drop rate and flow imbalance of the corresponding pipeline segment.

[0047] In this embodiment, when the distributed optical fiber sensor detects abnormal sound waves, it can further retrieve the processed status data corresponding to the pressure sensors and flow sensors arranged upstream and downstream of the pipeline segment where the target optical fiber is located, and then calculate the pressure drop rate and flow imbalance of the pipeline segment based on the processed status data.

[0048] Step S23: If the pressure drop rate exceeds the preset rate threshold and the flow imbalance is greater than the third threshold, then it is determined that there is a leakage abnormality in the water supply network.

[0049] In this embodiment, if the pressure drop rate exceeds the preset rate threshold and the flow imbalance between upstream and downstream is greater than 5%, it is determined that the corresponding pipeline section has indeed experienced leakage. This determination method can effectively eliminate false alarms caused by environmental factors such as vehicle movement or construction vibration.

[0050] In addition, the processed status data corresponding to water quality sensor and temperature sensor data can be integrated to further determine whether there is a leak in the water supply network. For example, when a leak causes external groundwater to seep in, the water quality sensor will detect an increase in turbidity. At the same time, the temperature sensor data of the distributed sensor can detect the local temperature drop of the soil caused by the cold water leak. Therefore, it is possible to analyze whether the turbidity increases and whether the soil temperature decreases to further determine whether there is a leak and the cause of the leak.

[0051] Through the above-mentioned multi-dimensional leakage detection mechanism, the severity of the leakage can be comprehensively assessed, whether it is a micro-leak, seepage, or pipe burst. Finally, the precise three-dimensional GIS (Geographic Information System) coordinates of the leakage point, the leakage type (such as joint damage or pipe crack), the severity level, and the predicted leakage amount are output, thus providing accurate decision-making basis for the timely repair of the water supply network.

[0052] Step S24: Calculate the temperature change of each section of the drainage pipe network based on the fifth processed state data corresponding to the drainage pipe network, and obtain the first temperature change.

[0053] For drainage pipe networks, data collected by distributed fiber optic sensors can be used primarily for detection. For example, in rainy conditions, distributed fiber optic sensors can detect sudden temperature changes caused by rainwater infiltration. This can be combined with the liquid level change trend measured by upstream liquid level sensors, such as a sudden drop in liquid level, to accurately locate pipe damage or misaligned interfaces. In addition, distributed fiber optic sensors can also identify low-frequency (<50Hz) sounds of mud and sand impact or air churning, and combined with data showing that downstream flow rate is consistently below a preset flow threshold, to further determine the degree of siltation and blockage risk inside the pipe.

[0054] In this embodiment, the temperature change of each section of the drainage network (such as the difference between the current temperature and the historical temperature at the previous moment) is calculated based on the fifth processed state data (corresponding temperature sensing data) of the drainage network to obtain the first temperature change.

[0055] Step S25: If the first temperature change exceeds the fourth threshold, the liquid level change trend and liquid level change amount of the corresponding pipeline segment are identified based on the processed status data corresponding to the liquid level sensor upstream of the corresponding pipeline segment.

[0056] In this embodiment, if the temperature change exceeds a preset threshold, such as the temperature drop exceeding a preset threshold, the processed status data of the level sensor upstream of the corresponding pipeline segment is obtained, and the liquid level change trend (such as liquid level drop) is analyzed based on the processed status data, and the liquid level change (i.e., drop) is calculated.

[0057] Step S26: If the liquid level change trend shows a downward trend and the liquid level change exceeds the fifth threshold, then it is determined that there is an abnormal leakage in the drainage network.

[0058] In this embodiment, if the liquid level changes downward and the amount of liquid level change exceeds a preset threshold, it indicates a sudden drop in liquid level. At this time, it can be determined that there is an abnormal leakage in the drainage network.

[0059] In addition, when the fiber optic sensor detects a clear anomaly, it can automatically dispatch nearby pre-set pipeline inspection robots for precise verification. For example, the pipeline inspection robot autonomously travels to the corresponding pipeline section based on the suspicious distance coordinates provided by the fiber optic cable, transmits video streams in real time, and inputs the video streams into a visual model to identify physical damage such as cracks, root intrusion, and deposition patterns.

[0060] Furthermore, visual features (such as crack width and deposition thickness) in the video stream collected by the pipeline inspection robot can be correlated with vibration intensity and temperature changes detected by fiber optic sensors to gradually establish a mapping library from fiber optic signal features to the actual physical damage level. In this way, the damage size can be accurately inferred from the fiber optic signal alone in the subsequent real-time leak detection process without robot intervention, thereby greatly improving the automation and intelligence level of drainage pipe network inspection.

[0061] Step S27: Calculate the temperature change of each section of the heating network based on the fifth processed state data corresponding to the heating network, and obtain the second temperature change.

[0062] In this embodiment, for leak detection in the thermal pipeline network, the detection can be primarily based on the processed state data corresponding to the temperature sensing data (which can provide a continuous temperature distribution at the centimeter level) collected by the fiber optic sensor. First, the temperature change of each pipeline segment in the thermal pipeline network (such as the difference between the current temperature and the historical temperature at the previous moment) is calculated based on the fifth processed state data corresponding to the fiber optic sensor at the thermal pipeline network, thus obtaining the second temperature change.

[0063] Step S28: If the second temperature change exceeds the fifth threshold, determine whether there is a target high-frequency jet sound wave in the fourth processed state data corresponding to the corresponding pipeline segment. If so, determine that there is a leakage abnormality in the thermal pipeline.

[0064] In this embodiment, if the second temperature change exceeds a preset threshold, for example, the temperature change of the supply and return water heating network is compared with the preset temperature gradient in real time. If an abnormal temperature drop (e.g., temperature drop > 2℃, indicating hot water leakage) or an abnormal temperature rise (e.g., temperature rise > 3℃, possibly indicating a return water short circuit) is found at a certain point, the corresponding node (e.g., a distributed fiber optic sensor) is used as a candidate leak point, and its precise spatial coordinates are obtained.

[0065] Furthermore, the system automatically retrieves acoustic wave sensing data (which is a continuous recording of acoustic wave signals along the line with high spatial resolution (e.g., 2m) and temporal resolution (e.g., 5kHz) from distributed sensors within a 10-meter range before and after the candidate leak point, corresponding to the fourth processed state data. Then, it determines whether there is a high-frequency jet sound (referring to a jet sound with a frequency exceeding a preset frequency) in the fourth processed state data for double confirmation and location. If it exists, it is determined that there is a leakage anomaly in the thermal pipeline network.

[0066] In addition, distributed fiber optic sensors can be used to dynamically assess the stress and lifespan of pipelines through strain sensing. For example, during the start-up and shutdown of a heating network, distributed fiber optic sensors measure the circumferential and axial strain of the pipeline in real time through strain sensing. Then, thermal stress is calculated according to σ = E·ε (where E is the elastic modulus and ε is the measured strain). Combined with the pipe's SN curve (fatigue curve of the material), the stress amplitude of each start-up and shutdown cycle is counted using the rainflow counting method, thereby accumulating the fatigue damage degree D = Σ (ni / Ni) in real time. When the D value exceeds a preset threshold (such as 0.8), an immediate warning is issued that the pipeline is nearing the end of its fatigue life.

[0067] Simultaneously, multi-sensor data fusion can be performed, cross-correlation analysis can be conducted between the rate of temperature change at the leak point measured by the DTS (Distributed Fiber Optic Temperature Measurement System) and the rate of pressure drop captured by the pressure sensor, thereby effectively eliminating false temperature fluctuations caused by normal valve operation. Finally, the precise location of the leak point, heat loss power (kW / m), and remaining fatigue life of the pipeline (years) are output, thus providing quantitative indicators for preventive maintenance and replacement of the heating network.

[0068] Step S29: Based on the fourth processed state data corresponding to the gas pipeline network, determine whether there are ultrasonic waves of a specific frequency band in each pipeline segment of the gas pipeline network.

[0069] Understandably, in a gas pipeline network, when a minor leak occurs, high-pressure gas will be ejected from tiny holes, generating ultrasonic waves of a specific frequency band (e.g., 200-500Hz). Distributed fiber optic sensors, using phase-sensitive optical time-domain reflectometry, can detect the minute vibrations caused by these ultrasonic waves propagating through the soil. Furthermore, a matched-filter algorithm can accurately extract this characteristic frequency from strong environmental noise, enabling meter-level location of the leak and estimation of the leak volume. Therefore, based on the fourth-processed state data (corresponding to acoustic sensing data) of the gas pipeline network, it can be determined whether ultrasonic waves of a specific frequency band (e.g., 200-500Hz) exist in each section of the gas pipeline network.

[0070] Step S210: If a specific frequency band of ultrasonic waves exists in a section of the gas pipeline, it is determined that there is an abnormal leakage in the gas pipeline.

[0071] In this embodiment, if there are ultrasonic waves of a specific frequency band (such as 200~500Hz) in a section of the gas pipeline, the temperature change trend and the amount of temperature change are further determined based on the fifth processed state data (corresponding temperature sensing data) of the corresponding pipeline section. If the temperature change trend is a temperature drop and the amount of temperature change meets the preset conditions (such as being between 0.1~0.5°C), it is determined that there is a leakage abnormality in the gas pipeline.

[0072] Furthermore, it can automatically notify inspection drones or dispatch staff with handheld gas detectors to the corresponding GPS coordinates to measure methane concentration (ppm), and use the methane concentration (ppm) as the final basis for determining leakage anomalies.

[0073] It should be noted that the entire detection process does not require on-site power supply, which can completely eliminate the risk of electric sparks. Ultimately, it can output the accurate location of the leak point, the estimated leakage amount, and the hazard level based on the explosion-proof zone, thereby providing real-time and reliable data-driven decision support for emergency repairs of gas pipeline networks.

[0074] In addition, after determining that there is a leakage anomaly, it is also possible to determine the probability of existence, precise location, type, and severity level of the corresponding leakage point, and generate corresponding early warning information.

[0075] Specifically, the self-attention mechanism of the basic probability assignment function in distributed fiber optic sensing evidence theory can be utilized, and the confidence levels of different modal features (referring to the features of the fused data) can be dynamically weighted through a pipeline leakage detection model to handle sensor noise and local false alarms. After multi-layer nonlinear mapping and cross-modal evidence fusion, a comprehensive evaluation result in four dimensions can be output, such as: the probability of a leak (normalized to a value between 0 and 1, with a preset alarm threshold of 0.7; exceeding this threshold triggers an early warning); precise location (after fusing GIS geographic coordinates, the overall positioning error can be controlled within 1 meter); type (distinguishing between various operating conditions such as pipe bursts, leaks, cracks, loose interfaces, blockages, deposits, and fatigue cracking); and severity level (from Level I minor anomaly to Level IV emergency pipe burst or leak). By outputting the above information, an efficient, accurate, and robust comprehensive evaluation of any pipeline event can be achieved, providing a reliable basis for subsequent early warning and emergency decision-making.

[0076] This application can also implement early warning operations through a pre-established multi-level prediction and early warning mechanism. For example, it can dynamically generate risk heat maps of underground pipe networks in various cities. Specifically, by accumulating historical leakage data and real-time monitoring results over a long period, the leakage probability of each monitoring block can be mapped to a red-yellow-green gradient heat map layer, with redder colors indicating a higher risk of leakage / failure in the current or future area. Then, differentiated early warning pushes can be implemented based on the severity level. For Level IV emergency events (such as pipe bursts, high-concentration gas leaks, etc.), upstream and downstream valves can be automatically shut down to isolate the affected section, and early warning information can be simultaneously pushed to the emergency command center, fire department, and corresponding water or gas companies. The push content can include precise 3D GIS coordinates, the expected water / gas outage area, and a list of affected populations / facilities. For Level I or II minor to moderate warnings (such as minor leaks, initial blockages, mild fatigue cracking, etc.), they can be included in the routine maintenance plan, and the optimal robot inspection window (such as low water load periods or non-heating seasons) can be indicated to arrange preventative maintenance and prevent the situation from escalating.

[0077] Furthermore, corresponding abnormal early warning information can be generated for different urban underground pipe networks, and this information can be pushed to the corresponding management terminals. Using a digital twin simulation model, leak points with abnormal leakage can be reproduced in a virtual environment, thereby simulating maintenance plans for these leak points. For example, the system can accurately reproduce located leak points in the virtual environment and simulate the impact range of water and gas outages and possible secondary disasters (such as ground subsidence and water pollution spread) under different valve closure schemes, pressure reduction strategies, or emergency repair paths. This assists emergency command personnel in intuitively comparing and selecting the optimal dispatch plan and feeding back decision instructions to the on-site execution terminals.

[0078] Through the organic linkage of the aforementioned dynamic risk heat map, multi-level early warning push and digital twin simulation, a closed loop of the entire chain from risk perception and intelligent assessment to accurate early warning and auxiliary decision-making has been truly realized. This enables all-weather, high-precision, multi-source fusion intelligent detection and early warning of leakage and faults in urban underground pipe networks, thereby significantly improving the accuracy of leak point location and the response time of emergency response in urban underground pipe networks.

[0079] Specifically, determining the location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply network, and obtaining the target optical fiber location, may include: extracting acoustic features from the fourth processed state data corresponding to the water supply network to obtain broadband acoustic features; the broadband acoustic features include any one or more of kurtosis features, short-time energy features, and spectral centroid features; detecting whether the broadband acoustic features meet preset requirements; the preset requirements are that the kurtosis feature is less than a preset kurtosis threshold, and / or the short-time energy feature is less than a preset energy threshold, and / or the spectral centroid feature is located within a preset frequency domain distribution centroid range; if the broadband acoustic features do not meet the preset requirements, then determining the location of the distributed optical fiber sensor corresponding to the fourth processed state data, and obtaining the target optical fiber location. In this embodiment, acoustic features are first extracted from the fourth processed state data corresponding to the water supply network along the optical fiber. If broadband acoustic features caused by leakage are extracted, such as kurtosis features, short-time energy features, and spectral centroid features, then it is further detected whether the broadband acoustic features meet preset requirements. If they are met, such as the kurtosis feature being less than the preset kurtosis threshold, the short-time energy feature being less than the preset energy threshold, and the spectral centroid feature being within the preset frequency domain distribution centroid range, then it is determined that there is no leakage. If not, then the suspected leakage point is locked within ±2 meters using dual-end optical time-domain positioning technology, that is, the location of the current distributed optical fiber sensor is locked.

[0080] Specifically, determining whether a target high-frequency jet sound wave exists in the fourth processed state data corresponding to the corresponding pipeline segment, and if so, determining that the heating pipeline has a leakage anomaly, may include: acquiring the fourth processed state data corresponding to the corresponding pipeline segment, and extracting acoustic features from the fourth processed state data corresponding to the heating pipeline to obtain high-frequency acoustic features; the high-frequency acoustic features refer to acoustic features with frequencies exceeding a preset frequency threshold; and determining whether a target high-frequency jet sound wave exists in the fourth processed state data based on the high-frequency acoustic features, and if so, determining that the heating pipeline has a leakage anomaly. It is understood that high-frequency jet sound waves caused by leakage (usually continuous) typically have a characteristic frequency range (energy concentrated in 200-800Hz). For example, when hot water leaks and sprays out, it generates broadband sound waves with significant high-frequency components. Therefore, leakage anomalies can be detected using high-frequency jet sound waves. In this embodiment, the fourth processed state data (corresponding to acoustic sensing data) corresponding to the corresponding pipeline segment can be obtained first. Then, the high-frequency jet sound wave in the fourth processed state data can be filtered by a bandpass filter to extract the sound wave component suspected of leakage, thus obtaining filtered state data. Next, the short-time energy envelope of the filtered state data is calculated to identify signal segments with longer duration and stable energy, and the spectral centroid of the signal segment is calculated. Then, it is determined whether it falls within the preset frequency band corresponding to the leakage sound wave, such as 300-600Hz. In addition, the frequency corresponding to the maximum amplitude can be determined and compared with the preset frequency corresponding to the leakage (such as 450Hz). Since the leakage sound wave usually manifests as continuous and stable random noise rather than transient impact (such as valve closing sound), the variance of the energy envelope can also be calculated, and it can be determined whether the variance is less than a preset threshold (if it is less, it indicates stability and continuity).

[0081] Specifically, the high-frequency sound wave characteristics can be matched with a preset leakage sound pattern template. If the maximum value of the energy envelope after bandpass filtering exceeds the background noise and the center of the spectrum is within a certain range, it is determined to be a leakage sound wave.

[0082] Accordingly, determining that the gas pipeline network has a leakage anomaly can specifically include: acquiring the corresponding fifth-processed state data of the relevant pipeline segment, and identifying the temperature change trend of the relevant pipeline segment based on the fifth-processed state data; if the temperature change trend shows a downward trend, then it is determined that the gas pipeline network has a leakage anomaly. It is understood that the Joule-Thomson effect caused by natural gas leakage will lead to a slight local temperature drop (0.1~0.5°C), and the temperature sensing of the distributed fiber optic sensor can capture this subtle change to assist the acoustic wave sensing of the distributed fiber optic sensor, thereby reducing the false alarm rate. That is, when the temperature is confirmed to be abnormal, and the acoustic wave is also confirmed to be abnormal, then the corresponding candidate point is finally confirmed as the actual leak point. Through the above method, the reliability and location accuracy of thermal pipeline network leak detection can be improved.

[0083] Additionally, if there is a temperature anomaly (e.g., a temperature drop exceeding a preset threshold) but the acoustic sensor does not detect a leakage sound wave, possible causes include: instantaneous temperature fluctuations caused by normal valve operation (e.g., short-term return water short circuit), localized damage to the insulation layer without leakage, and occasional sensor drift. In this case, no leakage alarm needs to be generated; it should only be recorded as a "temperature anomaly event" for manual review. If the acoustic sensor detects a leakage sound wave but there is no temperature anomaly, it may be from other sound sources (e.g., nearby construction or passing vehicles), and in this case, no leakage anomaly is determined.

[0084] As can be seen, in this embodiment, multiple different types of sensor acquisition devices, such as pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors, are first installed at different urban underground pipe networks. The data monitored by the multiple sensor acquisition devices are preprocessed, and the processed multimodal data is input into a neural network model based on a self-attention mechanism for leakage detection. Compared with the leakage detection method based on single-modal data, by fusing the data monitored by multiple sensor acquisition devices, the accuracy of leakage detection in urban underground pipe networks can be significantly improved. Since the entire leakage detection process does not require manual intervention, it saves labor and time costs. Furthermore, it can simultaneously perform leakage detection on multiple different types of urban underground pipe networks within the target area, thereby improving the efficiency of leakage detection.

[0085] Accordingly, this application also discloses a leakage monitoring device for urban underground pipe networks, see [link to relevant documentation]. Figure 3 As shown, the device includes: The acquisition module 11 is used to acquire data collected by multiple target devices installed at various urban underground pipe networks in the target area within a preset time period to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed optical fiber sensors, and level gauge sensors; the distributed optical fiber sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. Processing module 12 is used to preprocess the initial pipeline status data to obtain processed status data; Input module 13 is used to input the processed state data into the trained pipeline leakage detection model, so as to perform leakage detection on each of the urban underground pipelines based on the processed state data and obtain leakage detection results; The pipeline leakage detection model is a model obtained by training a neural network model based on a self-attention mechanism using historical pipeline status data.

[0086] The specific workflow of each of the above modules can be found in the relevant content disclosed in the foregoing embodiments, and will not be repeated here.

[0087] Furthermore, embodiments of this application also disclose an electronic device, Figure 4 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0088] Figure 4 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the urban underground pipe network leakage monitoring method disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0089] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0090] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0091] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the urban underground pipe network leakage monitoring method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0092] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for monitoring leakage in urban underground pipe networks. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0093] Furthermore, embodiments of this application also disclose a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the urban underground pipe network leakage monitoring method disclosed above.

[0094] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0095] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0096] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0097] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0098] The above provides a detailed description of the method, apparatus, equipment, and medium for monitoring leakage in urban underground pipe networks provided by this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for monitoring leakage in urban underground pipe networks, characterized in that, include: Data collected by multiple target devices installed at various urban underground pipe networks in the target area within a preset time period is obtained to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors; the distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. The initial pipeline network status data is preprocessed to obtain processed status data; The processed state data is input into the trained pipeline leakage detection model to perform leakage detection on each of the city's underground pipelines based on the processed state data, and the leakage detection results are obtained. The pipeline leakage detection model is a model obtained by training a neural network model based on a self-attention mechanism using historical pipeline status data. The preprocessing of the initial pipeline network status data to obtain processed status data includes: sampling the initial pipeline network status data collected by each target device according to a preset sampling interval to obtain sampled pipeline network status data; identifying and removing outliers in the sampled pipeline network status data to obtain first removed status data; counting the number of all outliers in each sampled pipeline network status data to obtain a first outlier count, and calculating the ratio of the first outlier count to the total number of status values ​​to obtain a first ratio; the total number of status values ​​is the number of all status values ​​in the sampled pipeline network status data; determining whether the first ratio is less than a first threshold; if the first ratio is less than the first threshold and the missing points at different outliers are isolated from each other, then using interpolation to fill the missing points in the first removed status data to obtain filled status data; identifying and removing outliers in the filled status data to obtain second removed status data; and performing preprocessing operations on the second removed status data to obtain processed status data. When the urban underground pipe network includes water supply, drainage, heating, and gas pipelines, the preprocessing operation on the second excluded state data to obtain processed state data includes: denoising the second excluded state data corresponding to the pressure sensor using low-pass filtering to obtain first denoised state data, and determining whether there is baseline drift in the pressure signal of the first denoised state data; if there is baseline drift in the first denoised state data, removing the trend term from the first denoised state data to obtain the first processed state data; denoising the second excluded state data corresponding to the flow sensor using median filtering to obtain second denoised state data, and performing node flow balance verification on the second denoised state data to obtain a verification result; if the verification result indicates that the inflow and outflow at each node of the urban underground pipe network are balanced... If the quantity does not satisfy the conservation relationship, the second denoised state data is corrected using the least squares method to obtain the second processed state data; temperature compensation is performed on the second rejected state data corresponding to the water quality sensor to obtain compensated state data, and piecewise linear interpolation is used to interpolate the compensated state data to obtain the third processed state data; the acoustic sensing data in the second rejected state data corresponding to the distributed optical fiber sensor is denoised to obtain the third denoised state data, and independent component analysis is used to separate the leakage acoustic waves in the third denoised state data from other interference sources to obtain the fourth processed state data; the temperature sensing data in the second rejected state data corresponding to the distributed optical fiber sensor is denoised to obtain the fourth denoised state data, and trend correction is performed on the fourth denoised state data to obtain the fifth processed state data; The process of detecting leakage in each of the urban underground pipe networks based on the processed state data to obtain leakage detection results includes: determining the location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply pipe network, and obtaining the target optical fiber location; calculating the pressure drop rate and flow imbalance of the corresponding pipe network segment based on the first and second processed state data corresponding to the target optical fiber location; determining that the water supply pipe network has a leakage anomaly based on the pressure drop rate exceeding a preset rate threshold and the flow imbalance exceeding a third threshold; calculating the temperature change of each pipe network segment of the drainage pipe network based on the fifth processed state data corresponding to the drainage pipe network, and obtaining the first temperature change; determining that the temperature change exceeds the fourth threshold based on the corresponding level sensor upstream of the pipe network segment. The processed state data is used to identify the liquid level change trend and amount of liquid level change in the corresponding pipeline segment. If the liquid level change trend shows a downward trend and the amount of liquid level change exceeds a fifth threshold, it is determined that the drainage pipeline has a leakage anomaly. Based on the fifth processed state data corresponding to the heating pipeline, the temperature change of each pipeline segment of the heating pipeline is calculated to obtain a second temperature change. If the second temperature change exceeds a fifth threshold, it is determined whether there is a target high-frequency jet sound wave in the fourth processed state data corresponding to the corresponding pipeline segment. If so, it is determined that the heating pipeline has a leakage anomaly. Based on the fourth processed state data corresponding to the gas pipeline, it is determined whether there is a specific frequency band of ultrasonic waves in each pipeline segment of the gas pipeline. If there is a specific frequency band of ultrasonic waves in the pipeline segment of the gas pipeline, it is determined that the gas pipeline has a leakage anomaly.

2. The method for monitoring leakage in urban underground pipe networks according to claim 1, characterized in that, The process of identifying and removing outliers from the sampled pipeline network status data to obtain the first removed status data includes: Identify state values ​​in the sampled pipeline status data that exceed a preset state threshold, and treat state values ​​that exceed the preset state threshold as outliers; Calculate the difference between any two adjacent state values ​​in the sampled pipeline network state data in sequence, and determine whether the difference exceeds the second threshold. If the difference exceeds the second threshold, the corresponding state value will be considered an outlier. The outliers in the sampled pipeline status data are removed to obtain the first set of removed status data.

3. The method for monitoring leakage in urban underground pipe networks according to claim 1, characterized in that, The step of determining the location of the distributed optical fiber sensor related to the fourth processed state data corresponding to the water supply network, and obtaining the target optical fiber location, includes: The acoustic wave features are extracted from the fourth processed state data corresponding to the water supply network to obtain broadband acoustic wave features; the broadband acoustic wave features include any one or more of kurtosis features, short-time energy features, and spectral centroid features. Detect whether the broadband acoustic wave characteristics meet preset requirements; the preset requirements are that the kurtosis characteristic is less than a preset kurtosis threshold, and / or the short-time energy characteristic is less than a preset energy threshold, and / or the spectral centroid characteristic is located within a preset frequency domain distribution centroid range; If the broadband acoustic wave characteristics do not meet the preset requirements, the position of the distributed optical fiber sensor corresponding to the fourth processed state data is determined to obtain the target optical fiber position.

4. The method for monitoring leakage in urban underground pipe networks according to claim 1, characterized in that, The step of determining whether a target high-frequency jet sound wave exists in the fourth processed state data corresponding to the corresponding pipeline segment, and if so, determining that the thermal pipeline has a leakage anomaly, includes: The fourth processed state data corresponding to the corresponding pipeline segment is obtained, and the acoustic wave features of the fourth processed state data corresponding to the heating pipeline are extracted to obtain high-frequency acoustic wave features; the high-frequency acoustic wave features refer to acoustic wave features with frequencies exceeding a preset frequency threshold. Based on the high-frequency acoustic wave characteristics, it is determined whether there is a target high-frequency jet acoustic wave in the fourth processed state data. If so, it is determined that there is a leakage anomaly in the thermal pipeline network.

5. The method for monitoring leakage in urban underground pipe networks according to claim 1, characterized in that, The determination that the gas pipeline network has leakage abnormalities includes: Obtain the corresponding fifth-processed state data for the corresponding pipeline segment, and identify the temperature change trend of the corresponding pipeline segment based on the fifth-processed state data; If the temperature change trend shows a downward trend, it is determined that there is an abnormal leakage in the gas pipeline network.

6. A device for monitoring leakage in urban underground pipe networks, characterized in that, The device is used to implement the urban underground pipe network leakage monitoring method as described in any one of claims 1 to 5, comprising: The acquisition module is used to acquire data collected by multiple target devices installed at various urban underground pipe networks located in the target area within a preset time period to obtain initial pipe network status data; the target devices include pressure sensors, flow sensors, water quality sensors, distributed fiber optic sensors, and level gauge sensors; the distributed fiber optic sensors are used to realize distributed acoustic wave sensing, distributed temperature sensing, and distributed strain sensing. The processing module is used to preprocess the initial pipeline network status data to obtain processed status data; The input module is used to input the processed state data into the trained pipeline leakage detection model, so as to perform leakage detection on each of the urban underground pipelines based on the processed state data and obtain leakage detection results.

7. An electronic device, characterized in that, It includes a processor and a memory; wherein, when the processor executes a computer program stored in the memory, it implements the urban underground pipe network leakage monitoring method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the urban underground pipe network leakage monitoring method as described in any one of claims 1 to 5.

Images