Water supply network leakage detection method based on dynamic DMA partition and transient pressure wave analysis

Abstract

The invention discloses a water supply network leakage detection method based on dynamic DMA partitioning and transient pressure wave analysis, and belongs to the technical field of water supply network leakage detection and intelligent management. DMA boundaries are dynamically adjusted through a self-adaptive partitioning algorithm in combination with a GIS and a hydraulic model, a high-density pressure monitoring network is constructed, and the leakage detection and intelligent management of a water supply network are realized. A high-frequency sensor is used for capturing transient pressure wave disturbance caused by tiny leakage, and accurate positioning of leakage points is achieved by fusing wavelet transform and a TDOA positioning model; and aiming at complex pipe network characteristics, a self-adaptive wave velocity correction algorithm is developed to overcome reflection and attenuation interference. A deep learning diagnosis model and a self-adaptive control system are integrated, leakage automatic identification and pipe network operation dynamic adjustment are achieved, and a closed-loop system from detection to management is formed. According to the method, the limitation of low resolution and large positioning error of traditional DMA is broken through, the tiny leakage detection sensitivity and the complex pipe network management efficiency are remarkably improved, and an innovative solution is provided for water resource saving and pipe network intelligent upgrading.

Description

Technical Field

[0001] The present invention belongs to the technical field of leakage detection and intelligent management of water supply pipe networks, and particularly relates to a method for detecting leakage in water supply pipe networks based on dynamic DMA zoning and transient pressure wave analysis. Background Technique

[0002] Foreign technologies and development situations: Abroad, the United States, Europe, and Japan are in the leading position in the technologies of leakage detection and management of water supply pipe networks. In Europe, the DMA (District Metered Area) technology is widely adopted, and leakage is monitored by combining static zoning with flow meters. Its concept originated in the UK in the 1980s, emphasizing regional water balance. However, traditional DMA zoning mostly has fixed boundaries, and the resolution is limited by the density of monitoring points, making it difficult to capture minor leaks (<0.5 L / s), and the leakage detection rate is usually lower than 70%. The United States focuses on transient pressure wave positioning technology. Based on the wave velocity analysis of a single pipe section, acoustic sensors are used to detect leakage points. Representative technologies such as leak acoustic locators are used, but their positioning error usually exceeds 10 meters, and they have poor adaptability to multi-branched complex pipe networks. Japan has developed rapidly in the field of intelligent water services and relies on the Internet of Things and low-power sensors to achieve real-time monitoring of pipe network pressure and flow, such as the intelligent monitoring network of the Tokyo Water Bureau. However, its technology mostly stays at the data collection level and lacks a closed-loop system from detection to control. Generally speaking, foreign technologies perform excellently in specific scenarios, but there is still room for improvement in the detection sensitivity of minor leaks, the positioning accuracy of complex pipe networks, and intelligent management.

[0003] Domestic technologies and development situations: Domestically, the technology of leakage detection in water supply pipe networks started relatively late, but certain progress has been made in recent years with the joint promotion of universities and enterprises. Universities such as Tsinghua University introduced hydraulic models and GIS technologies in the research of DMA zoning technology and tried to optimize the zoning boundaries, but mostly stayed in the theoretical simulation stage, lacking a dynamic adjustment mechanism and field verification. Zhejiang University made breakthroughs in transient pressure wave analysis and developed a leakage positioning method based on acoustic signals. However, due to the limited density of sensors and signal processing algorithms, the positioning accuracy still did not reach within 5 meters. At the enterprise level, for example, Huawei cooperated with water service groups to launch a pressure monitoring system based on NB-IoT, realizing remote transmission of data. However, the density of monitoring points laid out is low (less than 10 points per kilometer), making it difficult to meet the needs of minor leakage detection. In addition, some domestic water service companies borrowed the DMA concept from Europe and carried out leakage management in combination with flow meters. However, due to the complexity of old pipe networks and the diversity of pipe materials, the detection sensitivity and control efficiency are insufficient. Generally speaking, domestic technologies have made progress in theoretical research and preliminary applications, but compared with foreign countries, there is still a large gap in the construction of high-density monitoring networks, precise positioning algorithms, and the integration of intelligent control systems, especially the practicality in complex pipe network environments needs to be improved urgently.

[0004] Existing problems currently:

[0005] Insufficient detection resolution for minute leakage signals: Traditional DMA technology relies on low-density monitoring and static zoning, making it difficult to capture the weak pressure fluctuations caused by minute leakage (<0.5 L / s). The signals are easily masked by hydraulic noise, resulting in a low leakage detection rate.

[0006] Low leakage location accuracy in complex pipe networks: Existing transient pressure wave location technologies are interfered by reflections and attenuations in multi-branch and aging pipe networks. The location error often exceeds 10 meters, making it difficult to meet the requirements of precise repair.

[0007] Insufficient real-time performance in pipe network monitoring and control: Traditional technologies lack a rapid response mechanism after leakage detection, leading to an expansion of leakage and a delay in repair.

[0008] Poor adaptability of the monitoring network to diverse pipe materials: The coexistence of old and new pipe networks, along with the use of various pipe materials such as cast iron and PVC, results in significant differences in the pressure wave propagation characteristics, making it difficult for traditional technologies to be universal.

[0009] Insufficient robustness of intelligent diagnosis models: Existing diagnosis technologies are affected by insufficient on-site data, with limited generalization ability, making it difficult to accurately identify leakage signals in complex scenarios.

[0010] Low system integration of detection and management: Existing technologies mostly remain at the level of single monitoring or location, lacking a closed-loop design from detection to management. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide the present invention in view of the deficiencies in the background technology.

[0012] The present invention adopts the following technical solutions to solve the above technical problems:

[0013] A leakage detection method for water supply pipe networks based on dynamic DMA zoning and transient pressure wave analysis, specifically including the following steps;

[0014] Step 1, dynamically adjust the boundary of the zoned metering area through an adaptive zoning algorithm in combination with a geographic information system and a hydraulic model;

[0015] Step 2, construct a high-density pressure monitoring network;

[0016] Step 3, use high-frequency sensors to capture the transient pressure wave disturbances caused by minute leakage;

[0017] Step 4, fuse wavelet transform and TDOA location model to achieve precise location of the leakage point;

[0018] Step 5: For the characteristics of complex pipe networks, an adaptive wave velocity correction algorithm is adopted to overcome reflection and attenuation interference. An integrated deep learning diagnosis model and an adaptive control system are used to achieve automatic leakage identification and dynamic regulation of the operation of the pipe network, forming a closed-loop system from detection to management.

[0019] As a further preferred solution of the water supply pipe network leakage detection method based on dynamic DMA zoning and transient pressure wave analysis of the present invention, in Step 1, based on the topological structure and hydraulic characteristics of the water supply pipe network, the boundary of the zoned metering area is dynamically adjusted through an adaptive zoning algorithm. The geographic information system and the hydraulic model are used to analyze the flow distribution and pressure fluctuation of the pipe network, and small and medium-sized areas are divided. The zoning boundary is optimized in real time through a pressure stability index.

[0020] Among them, the boundary of the zoned metering area is dynamically adjusted through an adaptive zoning algorithm, and the specific calculation is as follows:

[0021] Among them, B opt is the optimized boundary, ΔP i is the pressure fluctuation of the i-th node, w i is the weight coefficient, V flow is the flow rate change rate, and λ is the penalty coefficient;

[0022] The zoning boundary is optimized in real time through a pressure stability index, and the specific calculation is as follows:

[0023] Among them, S p is the stability index, P i is the node pressure, is the average pressure, and N is the number of nodes;

[0024] The small and medium-sized areas are divided, and the specific calculation is as follows:

[0025] H min ≤H DMA ≤H max ; Among them, H DMA is the number of households in the zone, h j is the number of households in the j-th sub-region, H min and H max are the upper and lower limits of the number of households.

[0026] As a further preferred solution of the water supply pipe network leakage detection method based on dynamic DMA zoning and transient pressure wave analysis of the present invention, in Step 2, a high-density pressure monitoring network is constructed, and the specific calculation is as follows:

[0027] Calculation of the detection point density: 15≤D sensor ≤25; Among them, D sensoris the sensor density, N point is the number of monitoring points, L pipe is the pipeline length;

[0028] Data transmission delay: wherein, Td e l ay is the total delay, T sample is the sampling time, T proc is the processing time, D size is the data volume, R band is the bandwidth;

[0029] Pressure wave resolution: R p <0.05mmH2O; wherein, R p is the resolution, ΔP min is the minimum pressure change, f s is the sampling frequency.

[0030] As a further preferred solution of the water supply pipeline network leakage detection method based on dynamic DMA partitioning and transient pressure wave analysis of the present invention, in step 3, based on the high-density pressure data within the DMA partition, using the propagation characteristics of transient pressure waves including wave speed, attenuation rate, TDOA, separating signal noise through wavelet transform and Hilbert transform, and combining with an adaptive wave speed correction model to locate the leakage point; the specific calculation is as follows:

[0031] wherein, Δt ij is the arrival time difference between the i-th and j-th points, d i ,d j is the distance from the leakage point to the monitoring point, v w is the wave speed;

[0032] wherein, E is the elastic modulus of water, ρ is the density, D is the pipe diameter, E w is the elastic modulus of the pipe wall, t is the wall thickness;

[0033] Positioning error estimation: wherein, ε is the positioning error, Δt error is the time difference measurement error.

[0034] As a further preferred solution of the water supply pipeline network leakage detection method based on dynamic DMA partitioning and transient pressure wave analysis of the present invention, in step 4, through valve fine-tuning and pump station collaborative control, reducing the background noise within the DMA partition, improving the signal-to-noise ratio of the leakage signal, and providing high-quality data for subsequent positioning; based on real-time pressure feedback, dynamically adjusting the hydraulic condition to maintain pressure stability;

[0035] Pressure regulation target: P target =Pbase +ΔP adjust ; where P target is the target pressure, P base is the baseline pressure, and ΔP adjust is the adjustment amount;

[0036] Valve opening calculation: where K v is the valve flow coefficient, Q is the flow rate, and ΔP v is the valve pressure difference;

[0037] Noise reduction rate: where R n is the decibel of noise reduction, and σ before , σ after are the pressure variances before and after adjustment.

[0038] As a further preferred solution of the water supply network leakage detection method based on dynamic DMA partitioning and transient pressure wave analysis of the present invention, in step 5, the convolutional neural network CNN and the long short-term memory network LSTM are fused to extract the spatial and temporal features of the transient pressure wave based on high-density pressure data, realizing the automatic identification of minor leakage (<0.5 L / s) with a detection rate >95%; the robustness of the model is enhanced through transfer learning;

[0039] CNN feature extraction: F out =σ(W*X + b); where F out is the feature map, σ is the activation function, W is the convolutional kernel, X is the input data, and b is the bias;

[0040] LSTM state update: h t = o t ·tanh(C t ), where h t is the hidden state, o t , f t , i t are the output, forget, and input gates, and C t is the cell state;

[0041] Detection rate calculation: where DR is the detection rate, TP is the true positive, and FN is the false negative;

[0042] Based on the leakage diagnosis result, an adaptive control algorithm is designed to dynamically adjust the valve opening and pump station operation parameters within the DMA partition to reduce the expansion of leakage, with a response time <5 minutes;

[0043] Valve adjustment amount: where ΔK v is the change in opening, and Q leakis the leakage volume, P current , P target is the current and target pressure;

[0044] Pumping station power adjustment: Among them, P pump is the power, ρ is the water density, g is the acceleration due to gravity, Q is the flow rate, H is the head, and η is the efficiency;

[0045] Response time calculation: T response = T detect + T calc + T act ; Among them, T response is the total response time, T detect , T calc , T act are the detection, calculation, and execution times.

[0046] Compared with the prior art, the present invention adopts the above technical solutions and has the following technical effects:

[0047] 1. The present invention significantly improves the sensitivity of detecting minute leaks: Through a high-density pressure monitoring network (15 - 25 points per kilometer, sampling rate ≥ 200 Hz, resolution < 0.05 mmHO) and the wavelet transform algorithm, it captures the weak pressure fluctuations (< 0.1 mmHO) caused by minute leaks (< 0.5 L / s) and effectively separates the interference of hydraulic noise; Compared with the low resolution of the traditional DMA technology (leakage detection rate < 70%), the present application improves the detection sensitivity to over 95%, providing high-precision data support for water resource conservation, especially suitable for the early identification of minute leaks in old pipe networks;

[0048] 2. The present invention realizes high-precision positioning of leak points: By using the multi-point collaborative analysis of transient pressure waves and the TDOA positioning model, combined with the adaptive wave speed correction algorithm, the positioning accuracy of leak points in complex pipe networks is 3 - 5 meters, far exceeding the error of over 10 meters of traditional methods; The correction algorithm dynamically adjusts the influence of reflection and attenuation to ensure applicability in pipe networks of multiple materials such as cast iron and PVC, providing a reliable basis for precise repair and reducing the waste of resources caused by blind excavation;

[0049] 3. The present invention enhances the adaptability of monitoring and control of complex pipe networks: Through the integration of the dynamic DMA zoning algorithm and the high-density monitoring network, this technology can adapt to the characteristics of complex pipe networks with a coexistence of old and new and diverse pipe materials; The adaptive zoning adjusts the boundary in real time according to the pipe network topology and hydraulic changes, and the monitoring network overcomes the signal interference caused by branches and aging through multi-point data acquisition, significantly improving the universality and stability of the technology in diverse environments;

[0050] 4. The present invention improves the real-time performance and efficiency of leakage management: By integrating an intelligent diagnosis and adaptive control system, the response time of this technology from leakage detection to valve adjustment is controlled within 5 minutes; The deep learning models (CNN and LSTM) analyze pressure data in real time and trigger control instructions to dynamically adjust the operation parameters of the pipe network, reducing the expansion of leakage. Compared with the lag of traditional manual intervention, this application realizes efficient automated management;

[0051] 5. The present invention provides a closed-loop technology system from detection to management: Beyond single monitoring or positioning functions, through the system integration of diagnosis, positioning, and control, a closed-loop system from leakage identification to leakage control is formed; The visualization platform outputs the coordinates of leakage points, leakage volume estimation, and control suggestions in real time, supporting the decision-making of water service personnel, ensuring that the technical effects extend from the detection end to the management end, and improving the overall efficiency of pipe network management;

[0052] 6. The present invention optimizes data quality and algorithm robustness: The high-density monitoring network provides high-quality pressure data. Combining transfer learning enhances the generalization ability of the intelligent diagnosis model, and can maintain a detection rate of >95% even when field data is insufficient; The adaptive wave speed correction and pressure stability control technology further reduce background noise, providing a solid foundation for signal processing and positioning, and ensuring the stability and reliability of the algorithm under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is the flowchart of the pipe network data collection and basic preparation of the present invention;

[0054] Figure 2 is the flowchart of the dynamic DMA zoning design and optimization of the present invention;

[0055] Figure 3 is the flowchart of the deployment of the high-density pressure monitoring network of the present invention;

[0056] Figure 4 is the flowchart of the transient pressure wave signal collection and positioning of the present invention;

[0057] Figure 5 is the flowchart of the integration of the intelligent diagnosis and adaptive control system of the present invention;

[0058] Figure 6 is the flowchart of the system operation and maintenance of the present invention;

[0059] Figure 7 is the flowchart of the method for detecting leakage in a water supply pipe network based on dynamic DMA zoning and transient pressure wave analysis of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] The technical solutions of the present invention will be further described in detail below with reference to the accompanying drawings:

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention. The present invention will be described in detail below according to the accompanying drawings and preferred embodiments, and the purpose and effect of the present invention will become more apparent. 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.

[0062] The present invention proposes a precise detection and intelligent control technology for water supply network leakage based on dynamic DMA partition optimization, high-density pressure monitoring network construction, and transient pressure wave propagation analysis. By combining an adaptive partition algorithm with GIS and a hydraulic model to dynamically adjust the DMA boundary, a high-density pressure monitoring network is constructed. High-frequency sensors are used to capture transient pressure wave disturbances caused by minute leaks, and the wavelet transform and TDOA positioning model are integrated to achieve precise positioning of the leak point with a positioning accuracy of 3-5 meters. Aiming at the characteristics of complex pipe networks, an adaptive wave speed correction algorithm is developed to overcome reflection and attenuation interference. An integrated deep learning diagnosis model (CNN and LSTM) and an adaptive control system are used to achieve automatic leak identification (detection rate > 95%) and dynamic adjustment of the pipe network operation, forming a closed-loop system from detection to management. The present invention breaks through the limitations of traditional DMA with low resolution and large positioning errors, significantly improves the detection sensitivity of minute leaks and the management efficiency of complex pipe networks, and provides an innovative solution for water resource conservation and the intelligent upgrade of pipe networks.

[0063] Aiming at the insufficient detection resolution of minute leak signals, the present invention constructs a high-density pressure monitoring network, deploys high-frequency sensors and combines with the wavelet transform algorithm to separate leak signals from noise. At the same time, the DMA partition is optimized to stabilize the pressure baseline, improving the detection sensitivity of minute leaks and providing higher-resolution data support for water resource management.

[0064] Aiming at the low leak positioning accuracy in complex pipe networks, the present invention uses multi-point high-density pressure data within the DMA partition, extracts TDOA features and develops an adaptive wave speed correction algorithm. Combining with pipe network material parameters and hydraulic simulation, the positioning accuracy is significantly improved to 3-5 meters, solving the positioning problem in complex pipe network environments.

[0065] Aiming at the lack of real-time performance in pipe network monitoring and control, the present invention uses a deep learning model to identify leaks in real time and dynamically adjust the operation of valves and pumping stations, with the response time controlled within 5 minutes, improving the real-time performance and efficiency of pipe network management.

[0066] The present invention aims at the poor adaptability of the monitoring network to diverse pipe materials. Through an adaptive zoning algorithm and a wave velocity correction model, the monitoring and positioning parameters are dynamically adjusted to ensure the applicability of the technology in pipe networks of multiple materials.

[0067] Aiming at the insufficient robustness of the intelligent diagnosis model, the present invention introduces transfer learning, uses simulated data to enhance the diversity of training samples, and fuses CNN and LSTM to improve the recognition robustness of the model to minor leaks, with a detection rate of over 95%.

[0068] Aiming at the low system integration of detection and management, through the deep integration of intelligent diagnosis and adaptive control, a closed-loop management system is constructed to provide the coordinates of leakage points, leakage volume estimation, and control suggestions, realizing the systematic and intelligent upgrade of pipe network management.

[0069] The specific embodiments are as follows:

[0070] I. Dynamic DMA Zoning Optimization Technology

[0071] Based on the topological structure and hydraulic characteristics of the water supply pipe network, the boundaries of DMA (District Metered Area) are dynamically adjusted through an adaptive zoning algorithm, breaking through the low-resolution bottleneck of traditional static zoning. Using GIS (Geographic Information System) and hydraulic models to analyze the flow distribution and pressure fluctuations in the pipe network, small (<500 households) and medium-sized (<2000 households) areas are divided to reduce hydraulic interference and improve the ability to capture minor leakage signals. The zoning boundaries are optimized in real time through a pressure stability index to ensure the coverage of the monitoring network and data quality.

[0072] Among them, the boundaries of the district metered area are dynamically adjusted through the adaptive zoning algorithm, and the specific calculation is as follows:

[0073] Among them, among them, B opt is the optimized boundary, ΔP i is the pressure fluctuation at the i-th node, w i is the weight coefficient, Vf low is the flow rate change rate, and λ is the penalty coefficient;

[0074] The zoning boundaries are optimized in real time through a pressure stability index, and the specific calculation is as follows:

[0075] Among them, among them, S p is the stability index, P i is the node pressure, is the average pressure, and N is the number of nodes;

[0076] The small and medium-sized areas are divided, and the specific calculation is as follows:

[0077] H min≤H DMA ≤H max ; where, H DMA is the number of households in a sub - area, h j is the number of households in the j - th sub - area, and H min and H max are the upper and lower limits of the number of households.

[0078] Implementation plan

[0079] (1) Data collection and pre - processing:

[0080] ① Use the GIS system to collect pipe network topology data, including pipe length, diameter, material, and connection relationship.

[0081] ② Deploy flow meters and initial pressure sensors to collect baseline data on the flow and pressure distribution in the pipe network within 24 hours.

[0082] ③ Clean the data, remove outliers (such as jumps caused by sensor failures), and smooth the noise using the moving average method.

[0083] (2) Initial sub - area division:

[0084] ① Based on the pipe network topology, use the K - means clustering algorithm to initially divide the DMA areas and set the initial scales for small - sized (<500 households) and medium - sized (<2000 households).

[0085] ② Calculate the flow balance of each sub - area to ensure that the difference between the inflow and outflow is less than 5%, which serves as the initial boundary constraint.

[0086] (3) Hydraulic model construction and simulation:

[0087] ① Use hydraulic simulation software to establish a pipe network hydraulic model and input pipe parameters (such as roughness coefficient, elastic modulus) and actual flow data.

[0088] ② Simulate the pressure distribution under different working conditions to identify areas with large hydraulic fluctuations, which are the key points for boundary adjustment.

[0089] (4) Dynamic boundary optimization:

[0090] ① According to formula (1), combined with pressure fluctuations and flow change rates, iteratively optimize the DMA boundary. The step size for each iteration is set to 50 meters, and adjust the valve position to change the water flow path.

[0091] ② Use formula (2) to calculate the pressure stability index. If Sp>0.05, continue to adjust the boundary until the stability meets the standard.

[0092] ③ Constrain the sub - area scale through formula (3) to ensure that the number of households in the sub - area is within the specified range.

[0093] (5) Real-time monitoring and feedback:

[0094] ① Deploy control valves and sensors at the optimized DMA boundary to form a closed-loop monitoring network.

[0095] ② Update the hydraulic data every 1 hour and repeat the optimization process to ensure that the partitions adapt to the changes in the pipe network operation.

[0096] II. Construction technology of high-density pressure monitoring network

[0097] By deploying high-frequency and high-resolution pressure sensors (sampling rate ≥ 200Hz, resolution < 0.05mmHO), a refined grid with 15 - 25 monitoring points per kilometer is constructed, and low-power data transmission is achieved using 5G NB-IoT technology. The monitoring network captures transient pressure wave disturbances and provides high spatio-temporal resolution data, laying a foundation for leakage detection and location.

[0098] Calculation of detection point density: 15 ≤ D sensor ≤ 25; where D sensor is the sensor density, N point is the number of monitoring points, and L pipe is the pipeline length;

[0099] Data transmission delay: where Td e l ay is the total delay, T sample is the sampling time, T proc is the processing time, D size is the data volume, and R band is the bandwidth;

[0100] Pressure wave resolution: R p <0.05mmH2O; where R p is the resolution, ΔP min is the minimum pressure change, and f s is the sampling frequency.

[0101] Implementation plan:

[0102] (1) Sensor selection and calibration:

[0103] ① Select high-frequency pressure sensors (such as silicon-based MEMS sensors) with a sampling rate ≥ 200Hz, a resolution < 0.05mmHO, and a range of 0 - 10mH O.

[0104] ② Conduct static and dynamic calibrations in the laboratory, use a standard pressure source to verify the sensor accuracy, and control the error within ±0.02%.

[0105] (2) Monitoring point layout plan:

[0106] ① According to formula (1), combined with the pipeline length and complexity (such as the number of branches), calculate the number of monitoring points per kilometer, and preferentially arrange them near pipeline corners and valves.

[0107] ② Use GIS tools to draw the layout diagram to ensure the uniform distribution of monitoring points and avoid blind spots.

[0108] (3) Network hardware installation:

[0109] ① Install sensors and NB-IoT modules at each monitoring point. The sensors are fixed to the outer wall of the pipeline through waterproof connectors, and the modules are built-in batteries to support at least 2 years of operation.

[0110] ② Configure 5G base station coverage to ensure that the signal strength > 90%, and the test data upload success rate reaches 99%.

[0111] (4) Data collection and transmission:

[0112] ① The sensor continuously collects pressure data at a frequency of 200Hz, and generates a data packet (including timestamp and pressure value) every 5 seconds.

[0113] ② Optimize the transmission parameters through formula (2), compress the data volume to <1KB / packet, and control the delay <1 second.

[0114] ③ Upload the data to the cloud server and synchronously store it in the local SD card as a backup.

[0115] (5) Network operation and maintenance:

[0116] ① Check the sensor status daily and automatically alarm the nodes with battery power <10% or data interruption > 5 minutes.

[0117] ② Calibrate the sensor once every quarter and update the layout location to adapt to the pipeline network transformation.

[0118] III. Precise positioning technology for transient pressure waves

[0119] Based on the high-density pressure data within the DMA partition, using the propagation characteristics of transient pressure waves (wave speed, attenuation rate, TDOA), separate signal noise through wavelet transform and Hilbert transform, and combine with an adaptive wave speed correction model to locate the leakage point with an accuracy of 3-5 meters. The specific calculation is as follows:

[0120] Among them, Δt ij is the arrival time difference between the i-th and j-th points, d i , d j is the distance from the leakage point to the monitoring point, v w is the wave speed;

[0121] Among them, E is the elastic modulus of water, ρ is the density, D is the pipe diameter, and E w is the elastic modulus of the pipe wall, and t is the wall thickness;

[0122] Positioning error estimation: Among them, ε is the positioning error, and Δt error is the time difference measurement error.

[0123] IV. Partition boundary pressure stability control technology

[0124] Through the coordinated control of valve fine-tuning and pumping stations, the background noise in the DMA partition is reduced, and the signal-to-noise ratio of leakage signals is improved to provide high-quality data for subsequent positioning. Based on real-time pressure feedback, the hydraulic conditions are dynamically adjusted to maintain pressure stability.

[0125] Pressure regulation target: P target = P base + ΔP adjust ; among them, P target is the target pressure, P base is the baseline pressure, and ΔP adjust is the adjustment amount;

[0126] Valve opening calculation: Among them, K v is the valve flow coefficient, Q is the flow rate, and ΔP v is the valve pressure difference;

[0127] Noise reduction rate: Among them, R n is the decibel of noise reduction, and σ before , σ after are the pressure variances before and after adjustment.

[0128] V. Intelligent leakage diagnosis technology

[0129] Fusing convolutional neural network (CNN) and long short-term memory network (LSTM), extracting the spatial and temporal features of transient pressure waves based on high-density pressure data, realizing the automatic identification of minor leakage (<0.5 L / s), and the detection rate >95%. Enhancing the robustness of the model through transfer learning.

[0130] CNN feature extraction: F out = σ(W * X + b); among them, F out is the feature map, σ is the activation function, W is the convolutional kernel, X is the input data, and b is the bias;

[0131] LSTM state update: h t = o t · tanh(C t ), Among them, ht is in the hidden state, o t , f t , i t are the output, forget, and input gates, C t is the cell state;

[0132] Detection rate calculation: where DR is the detection rate, TP is the true positive, and FN is the false negative;

[0133] VI. Adaptive control system integration technology

[0134] Based on the leakage diagnosis results, design an adaptive control algorithm to dynamically adjust the valve opening and pump station operation parameters within the DMA partition, reduce the expansion of leakage, and the response time < 5 minutes. Integrate a visualization platform to provide real-time decision support.

[0135] Based on the leakage diagnosis results, design an adaptive control algorithm to dynamically adjust the valve opening and pump station operation parameters within the DMA partition, reduce the expansion of leakage, and the response time < 5 minutes;

[0136] Valve adjustment amount: where ΔK v is the change in opening, Q leak is the leakage volume, P current , P target are the current and target pressures;

[0137] Pump station power adjustment: where P pump is the power, ρ is the water density, g is the acceleration due to gravity, Q is the flow rate, H is the head, and η is the efficiency;

[0138] Response time calculation: T response = T detect + T calc + T act ; where T response is the total response time, T detect , T calc , T act are the detection, calculation, and execution times.

[0139] As Figure 7 shown, pipe network data collection and basic preparation: Obtain the topological structure, hydraulic parameters, and operation data of the pipe network to lay the foundation for subsequent zoning and monitoring.

[0140] Data collection: Through cooperation with the water service department, obtain the GIS data of the water supply pipe network, including the pipe length (accurate to meters), diameter (mm), material (such as cast iron, PVC), and valve positions. Collect historical operation data, such as the flow rate (m 3 / h) and pressure (mHO) recording.

[0141] On-site inspection: The technical team measured the geographical coordinates and pipeline status of key nodes (such as branch points and pumping stations) on-site, and recorded the sediment and corrosion conditions of old pipeline sections. A portable flowmeter and pressure gauge were used to verify the accuracy of GIS data, with the error controlled within ±5%.

[0142] Data preprocessing: Import the collected data into the database, use Python scripts to clean outliers (such as flow rate mutations >50%), and perform interpolation processing on missing data (linear interpolation method). Generate a pipeline network topology map and mark the candidate positions of monitoring points.

[0143] Dynamic DMA zoning design and optimization: Divide small (<500 households) and medium (<2000 households) DMA areas, and dynamically optimize the boundaries to improve the leakage detection resolution.

[0144] Initial zoning: Use the K-means clustering algorithm to initially divide the DMA areas according to the number of households and pipeline length, and generate the initial boundaries. Ensure that the water inflow and outflow in each zone are balanced, with a difference <5%.

[0145] Hydraulic model construction: Input the pipeline network parameters (roughness coefficient, elastic modulus) into the software, simulate the 24-hour hydraulic conditions, and output the pressure and flow distribution.

[0146] Boundary optimization: According to the pressure stability index (Sp<0.05), adjust the valve position to change the water flow path, with a step size of 50 meters each time. Iteratively calculate the optimization function until the zoning boundary is stable.

[0147] Verification: Implement zoning in the test area (such as a 100-km pipeline network), monitor the pressure fluctuations within 1 week, and confirm that the stability meets the standard.

[0148] Deployment of a high-density pressure monitoring network: Construct a network with 15 - 25 monitoring points per kilometer to capture transient pressure wave signals.

[0149] Sensor selection: Select high-frequency pressure sensors (sampling rate 200Hz, resolution <0.05mmHO), equipped with an NB-IoT communication module. Calibrate in the laboratory to ensure an accuracy of ±0.02%.

[0150] Layout planning: Calculate the monitoring point density (15 - 25 points / km) according to the pipeline network length and complexity, and prioritize deployment at corners and branches. Use GIS tools to generate the layout map to avoid signal blind spots.

[0151] Hardware installation: Technicians drill holes on the outer wall of the pipeline to fix the sensors, connect waterproof joints, and install the NB-IoT module (with a built-in battery, lifespan >2 years). Test the 5G signal coverage to ensure an upload success rate >99%.

[0152] Data Test: Start the sensor, continuously collect data for 24 hours, verify that the transmission delay < 1 second, and store it in the cloud server.

[0153] Transient Pressure Wave Signal Acquisition and Location: Use high-density data to locate the leakage point, with an accuracy of 3 - 5 meters.

[0154] Signal Acquisition: The sensor records the pressure wave at a frequency of 200Hz, lasting for 10 seconds each time, generating time series data.

[0155] Signal Processing: Apply wavelet transform (5-layer decomposition) to remove water hammer noise and extract high-frequency leakage signals.

[0156] Use Hilbert transform to calculate the arrival time, with an accuracy of ±0.001 seconds.

[0157] Wave Velocity Calibration: Input pipe material parameters (such as the elastic modulus of PVC is 1.4GPa), calculate the theoretical wave velocity, and adjust it in combination with the measured attenuation rate.

[0158] TDOA Location: Use data from more than 3 monitoring points to solve the coordinates of the leakage point, and the error assessment < 5 meters.

[0159] Verification: Simulate leakage (0.3L / s) in the test pipe network, compare the location results with the actual position, and optimize the algorithm.

[0160] Integrate intelligent diagnosis and adaptive control system to achieve automatic leakage identification (detection rate > 95%) and dynamic regulation of the pipe network (response time < 5 minutes).

[0161] Data Preparation: Collect 100,000 pressure data, label leakage samples, and simulate and generate diverse scenarios.

[0162] Model Training: Build a CNN (3-layer convolution) + LSTM (128 units) model, pre-train it on a public dataset, and fine-tune it on field data. Use the Adam optimizer, train for 50 epochs, and the loss < 0.01.

[0163] Control System Deployment: Install electric valves (accuracy ±0.1%) at the DMA boundary and connect to the pump station frequency converter. Develop a control algorithm to adjust the opening and power according to the leakage location.

[0164] Real-time Operation: Input pressure data every minute, output the diagnostic result. If the probability > 0.9, trigger a control instruction. Monitor the change in leakage volume after adjustment, with the goal of reducing it to < 0.1L / s. Visualization Platform: Build a Web interface to display the leakage coordinates, leakage volume estimation, and suggestions for water supply personnel to make decisions.

[0165] System Operation and Maintenance: Ensure the long-term stable operation of the system and continuously optimize the performance.

[0166] Pilot run: Operate in the pilot area for 1 month and record the leakage detection and control effects.

[0167] Performance evaluation: Calculate the detection rate (target > 95%), positioning accuracy (3 - 5 meters), and response time (< 5 minutes).

[0168] Maintenance plan: Check the sensor status daily and automatically alarm abnormal nodes. Calibrate the equipment quarterly and update the layout and algorithm parameters.

[0169] Data feedback: Generate an operation report monthly, analyze the reduction in water leakage, and optimize the system configuration.

[0170] Expansion application: Gradually promote to the entire regional pipe network according to the pilot results.

[0171] Example: Smart water service in a city in Xinjiang

[0172] Background: A certain city is located in the Yili River Valley of Xinjiang. The water supply pipe network covers about 150 kilometers in the urban area, with a total of about 100,000 households. Due to historical reasons, the pipe network includes both old cast iron pipes over 20 years old and newly built PVC pipes in recent years. Leakage problems occur frequently, with an annual leakage rate as high as over 20%, resulting in serious waste of water resources. In addition, the water demand in this city surges in summer, and the low temperature in winter easily causes pipe bursts. Traditional DMA zoning and manual inspections are difficult to meet the requirements of efficient management.

[0173] Implementation process

[0174] 1. Pipe network data collection and basic preparation

[0175] Cooperate with the water service bureau of a certain city to obtain GIS data, including pipe length (150 km), diameter (50 - 500 mm), material distribution (60% cast iron, 40% PVC), and valve positions.

[0176] The technical team conducted on-site inspections of 50 key nodes (such as branch points and old areas), verified the flow data using portable flow meters, and found that the recording errors at 5 locations were > 10%, and re-measured and corrected them.

[0177] Import the data into the database, clean the outliers (such as flow rate mutations > 50%), generate a pipe network topology map, and mark the candidate positions of monitoring points.

[0178] 2. Dynamic DMA zoning design and optimization

[0179] Use the K-means clustering algorithm to divide the pipe network into 20 DMA areas, including 10 small areas (< 500 households) and 10 medium-sized areas (< 2000 households).

[0180] Input the pipe parameters (cast iron roughness coefficient 0.015, PVC roughness coefficient 0.009) and simulate the summer peak flow (5000 m3 / h), identify the pressure fluctuation area.

[0181] According to the pressure stability index (Sp < 0.05), adjust the positions of 10 boundary valves with a step size of 50 meters, and iteratively optimize until the stability meets the standard.

[0182] Operate for 1 week in the pilot area (10 km pipe network), and the pressure fluctuation decreases from 0.2 mHO to 0.03 mHO.

[0183] 3. Deployment of high-density pressure monitoring network

[0184] Select MEMS pressure sensors (sampling rate 200 Hz, resolution 0.04 mmHO), deploy 2,500 monitoring points on a 150 km pipe network with a density of 16.7 points / km, and give priority to installing them on old cast iron pipe sections.

[0185] Using the GIS planning layout map, the installation team fixes the sensors on the outer wall of the pipeline, connects the NB-IoT module, and tests that the 5G signal coverage rate reaches 98%.

[0186] Continuously collect 24-hour data and upload it to the cloud server, with a delay < 0.8 seconds and a success rate of 99.5%.

[0187] 4. Acquisition and positioning of transient pressure wave signals

[0188] Simulate a 0.3 L / s leakage in the pilot area (10 km pipe network), and the sensor records 10 seconds of pressure wave data.

[0189] Apply wavelet transform (5-layer decomposition) to remove water hammer noise, use Hilbert transform to extract the arrival time, and calculate TDOA (accuracy ±0.001 seconds).

[0190] Input the pipe material parameters (cast iron elastic modulus 130 GPa, PVC 1.4 GPa), correct the wave speed (cast iron 1100 m / s, PVC 400 m / s), and solve the coordinates of the leakage point.

[0191] Verify that the error between the result and the actual position is 3.2 meters, and it is reduced to 2.8 meters after adjusting the monitoring points.

[0192] 5. Integration of intelligent diagnosis and adaptive control system

[0193] Collect 50,000 pressure data in the pilot area, train the CNN + LSTM model, fine-tune it after pre-training, and the detection rate reaches 96%.

[0194] Install electric valves (accuracy ±0.1%) at the boundaries of 5 DMAs, connect the pump station frequency converter, and deploy the control system.

[0195] Real-time diagnosis detected 1 leakage of 0.4 L / s, triggering an adjustment instruction. The valve opening decreased by 5%, and the pumping station frequency decreased by 10%. The leakage rate dropped to 0.1 L / s within 5 minutes.

[0196] The Web platform displays the leakage coordinates (latitude and longitude), the estimated leakage rate (0.4 L / s), and suggestions (close the upstream valve).

[0197] 6. System Operation and Maintenance

[0198] During 1 month of operation in the pilot area, 8 leakages were detected, and the positioning error was less than 5 meters for all. The leakage rate decreased from 20% to 12%.

[0199] Check the sensor status daily. An automatic alarm is triggered when the power of 2 sensors is less than 10%. Calibrate 50 sensors every quarter.

[0200] The monthly report shows a reduction in leakage of 5000 m 3 , and optimize the valve adjustment frequency.

[0201] Result: The system operates stably, and the leakage control effect is remarkable.

[0202] Implementation effect: Leakage detection: The detection rate of minor leakages (<0.5 L / s) reaches 96%, which is 26% higher than the traditional method (70%). Positioning accuracy: The positioning error of the leakage point is 3 - 5 meters, which is better than the traditional method by more than 10 meters. Leakage reduction: The leakage rate in the pilot area decreased by 8%, and the annual water saving is about 60,000 m 3 . Management efficiency: The response time is shortened from the hour level to 5 minutes, and the manual inspection is reduced by 80%.

[0203] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions described in the foregoing examples, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principles of the invention should be included within the protection scope of the invention. All technical features in this embodiment can be freely combined according to actual needs.

[0204] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A leakage detection method for water supply networks based on dynamic DMA partitioning and transient pressure wave analysis, characterized in that: Specifically, it includes the following steps; Step 1: Dynamically adjust the boundary of the district metering area by combining the adaptive zoning algorithm with the geographic information system and the hydraulic model; Step 2: Construct a high-density pressure monitoring network; Step 3: Use high-frequency sensors to capture the transient pressure wave disturbances caused by minor leaks; Step 4: Integrate wavelet transform and TDOA positioning model to achieve accurate positioning of the leak point; Step 5: Aiming at the characteristics of complex pipe networks, adopt an adaptive wave speed correction algorithm to overcome reflection and attenuation interference, integrate a deep learning diagnosis model and an adaptive control system, realize automatic leak identification and dynamic adjustment of pipe network operation, and form a closed-loop system from detection to management.

2. The leakage detection method for a water supply pipe network based on dynamic DMA partitioning and transient pressure wave analysis according to claim 1, wherein: In Step 1, based on the topological structure and hydraulic characteristics of the water supply pipe network, dynamically adjust the boundary of the district metering area by the adaptive zoning algorithm, analyze the flow distribution and pressure fluctuations of the pipe network using the geographic information system and the hydraulic model, divide small and medium-sized areas, and the zoning boundary is optimized in real time through the pressure stability index; Among them, the boundary of the district metering area is dynamically adjusted by the adaptive zoning algorithm, and the specific calculation is as follows: Among them, among them, B opt is the optimized boundary, ΔP i is the pressure fluctuation of the i-th node, w i is the weight coefficient, V flow is the flow rate change rate, and λ is the penalty coefficient; The zoning boundary is optimized in real time through the pressure stability index, and the specific calculation is as follows: Among them, S p is the stability index, P i is the node pressure, is the average pressure, and N is the number of nodes; Divide small and medium-sized areas, and the specific calculation is as follows: Among them, H DMA is the number of households in the partition, h j is the number of households in the j-th sub-region, H min and H max are the upper and lower limits of the number of households.

3. The method for detecting leakage in a water supply network based on dynamic DMA partitioning and transient pressure wave analysis according to claim 1, wherein: In Step 2, construct a high-density pressure monitoring network, and the specific calculation is as follows: Calculation of detection point density: where D sensor is the sensor density, N point is the number of monitoring points, and L pipe is the pipeline length; Data transmission delay: Among them, Td e l ay is the total delay, T sample is the sampling time, T proc is the processing time, D size is the data volume, R band is the bandwidth; Pressure wave resolution: R p <0.05 mmH2O; where R p is the resolution, ΔP min is the minimum pressure change, and f s is the sampling frequency.

4. The leakage detection method for water supply pipe networks based on dynamic DMA partitioning and transient pressure wave analysis according to claim 1, characterized in that: In Step 3, based on the high-density pressure data within the DMA zone, utilize the propagation characteristics of transient pressure waves including wave speed, attenuation rate, and TDOA, separate signal noise through wavelet transform and Hilbert transform, and combine with the adaptive wave speed correction model to locate the leak point; the specific calculation is as follows: where Δt ij is the arrival time difference between the i-th and j-th points, d i , d j is the distance from the leakage point to the monitoring point, and v w is the wave velocity; Among them, E is the elastic modulus of water, ρ is the density, D is the pipe diameter, and E w is the elastic modulus of the pipe wall, and t is the wall thickness; Positioning error estimation: where ε is the positioning error and Δt error is the time difference measurement error.

5. The leakage detection method for water supply pipe networks based on dynamic DMA partitioning and transient pressure wave analysis according to claim 1, characterized in that: In Step 4, through the coordinated control of valve fine-tuning and pump stations, reduce the background noise within the DMA zone, improve the signal-to-noise ratio of leak signals, and provide high-quality data for subsequent positioning; based on real-time pressure feedback, dynamically adjust the hydraulic conditions to maintain pressure stability; Pressure regulation target: P target = P base + ΔP adjust ; Among them, P target is the target pressure, P base is the baseline pressure, and ΔP adjust is the adjustment amount; Valve opening calculation: where K v is the valve flow coefficient, Q is the flow rate, and ΔP v is the valve pressure difference; Noise reduction rate: where R n is the decibel of noise reduction, and σ before , σ after are the pressure variances before and after adjustment.

6. The leakage detection method for water supply pipe networks based on dynamic DMA partitioning and transient pressure wave analysis according to claim 1, wherein: In Step 5, integrate the convolutional neural network CNN and the long short-term memory network LSTM, extract the spatial and temporal characteristics of transient pressure waves based on high-density pressure data, realize automatic identification of minor leaks (<0.5 L / s), and the detection rate > 95%; enhance the robustness of the model through transfer learning; CNN feature extraction: F out = σ(W * X + b); where F out is the feature map, σ is the activation function, W is the convolutional kernel, X is the input data, and b is the bias; LSTM state update: h t = o t ·tanh(C t ), where h t is the hidden state, o t , f t , i t are the output, forget, and input gates, and C t is the cell state; Detection rate calculation: where DR is the detection rate, TP is the true positive, and FN is the false negative; Based on the leak diagnosis results, design an adaptive control algorithm to dynamically adjust the valve opening and pump station operation parameters within the DMA zone, reduce the expansion of leakage, and the response time < 5 minutes; Valve adjustment amount: where, ΔK v is the opening change, Q leak is the leakage, P current , P target are the current and target pressures; Pumping station power adjustment: Among them, P pump is power, ρ is water density, g is gravitational acceleration, Q is flow rate, H is head, and η is efficiency; Response time calculation: T response = T detect + T calc + T act ; where, T response is the total response time, and T detect , T calc , T act are the detection, calculation, and execution times.

Images