Gis-based dma intelligent partitioning method for water supply network

Abstract

The application discloses a GIS-based DMA intelligent partitioning method for a water supply network, relates to the field of data processing, and can solve the technical problem that the current DMA intelligent partitioning method for the water supply network cannot accurately distinguish between sudden water taking abnormalities and leakage abnormalities, resulting in false partitioning boundary misjudgment or frequent fluctuations of a position boundary, and making it difficult to realize accurate positioning of a leakage area, and comprises the following steps: obtaining water flow state monitoring data; determining a water flow abnormal data set according to the water flow state monitoring data; and determining a water supply network partitioning preliminary result according to the water flow abnormal data set; wherein the water supply network partitioning preliminary result is used to represent the preliminary division range of each region in the water supply network; and the DMA partitioning of the water supply network is determined according to the water supply network partitioning preliminary result and physical equipment information of the water supply network. The application is used for DMA intelligent partitioning and leakage monitoring of the water supply network.

Description

Technical Field

[0001] This invention relates to the field of data processing, and specifically to a DMA intelligent zoning method for water supply networks based on GIS. Background Technology

[0002] As a crucial component of urban infrastructure, the stable operation of water supply networks is closely related to water supply security and resource utilization. District Metering Area (DMA) zoning is a key technical means for water supply network management. Typically, based on physical metering equipment within the network, the water supply network is divided into multiple clearly defined and controllable local network units. This zoning enables the monitoring of water supply facility leakage, and precise zoning can quickly locate leakage areas, thereby improving water supply security and water resource utilization efficiency, and providing support for the efficient operation and maintenance of urban water supply systems.

[0003] Because directly adjusting the physical partitions of the water supply network is too costly, and the adjustment process may cause problems such as water flow stagnation that affect the normal operation of the network, existing methods usually do not prioritize physical adjustments when performing DMA intelligent partitioning of the water supply network. Instead, they use techniques such as reverse simulation calculations to divide nodes that are sensitive to pressure changes into the same leakage partition. This allows for the location of abnormal areas such as pipe bursts, and the network anomalies can be quickly identified through virtual partitioning. This reduces adjustment costs while ensuring the basic stability of the network operation.

[0004] The physical water flow environment of actual water supply networks is quite complex. In actual operation, sudden and unrecorded water use events often occur, such as a factory taking a large amount of water at night. The pressure fluctuations caused by such sudden water use events are similar to leakage signals. However, existing DMA intelligent partitioning methods based on virtual partitions usually cannot accurately distinguish between these two types of signals, which leads to misjudgment of virtual partition boundaries or frequent fluctuations in the boundaries of a certain location. That is, a certain location is hydraulically ambiguous and affected by multiple water sources. When there is a sudden water use or leakage, the virtual partition boundary determined at that location will change repeatedly, affecting the accuracy of leakage location and the stability of partitioning. Summary of the Invention

[0005] To address the technical problem of current DMA (Discretionary Area Modulation) intelligent zoning methods for water supply networks, which fail to accurately distinguish between sudden water intake anomalies and leakage anomalies, leading to misjudgments of virtual zone boundaries or frequent fluctuations in boundaries at certain locations, thus hindering precise location of leakage areas, this invention aims to provide a GIS-based DMA intelligent zoning method for water supply networks. The specific technical solution adopted is as follows:

[0006] In a first aspect, the present invention provides a GIS-based DMA intelligent zoning method for water supply networks, comprising: acquiring water flow status monitoring data; wherein the water flow status monitoring data is used to characterize the dynamic characteristics of water flow at different monitoring points within the water supply network; determining a water flow anomaly dataset based on the water flow status monitoring data; wherein the water flow anomaly dataset includes anomaly time data and anomaly time period data for each monitoring point, the anomaly time data being used to characterize the time when an anomaly event occurs at the monitoring point, and the anomaly time period data being used to characterize the time range within which consecutive anomalies occur at the monitoring point; determining preliminary zoning results for the water supply network based on the water flow anomaly dataset; wherein the preliminary zoning results for the water supply network are used to characterize the preliminary division range of each area within the water supply network; and determining the DMA zoning of the water supply network based on the preliminary zoning results and the physical equipment information of the water supply network.

[0007] In one possible implementation, the abnormal water flow dataset is determined based on the water flow status monitoring data. Specifically, this includes: for each monitoring point, arranging the water flow status monitoring data in time series according to different indicators to obtain time series curves for each indicator; segmenting the time series curves for each indicator using adaptive principal component analysis (APCA); and determining the abnormal water flow dataset based on the segmented time series curves for each indicator.

[0008] In one possible implementation, the abnormal water flow dataset is determined based on the segmented time-series curves of each indicator. Specifically, this includes: acquiring historical comparison data, which includes water flow status monitoring data of the water supply network within historical time periods; for each moment in the time-series curve of each indicator, determining a moment anomaly probability parameter based on the corresponding historical comparison data and the duration of the time period; where the moment anomaly probability parameter characterizes the likelihood of an anomaly occurring in the indicator at the current moment; and determining multiple abnormal moments based on the moment anomaly probability parameter for each moment; where an abnormal moment is defined as one where the anomaly probability parameter is greater than the moment anomaly probability parameter. The threshold time is determined; based on multiple abnormal times, abnormal flow growth segments are identified; among them, the difference between the water flow status monitoring data corresponding to each time in the abnormal flow growth segment and the average value of the same period in the historical comparison data is greater than 0, and the monitoring value at the right end of the time period at each time is greater than the monitoring value at the left end; based on the time-anomaly probability parameter of the abnormal time corresponding to the abnormal flow growth segment and the number of abnormal indicators in the abnormal flow growth segment, the time-period anomaly probability parameter of the abnormal flow growth segment is determined; based on the time-period anomaly probability parameter of the abnormal flow growth segment, multiple abnormal time periods are identified; the abnormal times and abnormal time periods corresponding to each monitoring point are integrated to determine the water flow anomaly dataset.

[0009] In one possible implementation, the preliminary results of water supply network zoning are determined based on the water flow anomaly dataset. Specifically, this includes: determining the anomaly event differentiation results based on the water flow anomaly dataset; wherein the anomaly event differentiation results are used to indicate the anomaly type of anomaly event occurring within the water supply network, and the anomaly types include sudden water intake anomalies and leakage anomalies; adjusting the weights of each monitoring point in the GNN based on the anomaly event differentiation results to obtain the preliminary results of water supply network zoning output by the GNN; wherein the GNN is used to determine the preliminary results of water supply network zoning based on the topology of the water supply network and the adjusted weights of each monitoring point, and the preliminary results of water supply network zoning are used to characterize the preliminary division range of each area in the water supply network.

[0010] In one possible implementation, the abnormal event differentiation result is determined based on the water flow anomaly dataset, specifically including: identifying multiple monitoring points for water flow anomalies based on the water flow anomaly dataset; calculating the water flow anomaly similarity index between every two adjacent monitoring points; wherein, two adjacent monitoring points are two monitoring points directly connected in the topology, and the water flow anomaly similarity index is used to characterize the similarity of the abnormal events of the two adjacent monitoring points; determining local anomaly regions based on the water flow anomaly similarity index between every two adjacent monitoring points; wherein, local anomaly regions are used to characterize the set of monitoring points whose water flow anomalies are caused by the same abnormal event; and determining the anomaly type of the local anomaly region to obtain the abnormal event differentiation result.

[0011] In one possible implementation, the anomaly type of a local anomaly region is determined to obtain an anomaly event differentiation result. Specifically, this includes: determining the historical recurrence index for each monitoring point in the local anomaly region; wherein the historical recurrence index is used to characterize the similarity between the current anomaly event and historical anomalies at each monitoring point; determining the sudden water withdrawal probability index of the local anomaly region based on the historical recurrence index of each monitoring point, the anomaly recovery speed of the local anomaly region, and the duration of the anomaly event; wherein the anomaly recovery speed is used to indicate the speed at which the local anomaly region recovers to a normal state after an anomaly event occurs, and the sudden water withdrawal probability index is used to characterize the probability that the anomaly event occurring in the local anomaly region is a sudden water withdrawal anomaly; and determining the anomaly event differentiation result based on the sudden water withdrawal probability index of the local anomaly region.

[0012] In one possible implementation, the weights of each monitoring point in the GNN are adjusted based on the abnormal event differentiation results to obtain the preliminary results of the water supply network zoning output by the GNN. Specifically, this includes: determining the weight adjustment priority index for each monitoring point based on the historical leakage anomaly occurrence frequency, the number of monitoring points included in the corresponding local anomaly area, and the anomaly type of the corresponding abnormal event; for each monitoring point in the GNN, adjusting the weight of each monitoring point in the GNN according to the weight adjustment priority index, the initial weight, and the preset weight adjustment rules; wherein, the preset weight adjustment rules include: if the abnormal event differentiation results indicate that the monitoring point corresponds to a leakage anomaly, then the weight of the monitoring point is increased and maintained for a preset duration; if the abnormal event differentiation results indicate that the monitoring point corresponds to a sudden water intake anomaly, then the weight of the monitoring point is temporarily increased before the sudden water intake anomaly ends; the adjusted weights of each monitoring point are input into the GNN, and the GNN outputs the preliminary results of the water supply network zoning.

[0013] In one possible implementation, the DMA partitioning of the water supply network is determined based on the preliminary results of the water supply network partitioning and the physical equipment information of the water supply network. Specifically, this includes: mapping the preliminary results of the water supply network partitioning to the physical equipment information of the water supply network to determine multiple hydraulically non-independent regions; wherein, a hydraulically non-independent region is a region without a key dividing valve and where water flow cannot be controlled independently; for each hydraulically non-independent region, based on the topology of the water supply network, determining the region with the strongest hydraulic correlation adjacent to the hydraulically non-independent region; merging each hydraulically non-independent region into its corresponding adjacent region with the strongest hydraulic correlation to form multiple target regions; and determining the valve or pipe diameter change point closest to the boundary of the multiple target regions as the region boundary of the DMA, thus obtaining the DMA partitioning of the water supply network.

[0014] In one possible implementation, after acquiring the water flow status monitoring data, the method further includes: aligning the timestamps of the detection data corresponding to different indicators in the water flow status monitoring data according to a preset monitoring period; preprocessing the water flow status monitoring data; wherein, the preprocessing includes filling in missing values ​​and removing invalid data.

[0015] In one possible implementation, the water flow status monitoring data includes pressure data, flow rate data, flow velocity data, and water quality data at each monitoring point in the water supply network.

[0016] This invention offers the following advantages: First, by installing multiple types of sensors at key locations in the pipeline network, and then using timestamp alignment, preprocessing, and historical data to eliminate seasonal interference, high-quality water flow status monitoring data is obtained, laying a reliable foundation for subsequent analysis. Next, adaptive principal component analysis is used to segment the time-series curves of the indicators, and combined with anomaly probability parameters, anomaly times and periods are accurately selected to form a structured water flow anomaly dataset. Subsequently, based on the anomaly dataset, local anomaly regions are merged, and leakage and sudden water intake anomalies are distinguished. The weights of GNN nodes are adjusted to output preliminary results of water supply network zoning that align with the anomaly distribution. Finally, the preliminary results are mapped to the information of the pipeline network's physical equipment, identifying and merging hydraulically non-independent regions. The strongest associated merging objects are determined based on the topology, and after boundary calibration, a DMA zoning system with independent metering, independent water control, and identifiable leakage characteristics is formed. This comprehensively solves the problems of fragmented data, anomaly confusion, and the disconnect between theory and physical reality in traditional zoning, effectively improving the accuracy of water supply network leakage monitoring and the efficiency of refined operation and maintenance, achieving an intelligent closed loop from data acquisition to zoning management. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the architecture of a GIS-based water supply network DMA intelligent zoning system provided in one embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the architecture of a water flow anomaly analysis module provided in one embodiment of the present invention;

[0020] Figure 3 This is a schematic diagram of the architecture of a water supply network partitioning module provided in one embodiment of the present invention;

[0021] Figure 4 This is a flowchart illustrating a GIS-based DMA intelligent zoning method for water supply networks, provided in one embodiment of the present invention.

[0022] Figure 5 This is a flowchart illustrating another DMA-based intelligent zoning method for water supply networks provided in an embodiment of the present invention.

[0023] Figure 6 This is a flowchart illustrating another DMA-based intelligent zoning method for water supply networks provided in an embodiment of the present invention.

[0024] Figure 7 This is a flowchart illustrating another DMA-based intelligent zoning method for water supply networks provided in an embodiment of the present invention. Detailed Implementation

[0025] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a GIS-based DMA intelligent zoning method for water supply networks proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0027] The technical terms involved in this invention are described below:

[0028] 1. Geographic Information System

[0029] Geographic Information System (GIS) is a technology system that integrates geospatial data acquisition, storage, analysis, and visualization. In this solution, GIS plays a core role in data support and model building. It first imports the physical information of the water supply network (including pipe diameter, material, and equipment location), then constructs a three-dimensional topology model of the water supply network based on this information, clarifying the connection relationships between each network node. Simultaneously, GIS is used to record monitoring point locations, store water flow status monitoring data, and, in the subsequent zoning stage, map the zoning results to the actual geographic information of the network, assisting in optimizing zoning boundaries. Finally, it visualizes the DMA zoning results, providing staff with an intuitive reference for abnormal network areas.

[0030] 2. Independent metering area

[0031] A District Metering Area (DMA) is a clearly defined, independently manageable local network unit within a water supply network, divided using physical metering equipment or virtual partitioning. Its core objective is to monitor and locate leaks in the water supply network. By monitoring flow, pressure, and other data within a single DMA unit, the existence and approximate extent of leaks in that area can be quickly determined. In this solution, DMA partitioning must combine anomaly analysis results with information on the network's physical equipment to ensure that partition boundaries match actual valves and pipe diameter change points, forming independently manageable leak monitoring units. This addresses the problems of ambiguous partition boundaries and inaccurate leak location associated with traditional methods.

[0032] 3. Graph Neural Networks

[0033] Graph Neural Networks (GNNs) are deep learning models that excel at processing topological data, analyzing and computing data based on the "node-edge" relationships. In this scheme, the topology of the water supply network (water-using units as nodes and pipes as edges) is highly compatible with the processing characteristics of GNNs. GNNs first receive the initial weights of each network node, and then dynamically adjust the node weights based on the results of abnormal situations (sudden water intake anomalies / leakage anomalies)—the weights of nodes with leakage anomalies are increased and maintained, while the weights of nodes with sudden water intake anomalies are temporarily increased and then restored. Finally, based on the adjusted node weights and the network topology relationship, GNNs calculate and output the preliminary results of the water supply network partitioning, providing a foundation for subsequent boundary optimization.

[0034] 4. Adaptive Principal Component Analysis

[0035] Adaptive Principal Component Analysis (APCA) is a data preprocessing and segmentation technique that adaptively divides continuous time-series curves into multiple independently analyzable time periods based on the temporal characteristics of the data. In this scheme, APCA is used to process the time-series curves of indicators (pressure, flow rate, velocity, water quality, etc.) at various monitoring points. By segmenting the continuous water flow data into multiple time periods, it facilitates the subsequent calculation of the anomaly probability parameters at each moment. Simultaneously, APCA's adaptive nature can adapt to the dynamic changes in the water flow status of the water supply network, avoiding misjudgments of anomaly moments caused by fixed segmentation methods, and providing technical support for accurately screening abnormal water flow states.

[0036] The following description, in conjunction with the accompanying drawings, details a specific scheme for a GIS-based DMA intelligent zoning method for water supply networks provided by this invention.

[0037] Please see Figure 1This document illustrates a schematic diagram of the architecture of a GIS-based intelligent zoning system for water supply networks using the DMA (Digital Directional Analysis) mechanism, according to an embodiment of the present invention. The intelligent zoning system 10 includes: a data acquisition module 11, a water flow anomaly analysis module 12, a water supply network zoning module 13, a DMA zoning optimization module 14, and a GIS platform module 15. Through the coordinated operation of the data acquisition module 11, the water flow anomaly analysis module 12, the water supply network zoning module 13, the DMA zoning optimization module 14, and the GIS platform module 15, intelligent processing of the entire process from monitoring the water flow status of the water supply network to determining the DMA zoning is achieved. The modules are described below in sequence:

[0038] (1) Data acquisition module 11.

[0039] The data acquisition module 11 is the basic data input unit of the intelligent distribution system 10. It is responsible for acquiring the original water flow status data of the water supply network and preprocessing it to provide an accurate and standardized data foundation for subsequent water flow anomaly analysis. The preprocessed water flow status monitoring data it outputs is directly transmitted to the water flow anomaly analysis module 12.

[0040] Optionally, the data acquisition module 11 is used to acquire water flow status monitoring data; wherein, the water flow status monitoring data is used to characterize the dynamic characteristics of water flow at different monitoring points within the water supply network.

[0041] For example, water flow status monitoring data includes pressure data, flow rate data, flow velocity data, and water quality data at each monitoring point in the water supply network. This data is collected by installing pressure sensors, flow sensors, flow velocity sensors, and water quality sensors at each monitoring location. The monitoring points may be located at the inlet of the water supply network, valves, pumps, areas with historical leaks, and important water-using units.

[0042] Specifically, the data acquisition module 11 installs pressure sensors, flow sensors, velocity sensors, and water quality sensors at each monitoring point, and collects pressure data, flow data, velocity data, and water quality data at each monitoring point in a 24-hour monitoring cycle; at the same time, it synchronously records the timestamps of the data collected by each sensor to ensure the integrity of the data sequence.

[0043] After that, the data acquisition module 11 first aligns the timestamps of different sensor data to ensure time sequence consistency; then it completes the preprocessing by filling in missing values ​​and removing invalid data, and finally generates preprocessed water flow status monitoring data that can be directly used for water flow anomaly analysis, and transmits it to the water flow anomaly analysis module 12.

[0044] (2) Water flow anomaly analysis module 12.

[0045] The water flow anomaly analysis module 12, based on the water flow status monitoring data output by the data acquisition module 11, completes the screening of water flow anomaly status and the differentiation of anomaly types. Its output water flow anomaly dataset and anomaly event differentiation results provide the basis for weight adjustment for the water supply network partitioning module 13.

[0046] Optionally, the water flow anomaly analysis module 12 is used to determine the water flow anomaly dataset based on the water flow status monitoring data.

[0047] The water flow anomaly dataset includes abnormal time data and abnormal time period data for each monitoring point. Abnormal time data is used to characterize the time when an abnormal event occurs at the monitoring point, and abnormal time period data is used to characterize the time range within which an abnormal event occurs continuously at the monitoring point.

[0048] For example, such as Figure 2 As shown, the water flow anomaly analysis module 12 may include three sub-modules: a time-series curve processing sub-module 121, an anomaly screening sub-module 122, and an anomaly type region sub-module 123.

[0049] The time-series curve processing submodule 121 is used to arrange the pre-processed water flow state monitoring data of each monitoring point according to different indicators such as pressure, flow rate, flow velocity, and water quality, and generate time-series curves for each indicator. Then, APCA is used to segment each time-series curve, splitting the continuous data into time periods that can be analyzed independently, laying the foundation for subsequent screening of abnormal moments.

[0050] The anomaly screening submodule 122 is used to calculate the time anomaly probability parameter for each moment in the time series curve of each indicator using monitoring data from the past three months as historical comparison data. Moments with a normalized time anomaly probability parameter greater than 0.5 are marked as abnormal moments. Further screening is performed to select consecutive abnormal moments with monitoring values ​​greater than the historical average and the right endpoint of the time period being greater than the left endpoint, forming an abnormal flow growth segment. The remaining consecutive abnormal moments form an abnormal flow decrease segment. Then, monitoring points with a normalized time period anomaly probability parameter greater than 0.5 are selected by using the time period anomaly probability parameter, and their abnormal moments and abnormal time periods are integrated to generate a water flow anomaly dataset.

[0051] The anomaly type sub-module 123 is used to extract abnormal monitoring points from the water flow anomaly dataset, calculate the water flow anomaly similarity index between monitoring points directly connected in the topology (determined based on the difference in anomaly degree and the difference in anomaly time duration), merge monitoring points with a similarity index greater than 0.5 after normalization into local anomaly regions; then analyze the anomaly recovery speed, anomaly impact range, and anomaly historical recurrence degree of the local anomaly regions, distinguish between sudden water intake anomalies and leakage anomalies, generate anomaly event differentiation results, and transmit them to the water supply network partition module 13.

[0052] (3) Water supply network zoning module 13.

[0053] The water supply network partitioning module 13, based on the abnormal event differentiation results output by the water flow anomaly analysis module 12, adjusts the network node weights of the GNN and outputs preliminary partitioning results. This "preliminary water supply network partitioning result" is the core input of the DMA partitioning optimization module 14.

[0054] Optionally, the water supply network zoning module 13 is used to determine the preliminary results of the water supply network zoning based on the water flow anomaly dataset. The preliminary results of the water supply network zoning are used to characterize the preliminary division range of each area in the water supply network.

[0055] For example, such as Figure 3 As shown, the water supply network partitioning module 13 may include two sub-modules: a node weight adjustment sub-module 131 and a GNN partitioning calculation sub-module 132.

[0056] The node weight adjustment submodule 131 is used to set the initial weight of each pipeline node (corresponding to a monitoring point) in the GNN to 1, and the upper limit of weight increase is limited to 5; if the abnormal event differentiation result indicates that a certain node corresponds to a leakage abnormality, the weight of the node is increased and maintained (until the leakage is repaired); if it indicates a corresponding sudden water intake abnormality, the weight of the node is temporarily increased (until it returns to 1 after the abnormality ends), and the adjusted pipeline node weights are generated.

[0057] The GNN partition calculation submodule 132 is used to receive the adjusted weights output by the node weight adjustment submodule 131, combine them with the water supply network topology, calculate the partitions through the topology data processing capability of GNN, output the preliminary results of the water supply network partitions, clarify the preliminary division range of each area, and pass them to the DMA partition optimization module 14.

[0058] (4) DMA partition optimization module 14.

[0059] The DMA partitioning optimization module 14, in conjunction with the preliminary partitioning results output by the water supply network partitioning module 13 and the physical equipment information of the water supply network, optimizes the partitioning boundaries and determines the final DMA partitioning. Its output "water supply network DMA partitioning results" will be sent to the GIS platform module 15 for visualization.

[0060] Optionally, the DMA partitioning optimization module 14 is used to determine the DMA partitioning of the water supply network based on the preliminary results of the water supply network partitioning and the physical equipment information of the water supply network.

[0061] Specifically, the DMA partitioning optimization module 14 first maps the preliminary results of the water supply network partitioning with the physical equipment information of the water supply network (including pipe diameter, material, length, laying year, and geographical location and elevation of equipment such as valves and pumps) to identify "hydraulic non-independent areas" (i.e. areas without key dividing valves and where water flow cannot be controlled independently).

[0062] Following this, the DMA partitioning optimization module analyzes the hydraulic correlation strength between hydraulically non-independent regions and adjacent regions based on the water supply network topology (determined by the frequency of water flow interaction and pipeline connection relationship), and merges the hydraulically non-independent regions into the adjacent regions with the strongest hydraulic correlation to form the target region.

[0063] Finally, the DMA partitioning optimization module uses the valve or pipe section diameter change point (the location where the pipe diameter changes) closest to the boundary of each target area as a benchmark to adjust the boundary of each target area, ensuring that the boundary matches the actual physical equipment, and finally generates the "Water Supply Network DMA Partitioning Result", which is then transmitted to the GIS platform module 15.

[0064] (5) GIS platform module 15.

[0065] GIS platform module 15 is used to provide basic geographic and equipment data for data acquisition module 11, water flow anomaly analysis module 12, water supply network zoning module 13, and DMA zoning optimization module 14, and to visualize the final zoning results.

[0066] Specifically, the GIS platform module 15 first receives information on the physical equipment of the water supply network, constructs a three-dimensional topology model of the water supply network in the GIS platform, and clarifies the geographical location and equipment relationship of each network node. This model provides the basis for the data acquisition module 11 to determine the location of monitoring points and for the water supply network partitioning module 13 to call the topology structure.

[0067] In addition, the GIS platform module 15 is used to store the raw and preprocessed data of the data acquisition module 11, the water flow anomaly dataset and anomaly event differentiation results of the water flow anomaly analysis module 12, the weight data and preliminary partitioning results of the water supply network partitioning module 13, and the DMA partitioning results of the DMA partitioning optimization module 14, so as to realize the centralized management and retrieval of the entire process data.

[0068] It should be noted that the GIS platform module 15 is also used to load the DMA partitioning results of the water supply network output by the DMA partitioning optimization module 14, and to present the correspondence between the partition location and the actual network in the GIS map; if the partition contains leakage abnormal nodes, the partition will flash red to prompt key investigation; if it only contains sudden water intake abnormal nodes, the partition will remain yellow (until the abnormality ends and the normal display is restored), providing staff with an intuitive reference for abnormal areas.

[0069] The above provides an introduction to the GIS-based water supply network DMA intelligent zoning system 10 and its included modules.

[0070] For example, such as Figure 4The diagram illustrates a flowchart of a GIS-based DMA intelligent zoning method for water supply networks according to an embodiment of the present invention, comprising the following steps:

[0071] S401. Acquire water flow status monitoring data.

[0072] Among them, water flow status monitoring data is used to characterize the dynamic characteristics of water flow at different monitoring points within the water supply network.

[0073] Understandably, for the smooth execution of this step, monitoring points are first set up at key locations in the water supply network based on its topology and operation and maintenance needs. These monitoring points specifically include the network inlet, valves, pumps, historical leakage areas, and important water-using units. The inlet and pumps are core nodes for water flow input, reflecting the overall water supply load of the network; valves are key nodes for water flow control, capturing changes in state caused by local water flow interruption or regulation; historical leakage areas are high-risk nodes requiring focused monitoring of anomaly recurrence; and important water-using units (such as large residential areas and industrial users) are nodes with concentrated water demand, reflecting the impact of user-end water consumption fluctuations on the network.

[0074] Specifically, pressure sensors, flow sensors, velocity sensors, and water quality sensors are installed at each monitoring point. The pressure sensors collect water pressure data within the pipeline network at the monitoring point, reflecting the hydraulic stability of the network. The flow sensors collect water volume data passing through the monitoring point per unit time, reflecting changes in water demand. The velocity sensors collect water velocity data passing through the monitoring point, helping to determine whether the water flow within the pipeline is within the normal flow range. The water quality sensors collect water quality data such as turbidity and residual chlorine, indirectly reflecting whether there is damage to the pipeline network; for example, abnormal water quality may be accompanied by the infiltration of external impurities.

[0075] In one possible implementation, step S401 can be specifically executed by the data acquisition module 11 in the intelligent partitioning system 10 described above.

[0076] Optionally, the data acquisition module 11 uses a preset monitoring cycle of 24 hours to collect pressure, flow, velocity, and water quality data at each monitoring point in real time through the aforementioned sensors, while simultaneously recording the timestamp (accurate to the second) corresponding to each data point. Since different types of sensors have different response speeds, to avoid errors in subsequent data correlation analysis due to time-series discrepancies, the timestamps of the water flow status monitoring data collected by each sensor at the same monitoring point and during the same time period must be aligned. This ensures that the pressure, flow, velocity, and water quality data at the same monitoring point at the same moment can be matched, forming a time-consistent original water flow status monitoring dataset.

[0077] Furthermore, after forming a time-consistent original water flow state monitoring dataset, the data acquisition module 11 preprocesses the original water flow state monitoring dataset to finally generate preprocessed water flow state monitoring data that can be directly used for water flow anomaly analysis.

[0078] Preprocessing includes operations such as missing value completion and invalid data removal. For example, missing value completion can be achieved through interpolation of data from adjacent time points; invalid data removal can be achieved by eliminating values ​​that are outside the normal range due to sensor malfunctions.

[0079] In this way, the water flow status monitoring data acquired by the data acquisition module 11 comprehensively characterizes the dynamic characteristics of water flow at different monitoring points within the water supply network through multi-dimensional indicators: From a spatial perspective, it can reflect the differences in water flow status at different monitoring points (such as upstream and downstream of the pipeline, user end and water supply end). For example, the flow data at the inlet can reflect the overall water supply scale, while the flow data at the user end can reflect local water demand; From a temporal perspective, it can reflect the real-time changes and fluctuation patterns of water flow status at each monitoring point, such as the regular fluctuations of rising flow and falling pressure at the user end during the morning peak hours, or the step increase in flow caused by sudden water intake; From the perspective of abnormal correlation, it can provide data support for distinguishing between normal fluctuations and abnormal events through the coordinated changes of multiple indicators (such as a sudden increase in flow accompanied by a sudden drop in pressure).

[0080] In one possible implementation, to eliminate the interference of seasonal fluctuations in water use on subsequent anomaly judgment, the data acquisition module 11 acquires historical comparison data simultaneously while acquiring real-time water flow status monitoring data. This historical comparison data consists of water flow status monitoring data from the same monitoring point and the same monitoring period over the past three months.

[0081] S402. Based on the water flow status monitoring data, determine the water flow anomaly dataset.

[0082] The water flow anomaly dataset includes abnormal time data and abnormal time period data for each monitoring point. Abnormal time data is used to characterize the time when an abnormal event occurs at the monitoring point, and abnormal time period data is used to characterize the time range within which an abnormal event occurs continuously at the monitoring point.

[0083] In one possible implementation, step S402 can be specifically executed by the water flow anomaly analysis module 12 in the intelligent zoning system 10 described above. Specifically, for each monitoring point, the water flow anomaly analysis module 12 first arranges the water flow state monitoring data according to different indicators in time series to obtain the time series curves of each indicator; then, the water flow anomaly analysis module 12 segments the time series curves of each indicator according to adaptive principal component analysis (APCA); finally, the water flow anomaly analysis module 12 determines the water flow anomaly dataset based on the segmented time series curves of each indicator. It should be noted that the specific process of the water flow anomaly analysis module 12 executing the aforementioned steps is detailed in steps S501-S503 below, and will not be repeated here.

[0084] Therefore, the water flow anomaly analysis module 12 can transmit the determined water flow anomaly dataset to the water supply network partitioning module 13, providing core data support for calculating the water flow anomaly similarity index between monitoring points and distinguishing between sudden water intake anomalies and leakage anomalies, ensuring the accuracy of subsequent GNN node weight adjustment and water supply network partitioning.

[0085] S403. Based on the abnormal water flow dataset, determine the preliminary results of water supply network zoning.

[0086] Among them, the preliminary results of water supply network zoning are used to characterize the preliminary division range of each area in the water supply network.

[0087] In one possible implementation, step S403 can be specifically executed by the water supply network partitioning module 13 in the intelligent partitioning system 10 described above, including the following two steps:

[0088] (1) The water supply network partitioning module 13 determines the abnormal event differentiation results based on the water flow abnormal dataset.

[0089] The abnormal event differentiation results are used to indicate the type of abnormal event that occurred in the water supply network. The abnormal event types include sudden water intake abnormalities and leakage abnormalities.

[0090] Specifically, the water supply network zoning module 13 first filters out "water flow anomaly monitoring points" with anomaly records from the water flow anomaly dataset. Then, based on the water supply network topology, it calculates the water flow anomaly similarity index between adjacent anomaly monitoring points. This index is used to determine whether the anomalies of two monitoring points are caused by the same event. The higher the similarity index, the smaller the time difference between the anomalies and the closer the degree of anomaly, and the more likely they are to be classified into the same local anomaly area, that is, a set of monitoring points affected by a single anomaly event. Finally, by analyzing the recovery speed and stability of the impact range of the local anomaly area, it distinguishes between sudden water intake anomalies and leakage anomalies, forming an anomaly event differentiation result.

[0091] It should be noted that the purpose of the water supply network zoning module 13 in determining the abnormal event differentiation results is to clarify the type of abnormal event (sudden water intake abnormality or leakage abnormality) from the abnormal water flow data set (including the abnormal time and abnormal period of each monitoring point) and avoid zoning deviation due to confusion of abnormal types.

[0092] (2) The water supply network partitioning module 13 adjusts the weight of each monitoring point in the GNN based on the abnormal event differentiation results to obtain the preliminary results of the water supply network partitioning output by the GNN.

[0093] Among them, GNN is used to determine the preliminary results of water supply network zoning based on the topology of the water supply network and the adjusted weight of each monitoring point. The preliminary results of water supply network zoning are used to characterize the preliminary division range of each area in the water supply network.

[0094] It is understandable that GNN is a computational model adapted to the topology of water supply networks. It can perform partitioned calculations based on the relationship between network nodes and pipe connections. Each network node corresponds to a monitoring point, and the node weight directly determines the attention given to that location when partitioning.

[0095] When adjusting the weight of each monitoring point in the GNN, the water supply network partitioning module 13 follows the following adjustment rules: if the abnormal event classification result is "leakage abnormality", the weight of the corresponding node is increased and maintained (leakage needs to be controlled in the long term); if it is "sudden water intake abnormality", the weight is temporarily increased (the initial value of 1 is restored after the abnormality ends), and the upper limit of weight growth is limited (to avoid the weight of a single node being too high).

[0096] After this, the water supply network partitioning module 13 inputs the adjusted weights into the GNN. The GNN calculates the partitions based on the water supply network topology and outputs "preliminary results of water supply network partitioning". These results clarify the preliminary division range of each area and are directly used to optimize the DMA partition boundaries in conjunction with physical equipment information.

[0097] Therefore, the water supply network partitioning module 13 adjusts the weights of GNN nodes to prioritize high-priority anomalies (i.e., leakage anomalies) in the partitioning calculation, ensuring that the preliminary partitioning matches the actual anomaly situation in the network.

[0098] It should be noted that the specific process of the water supply network partition module 13 in performing the aforementioned two steps is detailed in steps S601-S602 below, and will not be repeated here.

[0099] S404. Based on the preliminary results of the water supply network zoning and the physical equipment information of the water supply network, determine the DMA zoning of the water supply network.

[0100] In one possible implementation, step S403 may be specifically performed by the DMA partition optimization module 14 in the intelligent partitioning system 10 described above.

[0101] Specifically, this step, based on the preliminary results of the water supply network zoning and combined with the physical equipment information of the water supply network, completes the DMA zoning in three steps: physical mapping, area optimization, and boundary determination. First, the preliminary zoning results are mapped with the physical equipment information to identify areas that cannot be independently managed. Then, these areas are merged into adjacent areas that are hydraulically closely related. Finally, the zoning boundaries are determined based on actual valves and pipe diameter change points to form the final DMA zoning. It should be noted that the specific process of the DMA zoning optimization module 14 in executing the aforementioned steps is detailed in steps S701-S704 below, and will not be repeated here.

[0102] Therefore, by integrating physical device information, the DMA partition optimization module 14 can correct unreasonable boundaries of the initial partition, ensuring that the final DMA partition has clear boundaries, can be independently measured, and can be individually controlled, thus achieving the core objective of accurate leakage monitoring and location.

[0103] Based on the above technical solution, this embodiment of the invention first sets up monitoring points at key locations in the water supply network, and uses multiple types of sensors to collect data at preset monitoring cycles. After timestamp alignment, preprocessing, and historical data assistance to eliminate interference, high-quality water flow status monitoring data is obtained. Then, the data is formed into time series curves according to indicators, and after segmentation by adaptive principal component analysis, abnormal times and periods are screened to construct a water flow anomaly dataset. Subsequently, local abnormal areas are merged by combining the network topology, and sudden water intake and leakage anomalies are distinguished. The weights of the calculation model adapted to the topology are adjusted according to the anomaly type to output preliminary results of water supply network partitioning. Finally, combined with the network physical equipment information, hydraulically non-independent areas are merged and the actual equipment is used to define the boundaries, ultimately forming a DMA partition with clear boundaries, independent metering, and independent water flow control. This effectively realizes accurate monitoring and location of water supply network leakage, and improves the operation and maintenance efficiency of the water supply network and the utilization rate of water resources.

[0104] For example, in combination Figure 4 ,like Figure 5 The diagram illustrates a flowchart of another GIS-based DMA intelligent zoning method for water supply networks, provided by an embodiment of the present invention. In this method, the abnormal water flow dataset is determined based on water flow status monitoring data, specifically including the following steps:

[0105] S501. For each monitoring point, the water flow status monitoring data are arranged in time series according to different indicators to obtain the time series curves of each indicator.

[0106] In one possible implementation, step S501 can be specifically executed by the time-series curve processing submodule 121 in the water flow anomaly analysis module 12 described above. Specifically, the time-series curve processing submodule 121 classifies the water flow status monitoring data by indicator type for each monitoring point in the water supply network; that is, pressure data is categorized separately, flow data is categorized separately, and so on, for the same monitoring point. The time-series curve processing submodule 121 uses time as the horizontal axis and indicator monitoring value as the vertical axis to arrange the monitoring data of the same indicator in chronological order according to the collection timestamp, generating the indicator time-series curve for each indicator corresponding to that monitoring point.

[0107] It should be noted that the purpose of the time series curve processing submodule 121 in generating time series curves is to transform discrete monitoring data into continuous time series, intuitively present the fluctuation pattern of each indicator over time, and provide a visual and analyzable basic data form for subsequent identification of abnormal fluctuations. In addition, the curves must completely retain the original timestamps of each monitoring value to ensure the time series consistency of subsequent segmentation and anomaly calculation.

[0108] S502. Segment the time series curves of each indicator according to the adaptive principal component analysis (APCA).

[0109] In one possible implementation, step S502 can be specifically executed by the time-series curve processing submodule 121 in the water flow anomaly analysis module 12 described above.

[0110] Specifically, the time series curve processing submodule 121 uses APCA to segment the time series curves of the above indicators. The core function of APCA is to adaptively divide the time series data of the indicators into multiple independently analyzable time periods based on the dynamic fluctuation characteristics of the time series data, rather than using fixed-duration segmentation.

[0111] Furthermore, the time-series curve processing submodule 121 extracts the principal component features (such as fluctuation amplitude and frequency of change) of the time-series curve, identifies the boundaries between stable fluctuation segments and abrupt fluctuation segments in the curve, and divides the continuous time-series curve into several segments of varying durations. For example, a flow-time time-series curve can be divided into a low-load stable segment in the early morning, a high-fluctuation segment during the morning peak, and a medium-load stable segment during the day. The advantage of this segmentation method is that it adapts to the dynamic changes in the water flow status of the water supply network, avoids the forced separation of stable and abrupt segments by fixed segmentation, and ensures that subsequent anomaly analysis can be carried out based on a relatively uniform time period, providing the key parameter of time period duration for accurately calculating the anomaly probability of a single moment.

[0112] S503. Based on the time series curves of each indicator after segmentation, determine the abnormal water flow dataset.

[0113] In one possible implementation, step S503 can be specifically executed by the anomaly filtering submodule 122 in the water flow anomaly analysis module 12 described above. For example, the anomaly filtering submodule 122 implements S503 through the following steps:

[0114] (1) Obtain historical comparison data.

[0115] The historical comparison data includes monitoring data on the water flow status of the water supply network during historical periods.

[0116] In this step, to eliminate the interference of seasonal fluctuations in water usage on anomaly detection, historical comparison data must first be obtained. This historical comparison data consists of water flow status monitoring data from the water supply network over the past three months, and must meet three conditions: same as the current monitoring point, same as the current analysis indicator (e.g., both are flow indicators), and same as the current monitoring cycle (both are 24-hour cycles). Specifically, the anomaly screening submodule 122 can obtain the historical comparison data collected by the data acquisition module 11 through a data connection with the data acquisition module 11.

[0117] Understandably, the selection of the past three months as the historical period is due to the significant seasonality of water supply network usage (e.g., higher water consumption in summer than in winter). This period not only covers the complete short-term water usage patterns but also avoids interference from factors such as pipeline aging and user changes in long-term data, ensuring that historical comparison data accurately reflects the fluctuation range of indicators under normal water usage conditions. Simultaneously, the historical comparison data needs to be preprocessed to ensure consistency in data quality and temporal completeness with the currently segmented indicator time-series curves, providing a reliable benchmark for subsequent anomaly detection.

[0118] (2) For each moment in the time series curve of each indicator, the abnormal probability parameter of the moment is determined based on the historical comparison data corresponding to each moment and the duration of the time period in which each moment is located.

[0119] Among them, the time-time anomaly probability parameter is used to characterize the degree of possibility that the corresponding indicator will be abnormal at the current time.

[0120] In this step, for each moment in the time series curves of each segmented indicator, an anomaly probability parameter is calculated by combining the aforementioned historical comparison data with the duration of the segment in which that moment falls. This parameter quantifies the degree to which the corresponding indicator deviates from its normal state at the current moment. For example, the moment anomaly probability parameter can be calculated using the following formula:

[0121]

[0122] in, Let be the time anomaly probability parameter at time i. Let i be the water flow rate measured at time i. This represents the average water flow rate measured at the same time point within a historical period. This indicates the duration of the time period in which time i is located.

[0123] It should be pointed out that when and The greater the deviation, The shorter the segment, the more intense the water flow fluctuation at that moment. The larger the value, the higher the probability that the indicator is abnormal at that moment.

[0124] (3) Determine multiple abnormal times based on the time-abnormal probability parameter at each time.

[0125] Among them, the abnormal moment is the moment when the abnormal probability parameter is greater than the time abnormal threshold;

[0126] Optionally, after calculating the time anomaly probability parameter for each time moment, the anomaly filtering submodule 122 will use the time anomaly probability parameter for each time moment... The time anomaly threshold is set by mapping the time anomaly probability parameter to the (0, 1) interval using a normalization method. Then, the anomaly filtering submodule 122 uses the time anomaly probability parameter... Moments exceeding the time anomaly threshold are marked as abnormal moments. Conversely, the anomaly filtering submodule 122 sets the time anomaly probability parameter... Moments that are less than or equal to the abnormal time threshold are marked as normal times.

[0127] For example, the above-mentioned time anomaly threshold can be set to 0.5. It should be noted that in practical applications, the time anomaly threshold can be determined according to requirements, and this embodiment of the invention does not impose specific limitations.

[0128] (4) Determine the abnormal flow growth segment based on multiple abnormal times.

[0129] Among them, the difference between the water flow status monitoring data corresponding to each moment in the abnormal flow growth segment and the average value of the same period in the historical comparison data is greater than 0, and the monitoring value at the right end of the time period at each moment is greater than the monitoring value at the left end.

[0130] In this step, the anomaly screening submodule 122 performs further classification and screening on the abnormal moments marked in the above steps to distinguish different types of abnormal fluctuations.

[0131] Specifically, the process for determining abnormal flow rate increases is as follows: if an abnormal time simultaneously meets two conditions—that the current monitored value is greater than the historical average for the same period and the monitored value at the right endpoint of the segment in which the current time falls is greater than the monitored value at the left endpoint—then the abnormal time is marked as an abnormal flow rate increase time. These two conditions are based on the characteristic that sudden water intake or leakage typically causes flow rate fluctuations, thus ensuring that the selected time matches the flow rate change characteristics caused by the abnormal event.

[0132] Similarly, the determination of abnormal flow reduction moments can also be performed. The process is as follows: if an abnormal moment meets two conditions simultaneously—the current monitored value is less than the historical average for the same period and the monitored value at the right end of the segment in which the current moment is located is less than the monitored value at the left end—then the abnormal moment is marked as an abnormal flow reduction moment. Such moments may be caused by sudden drops in pipeline pressure, valve closure, etc., and need to be analyzed separately from moments of increase to avoid confusion of abnormal types.

[0133] Furthermore, the anomaly screening submodule 122 merges consecutive abnormal traffic growth moments into a continuous time interval, denoted as an abnormal traffic growth segment; similarly, it merges consecutive abnormal traffic decrease moments into an abnormal traffic decrease segment. Together, they constitute the candidate abnormal time period for this indicator. When performing the aforementioned merging, the definition of "continuous" must be met, that is, the time interval between adjacent abnormal moments does not exceed a certain duration (e.g., 30 minutes).

[0134] (5) Determine the time period probability parameter of the abnormal flow growth segment based on the time abnormal probability parameter corresponding to the abnormal moment and the number of abnormal indicators of the abnormal flow growth segment.

[0135] For example, in this step, the anomaly filtering submodule 122 determines the time-period anomaly probability parameter of the abnormal traffic growth segment based on the time-period anomaly probability parameter corresponding to the abnormal moment of the abnormal traffic growth segment and the number of abnormal indicators of the abnormal traffic growth segment. Specifically, this can be achieved through the following formula:

[0136]

[0137] in, Let j be the time-period anomaly probability parameter for the abnormal traffic growth segment. This refers to the number of indicators that exhibited abnormal fluctuations after the occurrence of this abnormal increase in water flow (i.e., indicators with abnormal periods; for example, if water pressure data, water volume data, and flow velocity data are abnormal after this abnormal increase in water flow, then the number of indicators with abnormal fluctuations is 3). The value of is also 3). The abnormal probability parameter for indicators exhibiting abnormal fluctuations at corresponding times within the abnormal period. The mean, This represents the total number of indicators.

[0138] It is understandable that in the above formula, The larger the value, the better. The larger the value, the higher the degree of coordinated anomaly among multiple indicators, and the higher the probability that the candidate anomaly period was caused by a real abnormal event (rather than the accidental fluctuation of a single indicator); for example, when only the traffic indicator is abnormal... =1, If the value is low, the occurrence of this abnormal traffic increase is more likely due to sensor error; while when multiple indicators show abnormalities simultaneously... A higher value is more likely to indicate leakage or sudden water intake.

[0139] (6) Based on the time period abnormality probability parameters of the abnormal flow growth period, determine multiple abnormal time periods.

[0140] In this step, the anomaly screening submodule 122 determines whether each abnormal traffic growth segment is caused by a real abnormal event by using its corresponding time period anomaly probability parameter.

[0141] Specifically, the anomaly filtering submodule 122 will use the time period anomaly probability parameter Normalized to the (0, 1) interval, and an anomaly threshold for the time period is set. Then, the anomaly filtering submodule 122 applies the normalized time-period anomaly probability parameter. Abnormal traffic growth segments exceeding the time-period abnormal threshold are identified as abnormal time periods; and, the abnormality screening submodule 122 normalizes the time-period abnormality probability parameters. Abnormal traffic growth periods that are less than or equal to the time period abnormality threshold are identified as accidental fluctuation periods and removed to avoid invalid data interfering with subsequent anomaly analysis.

[0142] For example, the above-mentioned time period anomaly threshold can be set to 0.5. It should be noted that in practical applications, the time period anomaly threshold can be determined according to requirements, and this embodiment of the invention does not impose specific limitations.

[0143] (7) Integrate the abnormal times and abnormal periods corresponding to each monitoring point to determine the abnormal water flow dataset.

[0144] In this step, the anomaly screening submodule 122 integrates abnormal moment data and abnormal period data for each monitoring point in the water supply network. The integrated abnormal moment data includes: moments where the normalized abnormal parameter of any indicator at the monitoring point exceeds the abnormal threshold, which need to be labeled by indicator type to facilitate subsequent analysis of the anomaly correlation between different indicators. The integrated abnormal period data includes: abnormal flow growth segments where the normalized abnormal probability parameter of any period exceeds the abnormal threshold, requiring the recording of the start and end times of the period and the corresponding abnormal indicator type.

[0145] Based on the above technical solution, the embodiments of the present invention can integrate the abnormal time data and abnormal time period data of a single monitoring point into a subset of abnormal data of a single monitoring point, and then summarize the subsets of abnormal data of all monitoring points to finally form a water flow abnormal dataset covering the entire pipeline network. This provides basic data for the subsequent steps of determining the results of abnormal events, ensuring that subsequent steps can accurately identify sudden water intake abnormalities and leakage abnormalities.

[0146] For example, in combination Figure 5 ,like Figure 6 The diagram illustrates a flowchart of another GIS-based DMA intelligent zoning method for water supply networks according to an embodiment of the present invention. In this method, preliminary results for water supply network zoning are determined based on a water flow anomaly dataset, specifically including:

[0147] S601. Based on the abnormal water flow dataset, determine the results of abnormal event differentiation.

[0148] The abnormal event differentiation results are used to indicate the type of abnormal event that occurred in the water supply network. The abnormal event types include sudden water intake abnormalities and leakage abnormalities.

[0149] In one possible implementation, step S601 can be specifically executed by the anomaly type region sub-module 123 in the water flow anomaly analysis module 12 described above. For example, the anomaly type region sub-module 123 implements S601 through the following steps:

[0150] (1) Based on the water flow anomaly dataset, identify multiple monitoring points for water flow anomalies.

[0151] Specifically, the anomaly type area sub-module 123 extracts all monitoring points containing abnormal time data or abnormal time period data from the water flow anomaly dataset and records them as water flow anomaly monitoring points, in order to exclude normal monitoring points without any abnormal records.

[0152] (2) Calculate the similarity index of water flow anomalies between every two adjacent monitoring points among multiple monitoring points with water flow anomalies.

[0153] Among them, two adjacent monitoring points are two monitoring points that are directly connected in the topology, and the water flow anomaly similarity index is used to characterize the degree of similarity of abnormal events between two adjacent monitoring points.

[0154] Since anomalies at a single monitoring point can interfere with upstream and downstream operations, it is necessary to determine, based on the water supply network topology, whether anomalies at adjacent monitoring points are caused by the same abnormal event. This is specifically achieved by calculating a flow anomaly similarity index. For example, the flow anomaly similarity index is determined using the following formula:

[0155]

[0156] in, Let monitoring point k be the monitoring point at the time of the z-th abnormal event. The similarity index of water flow anomalies between them Let monitoring point k be the monitoring point at the time of the z-th abnormal event. The total duration corresponding to all abnormal periods The difference, The time of occurrence of the z-th abnormal event at monitoring point k is the distance from the monitoring point. The duration of the corresponding abnormal event is given by exp(), which is an exponential function with the natural constant as its base.

[0157] It should be noted that, The smaller the value, The smaller the value, the closer the degree of anomaly between two adjacent monitoring points and the more synchronous the occurrence time, the more likely it is caused by the same abnormal event, which is characterized by... The larger the value, the better.

[0158] (3) Determine the local abnormal area based on the similarity index of water flow between every two adjacent monitoring points.

[0159] Among them, the local anomaly region is used to characterize the set of monitoring points that cause water flow anomalies due to the same anomalous event.

[0160] In this step, the anomaly type region submodule 123 normalizes the flow anomaly similarity index to the (0, 1) interval and sets a similarity threshold. Then, the anomaly type region submodule 123 merges two adjacent monitoring points whose normalized flow anomaly similarity index is greater than the similarity threshold. Afterwards, the anomaly type region submodule 123 repeats the above calculation and merging operations for the newly merged monitoring points until no more monitoring points meeting the criteria can be added, ultimately forming a local anomaly region, i.e., a set of monitoring points affected by a single anomaly event.

[0161] Meanwhile, the anomaly type area molecular module 123 sorts the monitoring points in the local anomaly area from earliest to latest according to the time of anomaly occurrence, and generates a local anomaly transmission sequence for subsequent judgment of the source and propagation pattern of the anomaly.

[0162] (4) Determine the abnormality type of the local abnormal area to obtain the abnormal event differentiation result.

[0163] In this step, the anomaly type region sub-module 123 determines the anomaly event differentiation result through the following steps:

[0164] (4.1) For each monitoring point in the local abnormal area, determine the historical recurrence index of each monitoring point.

[0165] The historical recurrence index is used to characterize the similarity between current abnormal events and historical abnormal events at each monitoring point.

[0166] For example, the anomaly type region molecular module 123 determines the historical recurrence index of each monitoring point using the following formula:

[0167]

[0168] in, For the z-th anomalous event at monitoring point k, and the historical [missing information], The historical recurrence index of this anomalous event. For the z-th anomaly and the historical [number]th [event] The number of duplicate monitoring points in the local anomaly region corresponding to this anomaly event. For the z-th anomaly and the historical [number]th [event] The duration difference value of the sub-abnormal events For the z-th anomaly and the historical [number]th [event] The number of monitoring points corresponding to the longest common subsequence in the local anomaly transmission sequence of the two anomaly events. The aforementioned common subsequence is the sequence composed of the same monitoring points in the local anomaly transmission sequences corresponding to the two anomaly events.

[0169] (4.2) Determine the probability index of sudden water intake in local abnormal areas based on the historical recurrence index of each monitoring point, the abnormal recovery speed of local abnormal areas, and the duration of abnormal events.

[0170] Among them, the abnormal recovery rate is used to indicate the speed at which a local abnormal area recovers to a normal state after an abnormal event occurs, and the sudden water intake probability index is used to characterize the probability that the abnormal event occurring in the local abnormal area is a sudden water intake anomaly.

[0171] Optionally, the duration of an anomaly can be determined by anomaly time period data: take the start time and end time of the anomaly time period corresponding to an anomaly event, and the difference between the two is the anomaly duration of the anomaly event; if there are multiple consecutive anomaly time periods and they belong to the same anomaly event, then the duration difference between the start time of the first anomaly time period and the end time of the last anomaly time period is used for calculation.

[0172] For example, the anomaly type region molecular module 123 determines the probability index of sudden water intake in a local anomaly region using the following formula:

[0173]

[0174] in, Let z be the probability index of sudden water withdrawal for the z-th abnormal event at monitoring point k. The duration of the z-th abnormal event. The historical recurrence index of the z-th anomalous event and all historical anomalous events. The mean, This is the difference between the maximum value of the anomaly probability parameter at all times corresponding to the z-th anomaly event and the anomaly probability parameter at the first time of normal flow after the anomaly event. This represents the time interval between the time of the anomaly corresponding to the maximum value of the anomaly probability parameter among all times under the z-th anomaly event and the first time of normal water flow after the occurrence of this anomaly event. It should be noted that... The larger the value, the faster the recovery speed of abnormal events.

[0175] (4.3) Determine the results of abnormal event differentiation based on the probability index of sudden water intake in the local abnormal area.

[0176] Specifically, the molecular module 123 of the abnormal type area will include the probability index of sudden water intake. Normalized to the (0, 1) interval, and a threshold for determining sudden water withdrawal is set. Furthermore, the anomaly type region molecular module 123 assigns the sudden water withdrawal probability index... Anomalies in local anomaly regions exceeding the sudden water withdrawal threshold are classified as sudden water withdrawal anomalies; the anomaly type submodule 123 assigns the sudden water withdrawal probability index. Abnormal events in local abnormal areas that are less than or equal to the sudden water intake threshold are classified as leakage anomalies.

[0177] For example, the above-mentioned sudden water intake judgment threshold can be set to 0.5. It should be noted that in practical applications, the sudden water intake judgment threshold can be determined according to the needs, and the embodiments of the present invention do not impose specific limitations.

[0178] Furthermore, the anomaly type area sub-module 123 integrates the judgment results of all local anomaly areas to form anomaly event differentiation result, clarifying the type and scope of each anomaly area.

[0179] S602. Based on the results of the abnormal event differentiation, adjust the weight of each monitoring point in the GNN to obtain the preliminary results of the water supply network partitioning output by the GNN.

[0180] Among them, GNN is used to determine the preliminary results of water supply network zoning based on the topology of the water supply network and the adjusted weight of each monitoring point. The preliminary results of water supply network zoning are used to characterize the preliminary division range of each area in the water supply network.

[0181] In one possible implementation, S602 can be executed by the water supply network partitioning module 13 in the intelligent partitioning system 10 described above, specifically including the following steps:

[0182] (1) Determine the weight adjustment priority index for each monitoring point based on the historical leakage anomaly frequency of each monitoring point, the number of monitoring points included in the corresponding local anomaly area, and the anomaly type of the corresponding anomaly event.

[0183] In this step, the node weight adjustment submodule 131 in the water supply network partitioning module 13 can be used to perform the specific operation. The node weight adjustment submodule 131 first sets the initial weight of each network node (corresponding to a monitoring point) in the GNN to 1, and limits the weight increase to 5.

[0184] For example, the node weight adjustment submodule 131 determines the weight adjustment priority index for each monitoring point according to the following formula:

[0185]

[0186] in, Let k be the priority index for weight adjustment at time i. This represents the frequency of leakage anomalies at monitoring point k in historical comparison data. To monitor the degree to which point k conforms to the leakage characteristics at time i, Let be the number of monitoring points contained in the local anomaly region where monitoring point k is located at time i. Let be the sequence number of the local anomaly propagation sequence corresponding to monitoring point k at time i. It should be noted that... The larger the value, the larger the local abnormal area.

[0187] Furthermore, the node weight adjustment submodule 131 will adjust the weight priority index. Normalize to (0,1), and label the normalization result as... .

[0188] For example, the node weight adjustment submodule 131 adjusts the weight of monitoring point k according to the following formula:

[0189]

[0190] in, Let k be the weight of the monitoring point at time i. Let k be the weight of the monitoring point at time i-1. This is the weighted priority index after normalization.

[0191] (2) For each monitoring point in the GNN, adjust the weight of each monitoring point in the GNN according to the priority index, initial weight and preset weight adjustment rules of each monitoring point.

[0192] The preset weight adjustment rules include: if the abnormal event differentiation result indicates that the monitoring point corresponds to leakage abnormality, the weight of the monitoring point will be increased and will continue for a preset duration; if the abnormal event differentiation result indicates that the monitoring point corresponds to sudden water intake abnormality, the weight of the monitoring point will be temporarily increased before the sudden water intake abnormality ends.

[0193] (3) Input the adjusted weight of each monitoring point into GNN, and output the preliminary results of water supply network zoning from GNN.

[0194] In this step, the GNN partitioning calculation submodule 132 within the water supply network partitioning module 13 can be used for execution. Specifically, the GNN partitioning calculation submodule 132 receives the adjusted weights output by the node weight adjustment submodule 131, combines them with the water supply network topology, calculates partitions using the topology data processing capabilities of the GNN, outputs preliminary water supply network partitioning results, clarifies the preliminary division range of each region, and passes it to the DMA partitioning optimization module 14. It should be noted that nodes with higher weights are more likely to be partitioned into independent regions during partitioning.

[0195] Based on the above technical solution, the embodiments of the present invention can determine the preliminary results of water supply network partitioning according to the water flow anomaly dataset. The results are presented in the form of a list of regional ranges, which clearly defines the network nodes, monitoring points and corresponding anomaly types included in each preliminary region. This provides the core calculation basis for subsequent determination of DMA partitioning by combining physical equipment information, and ensures that subsequent optimization can accurately match the physical characteristics of the network.

[0196] For example, in combination Figure 6 ,like Figure 7 The diagram illustrates a flowchart of another GIS-based DMA intelligent zoning method for water supply networks according to an embodiment of the present invention. In this method, the DMA zoning of the water supply network is determined based on the preliminary results of the water supply network zoning and the physical equipment information of the water supply network, specifically including:

[0197] S701. Map the preliminary results of water supply network zoning with the physical equipment information of the water supply network to determine multiple hydraulically non-independent areas.

[0198] Among them, the hydraulically non-independent area is the area without key dividing valves and where water flow cannot be controlled independently.

[0199] Optionally, the physical equipment information of the water supply network includes two types of core data: pipeline attribute data and control equipment data. Among them, the pipeline attribute data covers the pipe diameter, material, laying length, laying year, and connection nodes at both ends of each pipe section; the control equipment data includes the geographical coordinates, elevation values, and equipment status of valves and water pumps.

[0200] In one possible implementation, the DMA partitioning optimization module 14 in the intelligent partitioning system 10 spatially correlates the above-mentioned physical equipment information with the preliminary results of the water supply network partitioning. Through spatial overlay analysis of the GIS system, the physical equipment such as pipes, valves, and water pumps included in each area of ​​the preliminary partitioning results are identified, forming a preliminary partitioning-physical equipment comparison table, which provides basic data for subsequent verification of the hydraulic independence of the area.

[0201] Furthermore, based on the aforementioned correlation model, the DMA partitioning optimization module 14 performs hydraulic independence verification on the regions of each preliminary partition, identifying hydraulically non-independent regions: if the connecting pipe section between the region and the adjacent region lacks available key dividing valves (such as check valves or damaged valves), water flow isolation cannot be achieved; or there are continuous pipe sections that are not cut off across regions, allowing water flow to freely cross the boundary; or there is no equipment with flow / pressure regulation function in the region, making it impossible to independently maintain normal water pressure, then the region is determined to be a hydraulically non-independent region, and a list of multiple hydraulically non-independent regions is compiled.

[0202] S702. For each hydraulically non-independent region, determine the region with the strongest hydraulic correlation adjacent to the hydraulically non-independent region based on the topology of the water supply network.

[0203] In one possible implementation, the DMA partitioning optimization module 14 in the intelligent partitioning system 10 identifies all preliminary areas directly adjacent to each hydraulically non-independent area based on the topology of the water supply network (pipeline connection relationships, node association paths), clarifies the physical connection between each adjacent area and the hydraulically non-independent area (such as shared pipe segment locations, connection node distribution), and constructs an "adjacent area association map".

[0204] Furthermore, based on the correlation map of adjacent areas, the hydraulic correlation strength between each adjacent area and the hydraulically non-independent area is comprehensively determined: the longer the shared pipe section and the larger the pipe diameter, the higher the correlation strength; the more frequent the historical water flow exchange and the larger the exchange flow, the higher the correlation strength; the stronger the valve and pump linkage control capability of adjacent areas, the higher the correlation strength. The adjacent preliminary area with the highest hydraulic correlation strength is identified as the merging object of the hydraulically non-independent area, ensuring that the merged area has unified control conditions in terms of physical connection.

[0205] S703. Merge each hydraulically independent region into the corresponding adjacent region with the strongest hydraulic correlation to form multiple target regions.

[0206] In this step, the DMA partitioning optimization module 14 merges the hydraulically independent regions into the adjacent preliminary regions with the "highest hydraulic correlation" to form multiple target regions, ensuring that the multiple target regions have a unified control basis in terms of physical connection.

[0207] S704. Determine the DMA area boundary by identifying the valve or pipe section diameter change point closest to the boundary of multiple target areas, thus obtaining the DMA partition of the water supply network.

[0208] In this step, the DMA partition optimization module 14 calibrates the partition boundaries based on the merged area and the actual location of the physical equipment. The specific rules are as follows: a. Prioritize valves as boundary references: Select the valve that can be opened and closed normally closest to the theoretical boundary of the merged area as the boundary point to ensure that water flow isolation between the area and the outside can be achieved by closing the valve; b. Secondarily select pipe diameter change points as boundary references: If there are no suitable valves near the theoretical boundary, select the location where the pipe diameter changes significantly as the boundary point to use the hydraulic resistance characteristics formed by the pipe diameter difference to assist in area isolation; c. Boundary continuity verification: Draw the calibrated boundary lines through the GIS system to ensure that the boundary lines are completely closed and consistent with the actual pipeline direction to avoid the logical loophole of "the boundary crosses the pipe section but does not cut off the water flow".

[0209] Furthermore, the DMA zoning optimization module 14 re-verifies whether all areas after the calibration boundary meet the core requirements of DMA zoning: a. Independent metering: whether a meter capable of measuring total flow is installed at the area entrance, and whether the meter data can be matched with the water consumption statistics of users in the area; b. Independent controllability: whether the pressure and flow rate in the area can be adjusted independently through the operation of boundary valves / pumps; c. Leakage identification: whether the distribution of monitoring points in the area can cover the main pipelines to ensure that leakage can be accurately located when it occurs.

[0210] Understandably, the area that passes the final verification is the DMA partition of the water supply network. The output results include the boundary coordinates of each partition, the list of physical equipment included, the location of control valves and metering parameters, providing an executable partitioning basis for leakage monitoring and refined operation and maintenance of the water supply network.

[0211] Based on the above technical solutions, this embodiment of the invention maps the preliminary results of water supply network zoning with physical equipment information to accurately identify hydraulically non-independent areas lacking key dividing valves, with abnormal pipeline connectivity, or insufficient control capabilities. This solves the problem of the preliminary zoning being based solely on theoretical calculations and disconnected from the actual physical characteristics of the network. Furthermore, by combining the water supply network topology with the characteristics of shared pipe sections, historical water flow exchange, and the synergy of control equipment, the invention identifies and merges the adjacent areas with the strongest hydraulic correlation for each hydraulically non-independent area, ensuring that the merged areas have a unified control basis in terms of physical connectivity. Finally, after boundary calibration and verification of DMA core requirements, the resulting DMA zoning not only conforms to the actual physical layout of the network but also meets the requirements of independent metering, independent water control, and leak identification. This effectively improves the accuracy of water supply network leak monitoring and the efficiency of refined operation and maintenance, realizing the implementation of a practically executable control scheme from theoretical zoning.

[0212] In this embodiment of the invention, the DMA intelligent zoning device for GIS-based water supply networks can be divided into functional modules or functional units according to the above method examples. For example, each function can be divided into its own functional modules or functional units, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or as software functional modules or functional units. The module or unit division in this embodiment is illustrative and represents only one logical functional division; other division methods may be used in actual implementation.

[0213] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0214] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A GIS-based DMA intelligent zoning method for water supply networks, characterized in that, The method includes: Acquire water flow status monitoring data; wherein, the water flow status monitoring data is used to characterize the dynamic characteristics of water flow at different monitoring points within the water supply network; Based on the water flow status monitoring data, a water flow anomaly dataset is determined; wherein, the water flow anomaly dataset includes abnormal time data and abnormal time period data for each monitoring point, the abnormal time data is used to characterize the time when an abnormal event occurs at the monitoring point, and the abnormal time period data is used to characterize the time range within which an abnormal event occurs continuously at the monitoring point. Based on the aforementioned water flow anomaly dataset, preliminary results of water supply network zoning are determined; wherein, the preliminary results of water supply network zoning are used to characterize the preliminary division range of each area in the water supply network; Based on the preliminary results of the water supply network partitioning and the physical equipment information of the water supply network, the DMA partitioning of the water supply network is determined; Specifically, determining the abnormal water flow dataset based on the water flow status monitoring data includes: For each monitoring point, the water flow status monitoring data are arranged in time sequence according to different indicators to obtain the time sequence curves of each indicator; The time series curves of the aforementioned indicators are segmented according to adaptive principal component analysis (APCA). Based on the segmented time-series curves of each indicator, the water flow anomaly dataset is determined, specifically including: Obtain historical comparison data; wherein, the historical comparison data includes water flow status monitoring data of the water supply network within a historical period; For each moment in the time series curve of each indicator, an anomaly probability parameter is determined based on the historical comparison data corresponding to each moment and the duration of the time period in which each moment is located; wherein, the anomaly probability parameter is used to characterize the probability that the corresponding indicator will be abnormal at the current moment; Based on the time anomaly probability parameter at each time moment, multiple abnormal times are determined; wherein, the abnormal time moment is the time moment when the anomaly probability parameter is greater than the time anomaly threshold. Based on the multiple abnormal moments, an abnormal flow growth segment is determined; wherein, the difference between the water flow status monitoring data corresponding to each moment in the abnormal flow growth segment and the average value of the same period in the historical comparison data is greater than 0, and the monitoring value at the right end of the time period at each moment is greater than the monitoring value at the left end. Based on the time-time anomaly probability parameter corresponding to the abnormal moment of the abnormal traffic growth segment and the number of abnormal indicators of the abnormal traffic growth segment, the time-period anomaly probability parameter of the abnormal traffic growth segment is determined. Based on the time-period anomaly probability parameters of the abnormal traffic growth segment, multiple abnormal time periods are determined; By integrating the abnormal times and periods corresponding to each monitoring point, the abnormal water flow dataset is determined. The preliminary results of determining the water supply network zoning based on the water flow anomaly dataset specifically include: Based on the water flow anomaly dataset, the abnormal event differentiation result is determined, including: identifying multiple monitoring points with water flow anomalies based on the water flow anomaly dataset; calculating the water flow anomaly similarity index between every two adjacent monitoring points; wherein, the two adjacent monitoring points are two monitoring points directly connected in the topology, and the water flow anomaly similarity index is used to characterize the similarity of the abnormal events of the two adjacent monitoring points; determining local abnormal regions based on the water flow anomaly similarity index between every two adjacent monitoring points; wherein, the local abnormal region is used to characterize the set of monitoring points whose water flow anomalies are caused by the same abnormal event; determining the anomaly type of the local abnormal region to obtain the abnormal event differentiation result; wherein, the abnormal event differentiation result is used to indicate the anomaly type of the abnormal event occurring in the water supply network, and the anomaly type includes sudden water intake anomalies and leakage anomalies; Based on the abnormal event differentiation results, the weights of each monitoring point in the GNN are adjusted to obtain the preliminary results of the water supply network zoning output by the GNN, including: Based on the historical leakage anomaly occurrence frequency of each monitoring point, the number of monitoring points included in the corresponding local anomaly area, and the anomaly type of the corresponding anomaly event, a weight adjustment priority index is determined for each monitoring point. For each monitoring point in the GNN, the weight of each monitoring point in the GNN is adjusted according to the weight adjustment priority index, the initial weight, and the preset weight adjustment rule. The preset weight adjustment rule includes: if the anomaly event differentiation result indicates that the monitoring point corresponds to the leakage anomaly, then the weight of the monitoring point is increased and maintained for a preset duration; if the anomaly event differentiation result indicates that the monitoring point corresponds to the sudden water intake anomaly, then the weight of the monitoring point is temporarily increased before the sudden water intake anomaly ends. The adjusted weights of each monitoring point are input into the GNN, and the GNN outputs the preliminary results of the water supply network zoning. The preliminary results of the water supply network zoning are used to characterize the preliminary division range of each area in the water supply network. The step of determining the DMA partition of the water supply network based on the preliminary results of the water supply network partitioning and the physical equipment information of the water supply network specifically includes: The preliminary results of the water supply network zoning are mapped with the physical equipment information of the water supply network to identify multiple hydraulically non-independent areas; wherein, the hydraulically non-independent areas are areas without key dividing valves and where water flow cannot be controlled independently; For each of the hydraulically non-independent regions, the region with the strongest hydraulic correlation adjacent to the hydraulically non-independent region is determined according to the topology of the water supply network. Each of the hydraulically non-independent regions is merged into the corresponding adjacent region with the strongest hydraulic correlation to form multiple target regions; The nearest valve or pipe section diameter change point to the boundary of multiple target areas is determined as the DMA area boundary, thus obtaining the DMA partition of the water supply network.

2. The DMA intelligent zoning method for water supply networks based on GIS according to claim 1, characterized in that, The abnormality type of the local abnormal region is determined to obtain the abnormal event differentiation result, specifically including: For each monitoring point in the local anomaly region, a historical recurrence index is determined for each monitoring point; wherein, the historical recurrence index is used to characterize the similarity between the current anomaly event and the historical anomaly event at each monitoring point; Based on the historical recurrence index of each monitoring point, the abnormal recovery rate of the local abnormal area, and the duration of the abnormal event, the probability index of sudden water withdrawal in the local abnormal area is determined; wherein, the abnormal recovery rate is used to indicate the speed at which the local abnormal area recovers to a normal state after an abnormal event occurs, and the probability index of sudden water withdrawal is used to characterize the probability that the abnormal event occurring in the local abnormal area is a sudden water withdrawal anomaly. The abnormal event differentiation result is determined based on the probability index of sudden water intake in the local abnormal area.

3. The DMA intelligent zoning method for water supply networks based on GIS according to claim 2, characterized in that, After acquiring the water flow state monitoring data, the method further includes: According to the preset monitoring cycle, the timestamps of the detection data corresponding to different indicators in the water flow status monitoring data are aligned. The water flow status monitoring data is preprocessed; wherein, the preprocessing includes missing value completion and invalid data removal.

4. The DMA intelligent zoning method for water supply networks based on GIS according to any one of claims 1-3, characterized in that, The water flow status monitoring data includes pressure data, flow rate data, flow velocity data, and water quality data at each monitoring point in the water supply network.

Images