A dma-based hierarchical zoning planning method for water supply network

Abstract

This invention relates to a DMA-based hierarchical and zonal planning method for water supply networks, belonging to the technical field of water supply network zonal planning. The method includes: acquiring network data, constructing a weighted undirected graph to extract the skeleton network; removing skeleton pipe segments and performing community detection on the remaining network to generate initial DMA boundaries; performing a first zero-pressure test to obtain measured pressure decay curves, which are then verified through implicit connectivity inversion and hydraulic decoupling; determining candidate valve positions based on the verification boundary pipe segment list for initial redundancy screening; using the NSGA-III algorithm to solve for the optimal boundary valve closure and flow meter layout scheme with metering coverage, reliability loss, and water age increment as objectives; and performing a second zero-pressure test based on the optimal boundary valve closure list. This invention achieves optimal synergy between reliable water supply and accurate metering through the extraction and protection of the skeleton network, dual verification via implicit connectivity inversion and hydraulic decoupling, multi-objective optimized layout, and a closed-loop secondary verification.

Description

Technical Field

[0001] This invention belongs to the field of water supply network zoning planning technology, specifically relating to a hierarchical zoning planning method for water supply networks based on DMA. Background Technology

[0002] Independent metering area (DMA) management is a core means of controlling leakage in water supply networks. Traditional DMA zoning methods typically define boundaries based on topography, administrative divisions, or experience, verify the sealing performance through a one-time valve closure test, and install flow meters in the boundary pipe sections for metering.

[0003] However, existing methods have the following shortcomings: First, they lack systematic identification and protection of backbone pipe sections that bear the main water transmission and transfer functions in the pipeline network. The selection of boundary valves can easily cut off key hydraulic channels and weaken the overall water supply resilience of the pipeline network. Second, relying solely on a single valve closure test makes it difficult to discover hidden connecting pipe sections left over from pipeline network archives or historical renovations. Furthermore, there is a lack of quantitative assessment of the degree of hydraulic coupling on both sides of the boundary, resulting in nominal DMA closure but serious actual hydraulic interference, affecting the accuracy of minimum flow analysis at night. Third, the determination of boundary valve closure and flow meter installation schemes fails to take into account the multi-objective coordination between metering coverage, hydraulic reliability, and water quality safety, and lacks quantitative optimization methods. Summary of the Invention

[0004] To address the aforementioned problems in existing technologies, this invention provides a hierarchical and zonal planning method for water supply networks based on DMA. By extracting and protecting the backbone network, performing dual verification through implicit connectivity inversion and hydraulic decoupling, optimizing the layout through multi-objective optimization, and implementing a secondary verification closed loop, the method achieves optimal synergy between reliable water supply and accurate metering.

[0005] The objective of this invention can be achieved through the following technical solutions: This disclosure provides a hierarchical and zoning planning method for water supply networks based on DMA, including the following steps: S1. Skeleton Extraction and Initial Hierarchical Division: Obtain pipeline network data, construct a weighted undirected graph to extract the skeleton network, remove skeleton pipe segments, perform community detection on the remaining pipeline network to generate the initial DMA boundary, and obtain the initial DMA boundary pipe segment list. S2, Dual-mode verification and boundary inversion: Based on the initial DMA boundary pipe segment list, the first zero-pressure test is performed to obtain the measured pressure decay curve, and then implicit connectivity inversion and hydraulic decoupling verification are performed in sequence to generate a verification boundary pipe segment list; S3. Redundancy Constraints and Multi-Objective Optimization: Based on the list of verification boundary pipe sections, candidate valve positions are determined for initial redundancy screening. With metering coverage, reliability loss and water age increment as objectives, the NSGA-III algorithm is used to solve the optimal boundary valve closure and flow meter layout scheme. S4. Zero-pressure retest and finalization: Perform a second zero-pressure test based on the optimal boundary valve closing list. Determine the effectiveness of the optimization by the zero-pressure time ratio and the percentage increase in the decay rate. If effective, finalize the decision; otherwise, backtrack to the implicit connectivity inversion iteration for correction. The redundant constraints and multi-objective optimization include the following steps: S31. Initial screening of redundancy in the candidate valve set: Based on the list of verification boundary pipe sections, determine the location of candidate valves, use the DIJKSTRA algorithm to calculate the increment of the shortest hydraulic path length for fire water supply after each valve is closed, and remove valves whose increment exceeds the preset threshold or are located on non-closable pipe sections of the skeleton network to obtain a candidate set of safety valves. S32. Multi-objective optimization of boundary valves and flow meters: Define valve state variables and flow meter installation variables. With the optimization objectives of maximizing metering coverage, minimizing hydraulic reliability loss, and minimizing water age increment, the NSGA-III algorithm is used to solve the Pareto optimal solution set under the constraints of fire head, prohibited installation of skeleton network valves, and inlet pressure fluctuation, to obtain the optimal boundary valve closure list and flow meter installation location scheme.

[0006] Furthermore, the skeleton extraction and initial hierarchical delineation include the following steps: S11. Pipeline network diagram model construction and attribute assignment: Obtain pipeline network GIS information data, pipe segment attribute extended data and SCADA historical operation data, abstract the pipeline network into a weighted undirected graph, and calculate the edge betweenness of each pipe segment and the degree centrality of each node as topological weights. S12. Skeleton Network Identification and Protection Marking: Define core score, sort all pipe segments in descending order of core score, extract the top-ranked pipe segments according to a preset ratio to form skeleton network, and mark all pipe segments in skeleton network as non-closable pipe segments. S13. Initial division of DMA metering cells: After removing the skeleton network segments, a community structure detection algorithm based on modularity optimization is executed on the remaining peripheral network. The nodes are automatically aggregated with the goal of maximizing modularity, the initial DMA partition boundaries are identified, and an initial DMA boundary segment list is generated.

[0007] Furthermore, the pipe segment side intermediate number Based on edge weight Calculate the percentage of times each node's shortest path passes through the current pipe segment:

[0008] In the formula, For nodes and The total number of shortest paths between [locations]. For the pipe section The number of shortest paths; The degree centrality of the nodes Defined as the normalized value of the number of pipe segments directly connected to a node:

[0009] In the formula, For nodes The degree value, This represents the total number of nodes in the network diagram model.

[0010] Furthermore, the core score is defined as: For connection nodes and pipe section The core score is expressed as: ; In the formula, For pipe segment edge intermediate number The normalized value, Degree centrality of nodes at both ends of the pipe segment and The value after mean normalization; , These are the weighting coefficients; The community structure detection algorithm uses the Louvain algorithm, based on modularity. Modularity is defined as the maximization of community partitioning quality as an evaluation metric.

[0011] In the formula, For the edge weights of the pipe segments, For nodes The weighting degree, For subgraph Half of the total edge weights For nodes Community affiliation number, This is an indicator function.

[0012] Furthermore, the dual-mode verification and boundary inversion include the following steps: S21. First zero-pressure field test: Close the corresponding valves according to the initial DMA boundary pipe section list, set up a temporary pressure recorder to perform a zero-pressure test, and collect the measured pressure decay curves of each DMA internal node. S22, Implicit Connectivity Inversion: Construct a pressure-driven node flow hydraulic model to simulate the theoretical pressure decay curve under ideal closed conditions; identify pressure anomaly regions by comparing the residuals of the measured curves and the theoretical curves; use the pressure gradient anomaly clustering algorithm to invert and calculate for pressure anomaly regions, output a list of suspected implicit connectivity pipe segments, update the network diagram model after on-site verification and repair, and generate a corrected boundary pipe segment list based on the initial DMA boundary pipe segment list; S23. Boundary hydraulic decoupling verification: Using the corrected boundary pipe segment list and SCADA historical pressure data, calculate the pressure correlation coefficient between any two nodes and construct a node pressure similarity matrix; identify node pairs with correlation coefficients on both sides of the boundary that are higher than a preset threshold, determine them as hydraulic coupling areas, mark the corresponding boundary pipe segments as boundaries to be adjusted and move them outward in the direction of low correlation to correct them, and generate a verification boundary pipe segment list.

[0013] Furthermore, in the pressure-driven nodal flow hydraulic model, the expression for nodal flow is:

[0014] In the formula, For nodes Under current pressure The actual outflow rate is below. For nodes Design water demand under reference pressure For reference pressure, The pressure-flow relationship function is described using a piecewise function to depict the effect of pressure on the outflow.

[0015] Furthermore, in the implicit connectivity inversion, the screening method for suspected implicit connectivity segments in abnormal regions is as follows: Construct the feature vector for each monitoring point ,in Here are the plane coordinates of the monitoring point. This is the normalized pressure gradient value; The DBSCAN algorithm is used to identify spatial clusters whose gradient values ​​are significantly different from the overall values. Suspected hidden connected regions are delineated with the geometric center of the abnormal cluster as the center and a preset search radius. Within this region, pipe segments in the pipe network diagram model that are located at one end inside the current DMA and the other end outside the current DMA or skeleton network nodes, and are not marked as boundary pipe segments in the initial DMA boundary pipe segment list, are searched to generate a list of suspected hidden connected pipe segments.

[0016] Furthermore, the optimization objectives are defined as follows: Objective function for maximizing meter coverage:

[0017] in, This represents the total number of DMA inlet pipe segments. Install variables for the flow meter. For the first The average daily water consumption of the DMA served by each DMA inlet pipe section For nodes Design water demand, This represents the total number of nodes in the network diagram model; Objective function for minimizing hydraulic reliability loss:

[0018] in, This represents the minimum cut set capacity of the skeleton mesh in its original state. The minimum cut set capacity of the skeleton mesh after closing the selected boundary valve; Objective function for minimizing water age increment:

[0019] in, and These represent the average water age at each DMA outlet node before and after valve closure. For the first The weight of water consumption percentage for each DMA.

[0020] Furthermore, the NSGA-III algorithm solution includes: Population individuals are randomly generated, and each individual is encoded as a decision variable vector. ,in The number of candidate valves; For each individual, update the graphical model and compute three objective function values, and check for constraint violations; search for the Pareto optimal frontier through non-dominated ordination and a reference point-based selection strategy; Offspring individuals are generated by simulating binary crossover and polynomial mutation. The iteration continues until the maximum number of iterations or Pareto front convergence is reached. The output consists of all individuals in the first non-dominated level, forming the Pareto optimal solution set.

[0021] Furthermore, the zero-pressure retest finalization includes: A second zero-pressure test was performed based on the optimal boundary valve closure list to obtain the optimized measured pressure decay curves for each DMA. For each monitoring point of each DMA, the pressure zeroing time ratio and the decay rate improvement percentage are calculated respectively. The zeroing time ratio is defined as the ratio of the optimized zeroing time to the zeroing time of the first test, and the decay rate improvement percentage is defined as the improvement of the optimized average decay rate relative to the first test. If the overall return-to-zero time ratio of a certain DMA is less than or equal to the effective threshold one, or the overall decay rate improvement percentage is greater than or equal to the effective threshold two, then the DMA optimization is deemed effective. If all DMAs are deemed valid, a verification report and the final solution are output. If an invalid DMA exists, it is marked as needing to be reviewed and the implicit connectivity inversion step is returned. The second test data is used for iterative correction until all DMAs pass verification.

[0022] The beneficial effects of this invention are as follows: This invention extracts the skeleton network based on the core degree score weighted by edge betweenness and degree centrality and marks it as non-closable pipe segments, thus ensuring the connectivity and fire redundancy of the skeleton network's water supply arteries after DMA boundary isolation from the source. Secondly, it locates implicitly connected pipe segments by inverting the pressure-driven model theory and the measured pressure attenuation residuals and clustering them, forming a true physical closed boundary after on-site correction. At the same time, it uses the historical pressure correlation coefficient matrix to identify hydraulic coupling areas and move the boundary outward, eliminating the phenomenon of hydraulic false closure and laying a strict boundary condition for leakage analysis. Thirdly, under the constraints of skeleton network, fire head, and pressure fluctuation, with the objectives of maximizing metering coverage, minimizing reliability loss, and minimizing water age increment, the NSGA-III algorithm is used to perform multi-objective optimization of the normally closed boundary valve scheme and flow meter layout, outputting a Pareto optimal scheme set that takes into account metering accuracy, water supply safety, and water quality stability. Finally, through a second zero-pressure verification and effective threshold determination, an iterative closed loop is constructed to ensure that the delivered scheme truly meets the standards for closure and hydraulic independence in actual operation. Attached Figure Description

[0023] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0024] Figure 1 A schematic diagram illustrating the steps of a DMA-based hierarchical and zoning planning method for water supply networks, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the steps of skeleton extraction and initial hierarchical division provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the steps of dual-mode verification and boundary inversion provided in an embodiment of the present invention; Figure 4 A schematic diagram illustrating the steps of redundant constraints and multi-objective optimization provided in an embodiment of the present invention. Detailed Implementation

[0025] To further illustrate the technical means and effects of the present invention in achieving the intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] This embodiment provides a hierarchical and zoning planning method for water supply networks based on DMA, such as... Figure 1 As shown, it includes the following steps: S1. Skeleton Extraction and Initial Hierarchical Delineation: Obtain pipeline network data, construct a weighted undirected graph, extract the skeleton network, remove skeleton pipe segments, perform community detection on the remaining pipeline network to generate initial DMA boundaries, and obtain a list of initial DMA boundary pipe segments, such as... Figure 2 As shown, it includes the following steps: S11. Pipeline network model construction and attribute assignment: Obtain pipeline network GIS information data, pipeline segment attribute extended data and SCADA historical operation data, abstract the pipeline network into a weighted undirected graph, and calculate the edge betweenness number of each pipeline segment and the degree centrality of each node as topological weights.

[0027] Specifically, the construction and attribute assignment of the pipeline network diagram model includes the following steps: Basic data acquisition: Pipeline network GIS information data, pipe segment attribute extended data, and SCADA historical operation data are obtained from the water supply enterprise's information management system. The pipeline network GIS information data includes the coordinates of the pipe segment's start and end nodes, pipe segment length, pipe diameter, pipe material type, node coordinates, node type, and ground elevation. The pipe segment attribute extended data includes roughness coefficients (Hayzen-Williams coefficient or Manning coefficient, determined according to industry standards based on pipe material and laying year), design flow limit, and current valve opening / closing status. The SCADA historical operation data includes time-series pressure data from minute-level pressure monitoring points and time-series flow data from the water source pumping station outlet for the past 30 consecutive days. Consistency verification and preprocessing are performed on these three types of data, including node-to-pipe segment matching verification, isolated record removal, and attribute null value completion.

[0028] Weighted undirected graph model construction: Abstracting the preprocessed pipeline network into a weighted undirected graph. ,in: Node set Each node Related attribute vector , respectively, are planar coordinates Ground elevation Design water demand and node type encoding ; Pipe segment assembly Each pipe section Connecting nodes and Related attribute vectors , respectively length Pipe diameter Roughness coefficient and valve status indicators ; edge weight To characterize the hydraulic impedance of the pipe section.

[0029] Topological weight calculation and assignment: in graph models G Based on this, calculate the edge intermediate number of each pipe segment. The degree centrality of each node is calculated as a topological weight used to measure its importance in the overall water transfer network. The topological weights are used to measure the local connectivity density. The calculated edge-interval values ​​and degree centrality values ​​are written into the corresponding fields of the pipe segment attribute table and the node attribute table, respectively, to form graph model data carrying topological weights.

[0030] Among them, the pipe section edge intermediate number Based on edge weight Calculate the percentage of times each node's shortest path passes through this pipe segment: ; In the formula, For nodes and The total number of shortest paths between [locations]. For the pipe section The shortest path number; the higher the edge number, the more prominent the transfer role of the pipeline segment in the water conveyance path.

[0031] Node degree centrality Defined as the normalized value of the number of pipe segments directly connected to a node, the formula is: ; In the formula, For nodes The degree value, This represents the total number of nodes in the pipeline network model; a higher degree centrality indicates that the node is located in a densely connected area of ​​the pipeline network.

[0032] After the above processing, the graphical model data includes the pipeline network geometry, hydraulic properties, and operational characteristics, and is supplemented with topological weight parameters that reflect the importance of the water conveyance structure. This integrates the originally scattered static GIS data and dynamic SCADA operational data into a unified pipeline network graphical model with quantifiable topological importance, providing a data foundation and algorithm input for the subsequent identification of the backbone network.

[0033] S12. Skeleton Network Identification and Protection Marking: Define core score, sort all pipe segments in descending order of core score, extract the top-ranked pipe segments according to a preset ratio to form skeleton network, and mark all pipe segments in skeleton network as non-closable pipe segments. Specifically, the skeleton mesh identification and protection markers include the following steps: Calculate the core score for each pipe segment For connecting nodes and pipe section The core score is defined as: ; In the formula, For pipe segment edge intermediate number The value after max-min normalization Degree centrality of nodes at both ends of the pipe segment and The value after mean normalization; weighting coefficients , satisfy .

[0034] In this embodiment, the following is taken , To emphasize the bridging role of the pipeline segment in the overall water conveyance path, while also taking into account local connectivity density characteristics, the core score is calculated. The range of values ​​is The higher the value, the closer the pipe section is to the core of the network in terms of topology, and the more important its water transmission function is.

[0035] All pipe sections are scored according to their core density. Sort in descending order and extract the top-ranked items. Pipe segments form a skeleton network of pipe segments. .

[0036] The preset ratio in this embodiment That is, the top 20% of pipe sections with the highest core score are selected.

[0037] Add a constraint field – a closable flag – to the pipe segment attribute table. For each pipe segment within the system, the "can be shut off" field is set to "No", indicating that it is a non-shut-off pipe segment; for the remaining pipe segments, the field is set to "Yes".

[0038] The pipe segment attribute table containing constraint fields and the skeleton network pipe segments ( The list is passed to subsequent steps along with the model; in the initial screening and multi-objective optimization of the valve candidate set, any candidate valve located on a pipe segment marked "No" for closability will be automatically eliminated, thereby ensuring that the setting of DMA boundary valves will not cut off the hydraulic connectivity of the skeleton network.

[0039] Based on the graph model carrying topological weights, the above process identifies the set of core pipe segments in the pipeline network that undertake the main water transmission and transfer functions, marks them as the protected skeleton network, and transfers the constraint condition to the subsequent boundary valve location stage.

[0040] S13. Initial division of DMA metering cells: After removing the skeleton network segments, a community structure detection algorithm based on modularity optimization is executed on the remaining peripheral network. The nodes are automatically aggregated with the goal of maximizing modularity, the initial DMA partition boundaries are identified, and an initial DMA boundary segment list is generated. Specifically, the initial division of DMA metering cells includes the following steps: From graph model Remove skeleton network segment set Preserve all nodes connected by the skeleton network. (These nodes will serve as the interface points connecting the peripheral pipeline network and the backbone network, and will be used in subsequent steps to determine the location of the DMA boundary valves), thus obtaining the peripheral pipeline network sub-graph. ,in The set of peripheral pipeline segments is represented as , This is the set difference operator.

[0041] Understandably, this subgraph It includes all branch or secondary ring pipe sections that do not undertake the main water transmission function, and is the direct operational object of DMA metering cell division.

[0042] Pair diagram The community structure detection algorithm is executed with the goal of maximizing modularity. Nodes are automatically aggregated into several communities with tight internal connections and sparse external connections. Each community corresponds to the prototype of a DMA metering cell.

[0043] This embodiment uses the Louvain algorithm, based on modularity. Modularity is defined as the maximization of community partitioning quality as follows: ; In the formula, Pipe segment edge weights , For nodes The weighting degree, For subgraph Half of the total edge weights For nodes Community affiliation number, This is an indicator function.

[0044] It should be noted that the Louvain algorithm gradually converges the modularity to its maximum value through alternating iterations of node movement and community aggregation, ultimately outputting the community affiliation of each node. All nodes belonging to the same community number together form an initial DMA partition.

[0045] Identify DMA boundary segments based on node community affiliation: If the nodes at both ends of a pipe segment belong to different communities, or if a node at one end of a pipe segment is connected to a node in the backbone network while the node at the other end belongs to a certain community, then the pipe segment is marked as the initial DMA boundary pipe segment; the closed area enclosed by all boundary pipe segments constitutes the initial DMA metering cell.

[0046] All marked initial DMA boundary pipe segments are summarized to generate an initial DMA boundary pipe segment list. This list is a structured data table, with each record corresponding to one boundary pipe segment, and includes at least the following fields: unique identifier of the pipe segment, starting node number, ending node number, DMA number (if one end is a skeleton network, then marked "skeleton network"), pipe diameter, pipe segment length, and current valve status identifier.

[0047] The above steps, based on the skeleton network marking information, perform community structure detection on the peripheral network after removing the skeleton network segments. The node community affiliation, the initial DMA boundary segment list, and the basic information of each DMA are used as the final output of the first stage of hierarchical delineation, and provide the initial partitioning basis for subsequent closed-loop tests.

[0048] S2. Dual-mode verification and boundary inversion: Based on the initial DMA boundary segment list, the first zero-pressure test is performed to obtain the measured pressure decay curve. This is followed by implicit connectivity inversion and hydraulic decoupling verification, ultimately generating a verification boundary segment list. Figure 3 As shown, it includes the following steps: S21. First zero-pressure field test: Close the corresponding valves according to the initial DMA boundary pipe section list, set up a temporary pressure recorder to perform a zero-pressure test, and collect the measured pressure decay curves of each DMA internal node.

[0049] Specifically, the first zero-pressure field test includes the following steps: Based on the initial DMA boundary pipe segment list output in step S13, identify the valves installed on each boundary pipe segment and compile a list of valves to be closed; for boundary pipe segments without valves or with valve failure, take measures such as installing temporary valves or freezing the pipes to ensure that the DMA boundary is completely cut off. The test was conducted during the period of lowest flow rate at night. At least three temporary pressure recorders were deployed inside each initial DMA, located at the most unfavorable point (the highest point or the end of the pipeline), near the geometric center, and near the inlet. The data sampling frequency was no less than once per minute. Close the boundary valves in sequence to hydraulically isolate each initial DMA from the external pipe network. Continuously collect pressure data at each monitoring point using a temporary pressure recorder and record the pressure decay process over time until the pressure stabilizes or reaches the preset test duration. After the test, the pressure time-series data of each monitoring point is exported to generate the measured pressure decay curve of each node within the DMA. For example, for each initial DMA node... 1 monitoring point, obtain time series ,in For monitoring points At any moment The measured pressure values ​​were obtained. For each DMA, a measured pressure decay curve was plotted with time on the horizontal axis and pressure on the vertical axis for each monitoring point. This curve visually reflects the rate of pressure decrease and the final stable value of the DMA after the valve is closed.

[0050] Through the above-mentioned zero-pressure field test, the actual hydraulic response data of each initial DMA in closed state were obtained, providing a measured benchmark for subsequent implicit connectivity inversion and hydraulic decoupling verification.

[0051] S22. Implicit Connectivity Inversion: Construct a pressure-driven node flow hydraulic model to simulate the theoretical pressure decay curve under ideal closed conditions; identify pressure anomaly regions by comparing the residuals of the measured curves and the theoretical curves; use the pressure gradient anomaly clustering algorithm to invert and calculate for pressure anomaly regions, output a list of suspected implicit connectivity pipe segments, update the network diagram model after on-site verification and repair, and generate a corrected boundary pipe segment list based on the initial DMA boundary pipe segment list.

[0052] Specifically, implicit connectivity inversion includes the following steps: Construct a pressure-driven nodal flow hydraulic model, expressing nodal flow as a function of nodal pressure: ; In the formula, For nodes Under current pressure The actual outflow rate; For nodes Design water demand under reference pressure; For reference pressure, the average service pressure of the node when the pipeline network is supplying water normally is usually taken, which is determined by SCADA historical data; The pressure-flow relationship function is described using a piecewise function to depict the effect of pressure on the outflow, such as the classic formula proposed by Wagner et al.: ; In the formula , The critical pressure ratio is the ratio of the minimum pressure at which flow begins to occur at the node to the reference pressure. In this embodiment, it is taken as... (Corresponds to approximately 2-3 meters of water head).

[0053] Embed the pressure-driven model into the hydraulic simulation engine (such as the pressure-driven extension module based on EPANET secondary development).

[0054] The pipeline network model (i.e., weighted undirected graph) from step S11 G Based on the topology, and using the boundary valve closure state recorded in step S21 as the boundary condition, hydraulic simulations are performed on each DMA, and the theoretical pressure decay curves for each monitoring point are output. .

[0055] Understandably, hydraulic simulation includes: The simulation start time is set to the valve closure completion time, and the initial state is the steady-state hydraulic condition before valve closure. The simulation period covers the complete duration of the actual test in step S21, and the hydraulic time step is consistent with the measured sampling interval (e.g., 1 minute). During the simulation, the flow rate at each node is dynamically updated according to the pressure-driven model, and the pipeline leakage is also adjusted synchronously according to the orifice outflow formula as the pressure changes. The simulated pressure time series values ​​at each monitoring point location (consistent with the temporary pressure recorder deployment locations in step S21) are output to form the theoretical pressure decay curve. ,in Number the monitoring point.

[0056] Calculate the measured pressure decay curve point by point Compared with the theoretical pressure decay curve Pressure residual If the measured pressure is higher than the theoretical pressure and the difference exceeds the preset threshold at the end of the test, or the measured attenuation rate is significantly lower than the theoretical value, the corresponding DMA and monitoring point will be marked as an abnormal pressure area.

[0057] Two types of anomaly detection, for example: Pressure cannot be returned to zero: At the end of the test (e.g., 120 minutes after valve closure), if the measured pressure at a certain monitoring point fails to reach zero... Higher than theoretical pressure And the difference exceeds the preset threshold (In this embodiment, a 1.0 meter water column is used). Therefore, the DMA is determined to have a physical sealing defect, i.e., an unidentified water inlet channel exists. The attenuation rate is significantly lower than expected: calculate the average attenuation rate of the measured and theoretical curves in the latter half of the test (e.g., from the 60th minute to the end). If the measured average attenuation rate is less than 50% of the theoretical average attenuation rate, and the pressure residual remains positive, then the DMA is determined to have hidden water replenishment. DMAs that meet any of the above conditions and their corresponding monitoring points are marked as pressure anomaly areas.

[0058] For abnormal areas, the pressure gradient of each monitoring point is calculated, and the density-based spatial clustering algorithm (DBSCAN) is used to jointly cluster the spatial coordinates and pressure gradients of the monitoring points to identify spatial clusters where pressure drop is obstructed. Based on the geometric center of the abnormal cluster, pipe segments that are not in the initial DMA boundary pipe segment list and are connected across DMAs are screened within a preset search radius to generate a list of suspected hidden connected pipe segments.

[0059] It should be noted that, for monitoring points The pressure gradient is defined as the rate of pressure change per unit time. Since the pressure should continuously decrease during a normal closed DMA zero-pressure test, the gradient should be negative; if there is implicit water replenishment in a certain area, the pressure decrease in that area is hindered, and the gradient value approaches zero or becomes positive. A density-based spatial clustering algorithm (DBSCAN) is used to jointly cluster the pressure gradient values ​​and spatial coordinates of the monitoring points. Specifically, a feature vector is constructed for each monitoring point: ; In the formula, Here are the plane coordinates of the monitoring point. This is the normalized pressure gradient value. Set the DBSCAN neighborhood radius. and minimum sample size It identifies spatial clusters whose gradient values ​​are significantly different from the overall value (i.e., pressure drop is hindered).

[0060] For the identified anomalous clusters, a suspected hidden connectivity region is delineated with the geometric center of the cluster as the center and a preset search radius (50 meters in this embodiment). Within this region, all pipe segments in the pipeline network model constructed in step S11 are retrieved, and pipe segments that simultaneously meet the following conditions are selected: One end of the pipe segment is located inside the current DMA, and the other end is located outside the current DMA or belongs to the skeleton network node marked in step S12; This segment was not marked as a boundary segment in the initial DMA boundary segment list in step S13.

[0061] Pipe segments meeting the above conditions are included in the list of suspected hidden connections, and their unique identifiers, start and end node coordinates, and pipe diameter information are marked for on-site verification. After the suspected hidden connections are confirmed to have been cut off or valves installed on-site, the pipeline network model constructed in step S11 is updated, and the newly added boundary pipe segments or adjusted boundary states resulting from the repair of hidden connections are synchronously updated to the initial DMA boundary pipe segment list, thereby forming a revised boundary pipe segment list (i.e., a list updated based on the results of the hidden connection verification of the initial DMA boundary pipe segment list) which is then passed to step S23.

[0062] It should be noted that the updates to the network diagram model and the initial DMA boundary segment list include: Pipeline network model update: For confirmed hidden connecting pipe sections, if truncation measures are taken, mark the pipe section as "truncation" in its attributes, and treat it as disconnected during hydraulic simulation; if valves are installed, update the valve status identifier for the pipe section. The status is set to "normally closed" or "closeable," and the connection relationship is preserved in the node-segment topology. Initial DMA boundary segment list update: If a hidden connected segment confirmed on-site is not marked as a boundary segment in the initial DMA boundary segment list generated in step S13, then the segment is added as a boundary segment, and its boundary type is marked as "hidden connected supplementary boundary" in the list, while its associated DMA number is recorded; if the segment has been marked as a boundary segment in the initial DMA boundary segment list, but on-site verification finds that its original valve is invalid or no valve is installed, then the segment's record in the initial DMA boundary segment list is maintained, its "current valve status" field is updated to "valve installed" or "valve repaired," and marked "confirmed closed by hidden connected verification."

[0063] Understandably, the above steps simulate the theoretical pressure decay process under ideal closed conditions by constructing a pressure-driven node flow hydraulic model based on the measured pressure decay curves of each DMA, and inversely locate the hidden connecting pipe segments based on the residual analysis of measured and theoretical data and pressure gradient clustering.

[0064] S23. Boundary hydraulic decoupling verification: Using the corrected boundary pipe segment list and SCADA historical pressure data, calculate the pressure correlation coefficient between any two nodes and construct a node pressure similarity matrix; identify node pairs with correlation coefficients on both sides of the boundary that are higher than a preset threshold, determine them as hydraulic coupling areas, mark the corresponding boundary pipe segments as boundaries to be adjusted and move them outward in the direction of low correlation to correct them, and generate a verification boundary pipe segment list.

[0065] Specifically, the boundary hydraulic decoupling verification includes the following steps: Historical SCADA pressure data was obtained as the basis for analysis, and node pressure time series data of the past 30 consecutive days with a sampling frequency of not less than 15 minutes were selected. For any two nodes in the pipeline diagram model and Extract its pressure time-series vector and calculate the Pearson correlation coefficient. As a quantitative indicator of the synchronicity of pressure fluctuations between two nodes, a node pressure similarity matrix is ​​then constructed. ; For each DMA boundary segment in the revised boundary segment list Extracting boundary pipe segments from the nodal pressure similarity matrix Two-end nodes and correlation coefficient ;like Greater than or equal to the preset threshold If so, the boundary pipe segment is marked as a hydraulic coupling boundary and added to the list of boundaries to be adjusted. In this embodiment... .

[0066] For boundary adjustments, an outward shift correction is performed, starting from the current boundary segment and searching in the adjacent direction until the correlation coefficient first drops below a low threshold. The location of the pipe section. In this embodiment... Move the DMA boundary from the current pipe segment to the pipe segment with low correlation coefficient, update the corrected boundary pipe segment list, and generate the verification boundary pipe segment list.

[0067] Perform outward relocation corrections, for example: for each boundary pipe segment to be adjusted. Perform the following boundary shift correction procedure: Starting from the current boundary segment, proceed to its two endpoints respectively. (Internal to DMA) and (DMA external or backbone network node) adjacent segment direction search. For each adjacent segment found. or Query the correlation coefficient between its two endpoints. Find the point where the correlation coefficient first drops below a preset low threshold. The location of the pipe section. In this embodiment... Move the DMA boundary from the current segment. The boundary pipe segment is moved outward to the low-correlation-coefficient pipe segment; that is, the new boundary pipe segment is a segment with a correlation coefficient below 0.4. If no segment with a correlation coefficient below 0.4 is found within the preset search depth (e.g., expanding outward by 3 pipe segments), the segment with the lowest correlation coefficient on the search path is selected as the new boundary location and marked in the correction report. Update the corrected boundary pipe segment list: the original boundary pipe segment... Remove from the boundary list and add the newly identified boundary segment to the boundary list. After the above relocation correction, the pressure correlation coefficients of the nodes on both sides of the new DMA boundary are both below the threshold. This ensures that each DMA is not only physically isolated, but also that the hydraulic fluctuations under normal operating conditions are independent of each other, thus guaranteeing the accuracy of subsequent nighttime minimum flow analysis.

[0068] Based on the identification and correction of hidden connecting pipe sections, the above steps utilize historical SCADA pressure data to quantitatively assess the degree of hydraulic coupling on both sides of each DMA boundary, identifying "hydraulic false closure" areas caused by synchronous pressure fluctuations. Through this process, a secondary verification of "hydraulic decoupling" of the DMA boundary is achieved on the basis of "physical closure," ensuring that internal pressure fluctuations of each DMA are not affected by external pipe networks, thus improving the accuracy of leakage measurement analysis. The list of verified boundary pipe sections is used as the output of the second stage and passed to the third stage for redundancy constraints and multi-objective optimization.

[0069] S3. Redundancy Constraints and Multi-Objective Optimization: Candidate valve positions are determined based on the check boundary pipe section list for initial redundancy screening. With metering coverage, reliability loss, and water age increment as objectives, the NSGA-III algorithm is used to solve for the optimal boundary valve closure and flow meter layout scheme. Figure 4 As shown, it includes the following steps: S31. Initial screening of redundancy in the valve candidate set: Based on the list of verification boundary pipe sections, determine the positions of candidate valves, use the DIJKSTRA algorithm to calculate the increment of the shortest hydraulic path length for fire water supply after each valve is closed, and remove valves whose increment exceeds the preset threshold or are located on non-closable pipe sections of the skeleton network to obtain a safety valve candidate set.

[0070] Specifically, the initial screening of redundancy in the valve candidate set includes the following steps: Based on the list of verified boundary pipe sections, determine the candidate valve locations on each boundary pipe section: if the pipe section already has an operable valve, then that valve is used as a candidate; if there is no valve, then the location of the planned new valve is used as a candidate; if the pipe section belongs to the skeleton network, then no candidate valves are generated for that pipe section. For each candidate valve The DIJKSTRA algorithm is used, with edge weights... For path cost, calculate the shortest hydraulic path length from each water source node to each municipal fire hydrant node before and after valve closure. and Then calculate the relative increment. ; It should be noted that the DIJKSTRA algorithm is used to calculate the water source To the fire hydrant Shortest hydraulic path length : ; in, To obtain water source To the fire hydrant The set of all possible paths, For one of the paths, The pipe segments traversed by the path. This represents the edge weight of the pipe segment.

[0071] In the pipeline diagram model In the middle, the candidate valves The status of the pipe segment is set to "disconnected" (i.e., from the pipe segment set). (Temporarily remove the pipe section) to obtain the modified graphical model. Starting from each water source node and ending at each fire hydrant node, the DIJKSTRA algorithm is executed again to calculate the valve closure. Later from the water source To the fire hydrant Shortest hydraulic path length .

[0072] If the relative increase in path length for any water source-fire hydrant combination exceeds a preset redundancy threshold... (In this embodiment, 30% is used), or if the pipe segment where the candidate valve is located has been marked as a non-closable pipe segment in step S12, then the valve is removed from the candidate set; otherwise, it is retained.

[0073] After initial screening, the retained valves constitute the candidate set of safety valves. The list of boundary pipe sections to be verified is then passed to step S32 as input feasible solution space for the multi-objective optimization model.

[0074] The above steps, based on the verification of the boundary pipe segment list, perform initial screening of candidate valves on the boundary pipe segments using redundancy constraints, eliminating valves that may significantly weaken the reliability of the pipe network's water supply. This initial redundancy screening effectively reduces the search scale of subsequent multi-objective optimization and ensures that the redundancy of the fire water supply path remains within a safe range after the boundary valves are closed.

[0075] S32. Multi-objective optimization of boundary valves and flow meters: With the objectives of maximizing metering coverage, minimizing hydraulic reliability loss, and minimizing water age increment, the NSGA-III algorithm is used to solve the Pareto optimal solution set under the constraints of fire head, prohibited installation of skeleton network valves, and inlet pressure fluctuation, to obtain the optimal boundary valve closure list and flow meter installation location scheme.

[0076] Specifically, the multi-objective optimization of boundary valves and flow meters includes the following steps: Define decision variables, including: Valve state variables This indicates whether each valve in the candidate set of safety valves is normally closed. Flow meter installation variables This indicates whether a flow meter is installed in each DMA inlet pipe section.

[0077] Understandably, Corresponding safety valve candidate set Each candidate valve in , This indicates that the valve has been selected as a normally closed valve in the DMA boundary scheme; This indicates that the valve remains open or has not been selected as a boundary isolation facility. The corresponding verification boundary pipe segment list includes all pipe segments connecting the skeleton network nodes and the internal nodes of the DMA (i.e., DMA entry pipe segments). This indicates that a flow meter is installed on this pipe section; This indicates that a flow meter will not be installed on this pipe section.

[0078] Three optimization objectives are defined, including: Maximize meter coverage, which is the ratio of DMA water consumption covered by the flow meter to the total water consumption; Minimize hydraulic reliability loss, i.e., the percentage decrease in minimum cut set capacity of the skeleton network after closing the boundary valves; Minimize the average water age increment of DMA exit nodes, which is the weighted average increase in water age of each DMA exit node compared to its original state.

[0079] Furthermore, metering coverage is represented by the percentage of water consumption covered by the selected DMA inlet pipe section where the flow meter is installed: ; in, This represents the total number of DMA inlet pipe segments. Install variables for the flow meter. For the first The average daily water consumption of each DMA inlet pipe segment served by the DMA (obtained from the basic information of each DMA in step S13). For nodes Design water demand, This represents the total number of nodes in the pipeline network model. The higher this target value, the greater the proportion of water usage covered by metering facilities, and the more complete the basic data for leakage analysis.

[0080] Hydraulic reliability loss is defined as the percentage decrease in the minimum cut set capacity of the skeleton network relative to its original state after closing selected boundary valves. The calculation process includes: With skeleton network node set Using the water source nodes as the source nodes and the remaining skeleton network nodes as the sink nodes, the minimum cut set capacity of the skeleton network is calculated using the maximum flow algorithm. Based on valve state variables Update graph model: All The pipe section containing the valve is set to the "disconnected" state, resulting in the modified graphical model. ;exist Recalculate the minimum cut set capacity of the skeleton network Define the reliability loss objective function: ; The smaller the target value, the less the disruption to the redundant connectivity of the skeleton network caused by closing the boundary valves.

[0081] Water age refers to the time it takes for water to travel from the water source into the pipe network to reach a certain node, and it is an important indicator for evaluating the water quality of the pipe network. After closing the boundary valves and changing the water flow path, some areas within the DMA (Diverterless Water Flow) may experience an increase in water age and a risk of water quality deterioration. The calculation process includes: In the original pipe network diagram model In this study, the average water age of each DMA outlet node (i.e., the node on the DMA boundary pipe segment closest to the DMA interior) was calculated using the EPANET water quality simulation module. ; In valve state variables Graphical model after application In the next step, the water quality simulation is re-executed to calculate the average water age of each DMA outlet node. The objective function for water age increment is defined as the weighted average of the water age increments at each DMA exit node: ; in, For the first The weight of each DMA can be taken as the proportion of water consumption of that DMA; The function ensures that the penalty is only applied as the water age increases.

[0082] Set constraints: the free head at the most unfavorable point under fire verification conditions shall not be less than 10 meters; no normally closed valves shall be installed in the backbone network section; the standard deviation of the inlet pressure fluctuation during the minimum flow period at night for each DMA shall not exceed 2 meters of water column.

[0083] It should be noted that, in the valve state variable Graphical model after application In accordance with fire protection regulations, simulated fire-fighting conditions are performed: an additional fire-fighting flow rate (e.g., 15 L / s) is applied at the most unfavorable point of each DMA (usually the highest point or the end node of the pipeline network), and the free head at that node is calculated. .Require If the free head at the most unfavorable point of any DMA under a certain scheme is less than 10m, then the scheme is considered infeasible. For any candidate valve If the pipe segment it belongs to belongs to the set of skeleton network pipe segments marked in step S12, Then, mandatory constraints are applied: This constraint has been guaranteed by the initial screening in step S31, and is here written as an explicit constraint in the optimization model. (Regarding the valve state variables...) Graphical model after application In the simulation of minimum flow conditions at night, the standard deviation of the inlet pressure of each DMA was calculated over several consecutive days. .Require: This constraint ensures stable DMA inlet pressure and avoids affecting the flow meter's measurement accuracy due to excessive pressure fluctuations.

[0084] The NSGA-III algorithm is used to solve the above multi-objective optimization model: the population size, crossover and mutation probability and maximum number of iterations are set, and the Pareto optimal frontier is searched through non-dominated sorting and reference point-based selection strategy, and a set of non-dominated solutions is output.

[0085] Furthermore, the execution flow of the NSGA-III algorithm is as follows: ① Population initialization: Randomly generated Each individual is encoded as a vector of decision variables. ,in The number of candidate valves, This represents the number of DMA entry segments.

[0086] ② Fitness evaluation: For each individual, update the graphical model based on the decision variables and calculate the values ​​of the three objective functions. It also checks for constraint violations. Individuals who violate the constraints are punished by having their non-dominant ranking rank lowered.

[0087] ③ Non-dominated ranking and reference point association: Individuals in the population are stratified according to Pareto dominance, and individuals with higher non-dominated ranks are selected to enter the next generation. For individuals in the critical stratum, a reference point-based selection strategy is used to maintain population diversity.

[0088] ④ Genetic operations: Offspring individuals are generated by simulating binary crossover and polynomial mutation, with crossover probability... Probability of mutation ( (The dimension of the decision variables).

[0089] ⑤ Iteration Termination: Repeat steps ② to ④ until the maximum number of iterations is reached. (In this example, 500 generations are used) or Pareto front convergence.

[0090] ⑥ Pareto optimal solution set output: When the algorithm terminates, output all individuals in the first non-dominated level, which constitute the Pareto optimal solution set.

[0091] A satisfactory solution is selected from the Pareto optimal solution set, and the optimal boundary valve closure list and flow meter installation location scheme are decoded and passed to step S41 along with the Pareto frontier data.

[0092] Furthermore, the optimal boundary valve closure list: lists all Valve identifier, pipe section, location, and closing method (normally closed / adjustable). Flow meter installation location plan: List all The DMA entry pipe segment.

[0093] Based on the candidate set of safety valves and the list of verification boundary pipe sections, a multi-objective optimization solution was performed on the boundary valve closing scheme and the flow meter installation location under multiple constraints. This achieved the synergistic optimization between metering accuracy, hydraulic reliability and water quality safety, providing a scientific basis for the final determination of the DMA hierarchical zoning scheme.

[0094] S4. Zero-pressure retest and finalization: Perform a second zero-pressure test based on the optimal boundary valve closure list. Determine the effectiveness of the optimization by the zero-pressure time ratio and the percentage increase in the decay rate. If effective, finalize the decision; otherwise, backtrack to the implicit connectivity inversion iteration for correction.

[0095] Specifically, the zero-pressure retest finalization process includes the following steps: Based on the optimal boundary valve closure list, the valve operation for the second zero-pressure test was performed. Under the same test period and monitoring point layout as the first zero-pressure field test, the second zero-pressure test was performed to obtain the optimized measured pressure decay curves of each DMA.

[0096] For each monitoring point of each DMA, the pressure return-to-zero time ratio is calculated separately. and percentage increase in decay rate Among them, the zeroing time ratio is defined as the ratio of the optimized zeroing time to the zeroing time of the first test; the percentage increase in decay rate is defined as the increase in the average decay rate after optimization relative to the first test.

[0097] The optimization effectiveness judgment criteria are set as follows: if the overall zeroing time ratio of a certain DMA is less than or equal to the effective threshold one (0.5 in this embodiment), or the overall decay rate improvement percentage is greater than or equal to the effective threshold two (50% in this embodiment), then the DMA optimization is determined to be effective.

[0098] If all DMAs are deemed valid, a verification report and the final solution are output. If an invalid DMA exists, it is marked as needing to be reviewed and the process returns to step S22 (implicit connectivity inversion). The second test data is used for iterative correction until all DMAs pass verification.

[0099] Furthermore, for DMAs that need to be reviewed, the implicit connectivity inversion process is re-executed, using the pressure attenuation data from the second test as a new input to improve the sensitivity of anomaly identification. After correction, the pipeline network model and the initial DMA boundary pipe segment list are updated, and steps S23 (hydraulic decoupling verification), S31 (redundancy screening), and S32 (multi-objective optimization) are re-executed. After the revised scheme is formed, this step is executed again for review.

[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A hierarchical and zoning planning method for water supply networks based on DMA, characterized in that: Includes the following steps: Skeleton Extraction and Initial Hierarchical Delineation: Obtain pipeline network data, construct a weighted undirected graph to extract the skeleton network, remove skeleton pipe segments, perform community detection on the remaining pipeline network to generate initial DMA boundaries, and obtain the initial DMA boundary pipe segment list; Dual-mode verification and boundary inversion: Based on the initial DMA boundary pipe segment list, the first zero-pressure test is performed to obtain the measured pressure decay curve. The implicit connectivity inversion and hydraulic decoupling verification are performed in sequence to generate a verification boundary pipe segment list. Redundancy constraints and multi-objective optimization: Based on the check boundary pipe section list, candidate valve positions are determined for initial redundancy screening. With metering coverage, reliability loss and water age increment as objectives, the NSGA-III algorithm is used to solve the optimal boundary valve closure and flow meter layout scheme. Zero-pressure retest finalization: A second zero-pressure test is performed based on the optimal boundary valve closure list. The effectiveness of the optimization is determined by the zero-pressure time ratio and the percentage increase in the decay rate. If effective, the decision is finalized; otherwise, the process is backtracked to the implicit connectivity inversion iteration correction. The redundant constraints and multi-objective optimization include the following steps: Initial screening of redundancy in the valve candidate set: Based on the list of verification boundary pipe sections, the positions of candidate valves are determined. The DIJKSTRA algorithm is used to calculate the increment of the shortest hydraulic path length for fire water supply after each valve is closed. Valves with increments exceeding a preset threshold or located on non-closable pipe sections of the skeleton network are removed to obtain a candidate set of safety valves. Multi-objective optimization of boundary valves and flow meters: Define valve state variables and flow meter installation variables. With the optimization objectives of maximizing metering coverage, minimizing hydraulic reliability loss, and minimizing water age increment, the NSGA-III algorithm is used to solve the Pareto optimal solution set under the constraints of fire head, valve prohibition in the skeleton network, and inlet pressure fluctuation, to obtain the optimal boundary valve closure list and flow meter installation location scheme.

2. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 1, characterized in that: The skeleton extraction and initial hierarchical delineation include the following steps: Pipeline network model construction and attribute assignment: Obtain pipeline network GIS information data, pipe segment attribute extended data and SCADA historical operation data, abstract the pipeline network into a weighted undirected graph, and calculate the edge betweenness of each pipe segment and the degree centrality of each node as topological weights; Skeleton network identification and protection marking: Define core score, sort all pipe segments in descending order of core score, extract the top-ranked pipe segments according to a preset ratio to form skeleton network, and mark all pipe segments in skeleton network as non-closable pipe segments. Initial division of DMA metering cells: After removing the skeleton network segments, a community structure detection algorithm based on modularity optimization is executed on the remaining peripheral network. The algorithm automatically aggregates nodes with the goal of maximizing modularity, identifies the initial DMA partition boundaries, and generates an initial DMA boundary segment list.

3. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 2, characterized in that: The number of side sections of the pipe segment Based on edge weight Calculate the percentage of times each node's shortest path passes through the current pipe segment: In the formula, For nodes and The total number of shortest paths between [locations]. For the pipe section The number of shortest paths; The degree centrality of the nodes Defined as the normalized value of the number of pipe segments directly connected to a node: In the formula, For nodes The degree value, This represents the total number of nodes in the network diagram model.

4. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 3, characterized in that: The core score is defined as: For connection nodes and pipe section The core score is expressed as: ； In the formula, For pipe segment edge intermediate number The normalized value, Degree centrality of nodes at both ends of the pipe segment and The value after mean normalization; , These are the weighting coefficients; The community structure detection algorithm uses the Louvain algorithm, based on modularity. Modularity is defined as the maximization of community partitioning quality as an evaluation metric. In the formula, For the edge weights of the pipe segments, For nodes The weighting degree, For subgraph Half of the total edge weights For nodes Community affiliation number, This is an indicator function.

5. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 1, characterized in that: The dual-mode verification and boundary inversion include the following steps: First zero-pressure field test: Close the corresponding valves according to the initial DMA boundary pipe section list, set up a temporary pressure recorder to perform zero-pressure test, and collect the measured pressure decay curves of each DMA internal node; Implicit connectivity inversion: Construct a pressure-driven node flow hydraulic model to simulate the theoretical pressure decay curve under ideal closed conditions; identify pressure anomaly regions by comparing the residuals of the measured curves and the theoretical curves; use the pressure gradient anomaly clustering algorithm to invert and calculate for pressure anomaly regions, output a list of suspected implicit connectivity pipe segments, update the network diagram model after on-site verification and repair, and generate a corrected boundary pipe segment list based on the initial DMA boundary pipe segment list. Boundary hydraulic decoupling verification: Using the corrected boundary pipe segment list and SCADA historical pressure data, calculate the pressure correlation coefficient between any two nodes and construct a node pressure similarity matrix; identify node pairs with correlation coefficients on both sides of the boundary that are higher than a preset threshold, determine them as hydraulic coupling areas, mark the corresponding boundary pipe segments as boundaries to be adjusted and move them outward in the direction of low correlation to correct them, and generate a verification boundary pipe segment list.

6. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 5, characterized in that: In the pressure-driven nodal flow hydraulic model, the expression for nodal flow is: In the formula, For nodes Under current pressure The actual outflow rate is below. For nodes Design water demand under reference pressure For reference pressure, The pressure-flow relationship function is described using a piecewise function to depict the effect of pressure on the outflow.

7. A hierarchical and zoning planning method for water supply networks based on DMA according to claim 6, characterized in that: In the implicit connectivity inversion, the screening method for suspected implicit connectivity segments in abnormal areas is as follows: Construct the feature vector for each monitoring point ,in Here are the plane coordinates of the monitoring point. This is the normalized pressure gradient value; The DBSCAN algorithm is used to identify spatial clusters whose gradient values ​​are significantly different from the overall values. Suspected hidden connected regions are delineated with the geometric center of the abnormal cluster as the center and a preset search radius. Within this region, pipe segments in the pipe network diagram model that are located at one end inside the current DMA and the other end outside the current DMA or skeleton network nodes, and are not marked as boundary pipe segments in the initial DMA boundary pipe segment list, are searched to generate a list of suspected hidden connected pipe segments.

8. The hierarchical and zoning planning method for water supply networks based on DMA according to claim 1, characterized in that: The optimization objectives are defined as follows: Objective function for maximizing meter coverage: in, This represents the total number of DMA inlet pipe segments. Install variables for the flow meter. For the first The average daily water consumption of the DMA served by each DMA inlet pipe section For nodes Design water demand, This represents the total number of nodes in the network diagram model; Objective function for minimizing hydraulic reliability loss: in, This represents the minimum cut set capacity of the skeleton mesh in its original state. The minimum cut set capacity of the skeleton mesh after closing the selected boundary valve; Objective function for minimizing water age increment: in, and These represent the average water age at each DMA outlet node before and after valve closure. For the first The weight of water consumption percentage for each DMA.

9. A hierarchical and zoning planning method for water supply networks based on DMA according to claim 8, characterized in that: The NSGA-III algorithm solution includes: Population individuals are randomly generated, and each individual is encoded as a decision variable vector. ,in The number of candidate valves; For each individual, update the graphical model and compute three objective function values, and check for constraint violations; search for the Pareto optimal frontier through non-dominated ordination and a reference point-based selection strategy; Offspring individuals are generated by simulating binary crossover and polynomial mutation. The iteration continues until the maximum number of iterations or Pareto front convergence is reached. The output consists of all individuals in the first non-dominated level, forming the Pareto optimal solution set.

10. A hierarchical and zoning planning method for water supply networks based on DMA according to claim 1, characterized in that: The zero-pressure retest finalization includes: A second zero-pressure test was performed based on the optimal boundary valve closure list to obtain the optimized measured pressure decay curves for each DMA. For each monitoring point of each DMA, the pressure zeroing time ratio and the decay rate improvement percentage are calculated respectively. The zeroing time ratio is defined as the ratio of the optimized zeroing time to the zeroing time of the first test, and the decay rate improvement percentage is defined as the improvement of the optimized average decay rate relative to the first test. If the overall return-to-zero time ratio of a certain DMA is less than or equal to the effective threshold one, or the overall decay rate improvement percentage is greater than or equal to the effective threshold two, then the DMA optimization is deemed effective. If all DMAs are deemed valid, a verification report and the final solution are output. If an invalid DMA exists, it is marked as needing to be reviewed and the implicit connectivity inversion step is returned. The second test data is used for iterative correction until all DMAs pass verification.

Images