DMA modeling method for urban water supply network based on multi-dimensional data fusion

Abstract

The present application relates to the technical field of water supply network partitioning, and specifically to a DMA modeling method for an urban water supply network based on multidimensional data fusion, the method comprising: analyzing the consistency of node hydraulic data of adjacent network nodes and the stability of actual hydraulic conditions of pipelines, obtaining the hydraulic characteristic similarity of network nodes, and constructing a weighted undirected graph of the network; based on the topological flexibility of important nodes in a ring network and amplifying the difference in importance scores of important nodes in a local water supply network area, constructing a screening index, screening initial partition centers, and obtaining the number of partitions; based on the weighted undirected graph of the network, the initial partition centers and the number of partitions, combining a spectral clustering algorithm and network hydraulic simulation software to perform DMA partitioning and modeling simulation, thereby avoiding the problem of an unbalanced number of DMA partition nodes caused by the presence of multiple initial partition centers in a local water supply network area, and improving the stability of DMA partitioning by the spectral clustering algorithm.

Description

Technical Field

[0001] The present application relates to the technical field of water supply network zoning, and in particular to a DMA modeling method for urban water supply network based on multi-dimensional data fusion. Background Art

[0002] The water supply network is an important urban infrastructure. With the advancement of urbanization, urban infrastructure is constantly improving. The length and scale of urban water supply networks are becoming larger and larger, and the structure and layout are becoming more complex, which makes the operation and management of water supply networks more difficult. The independent metering district partition (District Metered Area, DMA) can divide the entire urban water supply network into multiple DMA partitions. By monitoring the inflow and outflow of each DMA partition, the leakage of each DMA partition can be identified, and the leakage point in the urban water supply network can be found to ensure the safety of urban water use and improve economic benefits.

[0003] The node clustering method is the most common DMA partitioning method, and the most commonly used one is the spectral clustering algorithm, which regards the pipeline as the edge in the graph and the pipeline network node as the node in the graph. The similarity of hydraulic characteristics between pipelines and pipeline network nodes is used for clustering to complete DMA partitioning. However, due to the complex distribution of pipeline network characteristics, the traditional spectral clustering algorithm randomly selects the initial partition center during the DMA partitioning process, which is prone to the phenomenon that the initial partition center does not match the distribution of pipeline network characteristics, which leads to poor stability of DMA partitioning results and makes it difficult to identify the leakage location in the urban water supply pipeline network. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a DMA modeling method for urban water supply network based on multi-dimensional data fusion to solve the existing problems.

[0005] The DMA modeling method for urban water supply network based on multi-dimensional data fusion in this application adopts the following technical solutions:

[0006] An embodiment of the present application provides a DMA modeling method for an urban water supply network based on multi-dimensional data fusion, the method comprising the following steps:

[0007] Obtain a water supply network topology diagram of the urban water supply network; obtain various pipeline hydraulic data of each pipeline at each time in the water supply network topology diagram and various node hydraulic data of each network node through the network hydraulic simulation software;

[0008] Based on the degree of disorder of the pipeline hydraulic data of each pipeline in the water supply network topology diagram, each pipeline is evaluated in combination with an evaluation algorithm to obtain the hydraulic condition stability of each pipeline; based on the similarity of the node hydraulic data fluctuations of the network nodes at both ends of each pipeline in the water supply network topology diagram and the hydraulic condition stability, the hydraulic characteristic similarity between the network nodes at both ends of each pipeline is obtained;

[0009] A weighted undirected graph of the pipe network is constructed based on the hydraulic characteristic similarity, and the importance score of each pipe network node in the weighted undirected graph of the pipe network is calculated to obtain the important nodes in the water supply pipe network topology graph; a cost undirected graph is constructed based on the inverse of the hydraulic characteristic similarity, and a screening index of each important node is calculated based on the minimum path length between important nodes in the cost undirected graph and the node degree of each important node in the water supply pipe network topology graph; the initial partition center is screened based on the screening index and the importance score to obtain the number of partitions;

[0010] Based on the weighted undirected graph of the pipe network, the initial partition center and the number of partitions, the spectral clustering algorithm is combined to carry out DMA partitioning of the urban water supply network. The installation positions of water meters and valves are located on the boundary pipe sections through the Dijkstra algorithm, and the DMA partitioning results and valve closing conditions are modeled and simulated through the pipe network hydraulic simulation software.

[0011] In one embodiment, the various pipeline hydraulic data of each pipeline and the various node hydraulic data of each pipeline network node include: flow rate, flow velocity, water head and water pressure of each pipeline, and water demand, total water head and pressure of each pipeline network node.

[0012] In one embodiment, the process of obtaining the hydraulic working condition stability is:

[0013] Calculate the standard deviation of all the time data of each item of pipeline hydraulic data of each pipeline in the water supply network topology diagram, and record the matrix composed of the standard deviations of all items of pipeline hydraulic data of all pipelines as the hydraulic condition stability matrix;

[0014] The hydraulic condition stability matrix is ​​used as the input of the evaluation algorithm, and the standard deviation of each pipeline hydraulic data is used as a very small index, and the output is a comprehensive evaluation score for each pipeline, which is recorded as the hydraulic condition stability of each pipeline.

[0015] In one embodiment, the process of obtaining the hydraulic characteristic similarity is:

[0016] The time series of each node hydraulic data of each pipe network node is recorded as the node hydraulic data sequence; for any pipe C, the pipe network nodes at both ends of pipe C are recorded as , , calculate the network nodes and The similarity between the hydraulic data sequences of the same item nodes is recorded as the first similarity;

[0017] Pipeline network node and The average value of all the first similarities between the nodes is taken as the and The hydraulic demand consistency between ; Pipeline network nodes and The hydraulic characteristics similarity between , The expression is: , where It's a pipeline The hydraulic stability of is a normalization function.

[0018] In one embodiment, the process of constructing the pipe network weighted undirected graph is as follows: the hydraulic characteristic similarity between the pipe network nodes at both ends of each pipe is used as the weight of the corresponding pipe in the water supply pipe network topology diagram to obtain the pipe network weighted undirected graph.

[0019] In one embodiment, the importance score is obtained by taking the weighted undirected graph of the pipe network as the input of the sorting algorithm, and outputting the importance score of each pipe network node.

[0020] In one embodiment, the important node acquisition process is: taking the importance scores of all network nodes as input of the classification algorithm, outputting two categories of network nodes, and taking all network nodes in the category of network nodes with the highest average importance score as important nodes.

[0021] In one embodiment, the cost undirected graph is constructed by taking the inverse of the hydraulic characteristic similarity between the pipe network nodes at both ends of each pipe as the cost weight of the corresponding pipe in the water supply pipe network topology diagram to obtain the cost undirected graph.

[0022] In one embodiment, the process of obtaining the screening index is:

[0023] Obtain the length of the minimum cost path between any two important nodes in the cost undirected graph; construct an adjacency matrix of all important nodes, where the inverse of the minimum cost path length between any two important nodes is the value of the corresponding parameter in the adjacency matrix; use the adjacency matrix as the input of the clustering algorithm, and output each cluster; record the screening index of the ath important node in the kth cluster as , The expression is:

[0024] , where is the difference between the node degrees of the ath and rth important nodes in the kth cluster, is the difference between the importance scores of the ath and rth important nodes in the kth cluster, is the total number of important nodes in the kth cluster, is a normalization function.

[0025] In one embodiment, the process of screening the initial partition centers and obtaining the number of partitions is as follows:

[0026] The probability of selecting the partition center of the important node x in the water supply network topology is recorded as , The expression is: , where and are the importance score and screening index of important node x in the water supply network topology graph, respectively; is the preset allocation weight coefficient;

[0027] The partition center selection probability of all important nodes is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. The important nodes whose partition center selection probability is greater than the segmentation threshold are used as the initial partition center; the number of all initial partition centers is counted as the number of partitions.

[0028] This application has at least the following beneficial effects:

[0029] The present application analyzes the consistency of the node hydraulic data of adjacent pipe network nodes and the stability of the actual hydraulic working conditions of the pipelines, and obtains the similarity of the hydraulic characteristics of the pipe network nodes. Its beneficial effect is that all the pipes in the entire water supply network topology diagram are considered to avoid the problem that the actual water supply working conditions of the urban water supply network itself have obvious peak and valley characteristics, and the actual hydraulic working condition stability of a single pipe is relatively low, thereby improving the accuracy of evaluating the similarity of the hydraulic characteristics of the pipe network nodes; based on the topological flexibility of important nodes in the ring network and amplifying the differences in the importance scores of each important node in the local water supply network area, a screening index is constructed to correct the importance scores of important nodes, thereby avoiding the problem of multiple initial partition centers in the local water supply network area, resulting in an unbalanced number of DMA partition nodes; through the topological structure and actual hydraulic characteristics analysis of the urban water supply network, the initial partition center of the spectral clustering algorithm is adaptively obtained, the stability of the spectral clustering algorithm for DMA partitioning is improved, and it is helpful for the urban water supply network to identify the location of leaks. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 A flow chart of the DMA modeling method for urban water supply network based on multi-dimensional data fusion provided in this application;

[0032] Figure 2 The water supply network topology diagram of the urban water supply network;

[0033] Figure 3 It is a weighted undirected graph of the urban water supply network;

[0034] Figure 4 This is a local water supply network topology diagram of a ring-shaped urban water supply network. DETAILED DESCRIPTION

[0035] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the DMA modeling method for urban water supply network based on multidimensional data fusion proposed by the present application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0036] Unless defined otherwise, 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 application belongs.

[0037] The specific scheme of the urban water supply network DMA modeling method based on multi-dimensional data fusion provided by the present application is described in detail below with reference to the accompanying drawings.

[0038] An embodiment of the present application provides a DMA modeling method for an urban water supply network based on multi-dimensional data fusion.

[0039] Specifically, the following DMA modeling method for urban water supply network based on multi-dimensional data fusion is provided. Figure 1 , the method comprises the following steps:

[0040] Step S1, obtaining a water supply network topology diagram of the urban water supply network; obtaining various pipeline hydraulic data of each pipeline at each moment and various node hydraulic data of each network node in the water supply network topology diagram through the network hydraulic simulation software.

[0041] Get the water supply network topology diagram of the city water supply network, such as Figure 2 As shown, Figure 2Each edge represents each pipeline, and each node represents each network node. The modeling-related data of the urban water supply network are obtained through the water company, including the length, diameter, material, and pipeline number data of each pipeline, the elevation, pressure, and node number data of each node, and the number, location, and flow data of each water source. The modeling-related data of the pipeline and the node are input into the EPANET software to draw the urban water supply network, and then the hydraulic simulation analysis is performed in the MATLAB programming environment through the EPANET-MATLAB Toolkit to obtain the pipeline hydraulic data of each pipeline at each hour of the day, including flow, flow velocity, head, and water pressure, and obtain the node hydraulic data of each network node at each hour of the day, including water demand, total head, and pressure, wherein the hourly time is 1:00, 2:00, 3:00...22:00, 23:00, 24:00.

[0042] Each pipeline hydraulic data of each pipeline and each node hydraulic data of each pipe network node are normalized by Z-score normalization algorithm to eliminate the dimensional influence of each pipeline hydraulic data and each node hydraulic data. The Z-score normalization algorithm is a well-known technology, and the specific process is not repeated here.

[0043] It should be noted that for the normalization of various pipeline hydraulic data and various node hydraulic data, this application only provides one normalization method. There are many existing normalization methods, and implementers can also use other normalization algorithms to normalize various pipeline hydraulic data and various node hydraulic data. This application does not make specific restrictions.

[0044] Arrange any item of pipeline hydraulic data of any pipeline in ascending order of time, and the resulting sequence is recorded as the pipeline hydraulic data sequence of that item of pipeline hydraulic data of the pipeline; similarly, arrange any item of node hydraulic data of any pipeline network node in ascending order of time, and the resulting sequence is recorded as the node hydraulic data sequence of that item of node hydraulic data of the pipeline network node.

[0045] Step S2, based on the degree of confusion of the pipeline hydraulic data of each pipeline in the water supply network topology diagram, each pipeline is evaluated in combination with an evaluation algorithm to obtain the hydraulic condition stability of each pipeline; based on the similarity of the node hydraulic data fluctuations of the pipeline network nodes at both ends of each pipeline in the water supply network topology diagram and the hydraulic condition stability, the hydraulic characteristic similarity between the pipeline network nodes at both ends of each pipeline is obtained.

[0046] By zoning the network nodes in the urban water supply network through DMA, the entire urban water supply network can be divided into multiple water supply sub-areas, which helps to shorten the scope and time of discovering water leaks. Without or with little help from other DMA partitions, the network nodes and pipelines in the DMA partition should have good water balance and pressure control functions. Therefore, the DMA zoning of the urban water supply network is extremely dependent on the network structure and hydraulic characteristics. Two adjacent network nodes belong to the same DMA partition and should have similar hydraulic characteristics.

[0047] (1) In the actual operation of the urban water supply network, the stronger the consistency of the node hydraulic data of two adjacent network nodes, the more similar the water use pattern is, and the more helpful it is for the DMA partition to control the water supply in the area more accurately. Therefore, based on the similarity between the node hydraulic data sequences of the network nodes at both ends of any pipeline, the hydraulic demand consistency between the network nodes at both ends of any pipeline is constructed, specifically:

[0048] For any pipeline C, the network nodes at both ends of the pipeline are recorded as , , calculate the network nodes and The similarity between the hydraulic data sequences of the same item nodes is recorded as the first similarity;

[0049] Pipeline network node and The average value of the first similarity of the hydraulic data of all nodes is taken as the network node at both ends of the pipeline and Consistency of hydraulic demands between.

[0050] Preferably, in one embodiment of the present application, the first similarity can be and The dtw distance between the hydraulic data series of the same item node is a positive number that is preset manually. The reciprocal of the sum of The purpose is to prevent the denominator from being 0. The value of is set to 0.01. As other embodiments of the present application, the implementer can set it according to the actual situation. The value of .

[0051] In other embodiments of the present application, the implementer may also use other similarity algorithms, such as the Pearson correlation coefficient, to calculate the first similarity.

[0052] Pipeline network node and The greater the first similarity between them, the more likely it is that the two pipe network nodes have similar water use patterns during the actual operation of the urban water supply network, and the greater the consistency of hydraulic demand between the two pipe network nodes.

[0053] (2) In the water supply network topology diagram of the urban water supply network, each pipeline connects two network nodes. The similarity of hydraulic characteristics between network nodes depends not only on the consistency of hydraulic demand, but also on the actual hydraulic conditions of the pipeline. The more unstable the hydraulic conditions of the pipeline, the more difficult it is to perform hydraulic control through DMA partitioning. At the same time, since the actual water supply conditions of the urban water supply network have obvious peak and valley characteristics, the stability of the pipeline hydraulic characteristics should consider all the pipelines in the entire water supply network topology diagram. Therefore, based on the degree of chaos of the data in each pipeline hydraulic data sequence of each pipeline, combined with the evaluation algorithm, the comprehensive evaluation score of each pipeline is calculated, which is recorded as the hydraulic condition stability of each pipeline, specifically:

[0054] For any pipeline in the water supply network topology diagram, the standard deviation of all data in the pipeline hydraulic data sequence of the pipeline is calculated, and the vector composed of the standard deviation of all pipeline hydraulic data sequences of the pipeline is recorded as the hydraulic condition stability vector of the pipeline; the matrix composed of the hydraulic condition stability vectors of all pipelines in the water supply network topology diagram is recorded as the hydraulic condition stability matrix of the water supply network, wherein the hydraulic condition stability vector of each pipeline is used as a row vector, and the vector composed of the hydraulic condition stability of all pipelines at each moment is used as a column vector;

[0055] The hydraulic working condition stability matrix is ​​used as the input of the TOPSIS algorithm based on the entropy weight method, and the standard deviation of each pipeline hydraulic data sequence of each pipeline is used as a very small index. The entropy weight method is used to obtain the weight of each index, and the output is a comprehensive evaluation score for each pipeline, which is recorded as the hydraulic working condition stability of each pipeline, reflecting the actual hydraulic working condition stability of each pipeline relative to all pipelines in the entire water supply network topology diagram. Among them, the TOPSIS algorithm based on the entropy weight method is a well-known technology, and the specific process will not be repeated.

[0056] It should be noted that for the weight setting of each indicator and the evaluation of each pipeline, this application only provides a weighting method and an evaluation method. There are many existing weighting methods and evaluation methods. Implementers can also use other weighting algorithms and evaluation algorithms to set the weight of each indicator and evaluate each pipeline respectively. This application does not make specific restrictions.

[0057] (3) Based on the consistency of hydraulic demand between the network nodes at both ends of each pipeline and the hydraulic condition stability of each pipeline, the hydraulic characteristic similarity between the network nodes at both ends of each pipeline is determined, and the expression is:

[0058] , where For pipe network nodes and The hydraulic characteristics of It's a pipeline The hydraulic stability of It is a pipe network node and The consistency of hydraulic demand between is a normalization function.

[0059] The hydraulic characteristic similarity reflects the similarity of the actual hydraulic working condition characteristics of the pipe network nodes at both ends of the pipeline. The larger the value, the more stable the pipeline can maintain its hydraulic characteristics when facing the peak change of hydraulic conditions, and the more hydraulic control can be performed through DMA partitions. The more likely the two network nodes should be divided into the same DMA partition. The bigger; when The larger the value is, the more likely it is that the two network nodes have similar water use patterns in the actual operation of the urban water supply network, and the more likely they should be divided into the same DMA partition. The bigger.

[0060] Step S3, constructing a weighted undirected graph of the pipe network based on the similarity of the hydraulic characteristics, calculating the importance score of each pipe network node in the weighted undirected graph of the pipe network, and obtaining the important nodes in the water supply pipe network topology graph; constructing a cost undirected graph based on the inverse of the hydraulic characteristic similarity, and calculating the screening index of each important node based on the minimum path length between the important nodes in the cost undirected graph and the node degree of each important node in the water supply pipe network topology graph; screening the initial partition center based on the screening index and the importance score to obtain the number of partitions.

[0061] Due to the complex distribution of pipe network characteristics, the initial partition center of the spectral clustering algorithm is difficult to adaptively obtain. The hydraulic characteristics of the initial partition center directly affect other pipe network nodes in the water supply network topology diagram and have extremely high hydraulic characteristics importance.

[0062] (1) Based on the similarity of hydraulic characteristics between the network nodes at both ends of each pipeline in the water supply network topology diagram, combined with the sorting algorithm, the importance score of each network node is calculated, specifically:

[0063] The hydraulic characteristics similarity between the network nodes at both ends of each pipeline is used as the weight of the corresponding edge in the water supply network topology graph to construct a weighted undirected graph of the urban water supply network, such as Figure 3As shown. The weighted undirected graph of the urban water supply network is used as the input of the TextRank algorithm, wherein the damping factor of the TextRank algorithm is set, preferably, in one embodiment of the present application, the damping factor is set to 0.85. As other embodiments of the present application, the implementer can set the damping factor according to the actual situation. The output is the importance score of each network node. Among them, the TextRank algorithm is a well-known technology, and the specific process will not be repeated.

[0064] It should be noted that for the calculation of the importance score of each pipeline network node, this application only provides a sorting method. There are many existing sorting methods, and implementers can also use other sorting algorithms to calculate the importance score of each pipeline network node. This application does not make specific restrictions.

[0065] (2) Based on the importance scores of all pipeline network nodes and combined with the classification algorithm, the important nodes among all pipeline network nodes are obtained, specifically:

[0066] The importance scores of all network nodes are used as input of the cross-validation method, and two types of network nodes are output, and all network nodes in the type of network nodes with the highest average importance scores are used as important nodes. The cross-validation method is a well-known technology, and the specific process is not repeated here.

[0067] It should be noted that for the classification of pipeline network nodes, this application only provides one classification method. There are many existing classification methods, and implementers can also use other classification algorithms to classify pipeline network nodes. This application does not make specific restrictions.

[0068] (3) Based on the inverse of the hydraulic characteristics similarity between the network nodes at both ends of each pipeline in the water supply network topology diagram, combined with the minimum path algorithm, the minimum cost path and the minimum cost path length between any two important nodes are obtained; based on the inverse of the minimum cost path length, combined with the clustering algorithm, each clustering cluster is obtained, specifically:

[0069] In the ring-shaped urban water supply network, the network nodes are connected to form a closed ring structure, which is helpful to adjust the hydraulic characteristics. Local water supply network areas are prone to have similar hydraulic characteristics, which is very likely to cause the dense distribution of important nodes in the local water supply network area. The hydraulic conditions of the local water supply network area are stable and high, and there is no need to perform joint hydraulic control through multiple DMA partitions. Therefore, this application corrects the importance scores of important nodes through the topological relationship between important nodes to avoid the problem of multiple initial partition centers in the local water supply network area, resulting in an unbalanced number of DMA partition nodes.

[0070] First, the inverse of the hydraulic characteristics similarity between the network nodes at both ends of each pipeline is used as the cost weight of the corresponding edge in the water supply network topology graph to construct the cost undirected graph of the urban water supply network. The larger the cost weight, the less similar the node hydraulic characteristics of the network nodes at both ends of the pipeline are, and the less likely they should be divided into the same DMA partition.

[0071] Then, the cost undirected graph is used as the input of Dijkstra algorithm, and the output is the minimum cost path and minimum cost path length between any two important nodes. The smaller the minimum cost path length, the less it is necessary for the two important nodes to be jointly hydraulically controlled through multiple DMA partitions.

[0072] Finally, an adjacency matrix is ​​constructed for all important nodes, in which the inverse of the minimum cost path length between any two important nodes is used as the value of the corresponding parameter in the adjacency matrix, and the adjacency matrix is ​​used as the input of the Markov clustering algorithm, and the output is a cluster of K important nodes.

[0073] Among them, Dijkstra algorithm and Markov clustering algorithm are both well-known calculations, and the specific process will not be repeated. It should be noted that for the acquisition of the minimum cost path between any two important nodes and the clustering of important nodes, this application only provides a method for finding the minimum path and a clustering method. There are many existing methods for finding the minimum path and clustering methods. Implementers can also use other algorithms for finding the minimum path to obtain the minimum cost path between any two important nodes, and use other clustering algorithms to cluster important nodes. This application does not make specific restrictions.

[0074] (4) There are multiple important nodes in the cluster. These important nodes may be concentrated in a local water supply network area. In order to avoid the existence of multiple initial partition centers in the local water supply network area, the important nodes should be screened and the important nodes with great influence on DMA partition should be retained. Moreover, in the ring-shaped urban water supply network, when a pipeline fails, the network nodes need to redistribute the flow. The network nodes connecting different pipelines have higher flow distribution flexibility. Therefore, the important nodes connecting different pipelines should also have higher topological flexibility. Among them, the local water supply network topology diagram of the ring-shaped urban water supply network is as follows: Figure 4 shown.

[0075] First, the node degree of each important node in the water supply network topology is calculated as the topological flexibility of each important node. The greater the topological flexibility, the higher the flexibility of the important node in flow distribution, and the more it should be selected as the initial partition center of the DMA partition. The calculation of the node degree is a well-known technology, and the specific process will not be repeated here.

[0076] Then, based on the differences between the topological flexibility of different important nodes in each cluster and the differences between the importance scores, the screening index of each important node in each cluster is obtained, and the expression is:

[0077] , where is the screening index of the ath important node in the kth cluster, is the difference between the topological flexibility of the ath important node in the kth cluster and the topological flexibility of the rth important node, is the difference between the importance score of the ath important node and the importance score of the rth important node in the kth cluster. is the total number of important nodes in the kth cluster, This means that there is more than one important node in the k-th cluster. This means that there is only one important node in the k-th cluster. is a normalization function.

[0078] when hour, The larger the value, the higher the flexibility of the a-th important node in the topology when distributing traffic. The larger the value is, the more important the a-th important node is in the local water supply network area, and the more it should be selected as the initial partition center of the DMA partition. The closer it is to 1. Calculate The beneficial effect is to amplify the difference in importance scores of important nodes in the local water supply network area and avoid the existence of multiple initial partition centers in the local water supply network area.

[0079] when When , there is only one important node in the kth cluster, that is, the more likely it is that the important node is the only important node in the local water supply network area, the less likely it is that there are multiple initial partition centers in the local water supply network area. The closer to 1.

[0080] (6) Based on the screening index and importance score of each important node, the probability of selecting the partition center of each important node is calculated, and the expression is:

[0081] , where is the probability of selecting the partition center of the important node x in the water supply network topology; and are the importance score and screening index of important node x in the water supply network topology graph, respectively; is a preset allocation weight coefficient, which is used to give a greater weight to the importance score to avoid a too large range of screening index values. Preferably, in the embodiment of the present application, The value of is set to 0.8. As other embodiments of the present application, the implementer can set it according to the actual situation. The value of .

[0082] when The larger the value, the higher the criticality of the important node in the entire water supply network topology. The larger the value is, the more suitable the important node is as the DMA partition center in the local water supply network area, which is used to avoid the existence of multiple initial partition centers in the local water supply network area. The higher the probability of selecting the important node x as the initial partition center, The bigger.

[0083] (7) The partition center selection probability of all important nodes is used as the input of the Otsu method, and the output is the segmentation threshold. The important nodes whose partition center selection probability is greater than the segmentation threshold are used as the initial partition center. The number of all initial partition centers is counted as the number of partitions. Among them, the Otsu method is a well-known technology, and the specific process will not be described in detail.

[0084] It should be noted that for obtaining the segmentation threshold, this application only provides a threshold segmentation method. There are many existing threshold segmentation methods, and implementers can also use other threshold segmentation algorithms to obtain the segmentation threshold. This application does not make specific restrictions.

[0085] Step S4, based on the weighted undirected graph of the network, the initial partition center and the number of partitions, combined with the spectral clustering algorithm, DMA partitioning of the urban water supply network is performed.

[0086] (1) According to the weighted undirected graph of the water supply network topology corresponding to the pipeline network, the edge weights of the pipeline network nodes at both ends of the pipeline are used as the similarity between the two pipeline network nodes. , construct the similarity matrix of the water supply network topology map, and record the similarity matrix as , where m is the total number of network nodes in the water supply network topology diagram;

[0087] (2) Based on the similarity matrix of the water supply network topology , calculate the degree matrix, the degree matrix is ​​a diagonal matrix, and the degree matrix is ​​denoted as D; wherein, the calculation of the degree matrix is ​​a well-known technology, and the specific process is not repeated here.

[0088] (3) Construct the normalized Laplace matrix of the water supply network topology , where L is the standard Laplace matrix and I is the identity matrix.

[0089] (4) Before calculation The eigenvectors of the smallest eigenvalues ​​are composed of dimensional matrix, The row vectors in the dimensional matrix are standardized to form the feature matrix F.

[0090] (5) Use the K-mediods clustering algorithm to divide the feature matrix F into Clusters, where the number of clusters in the K-mediods clustering algorithm is the number of partitions , the initial cluster centers are the centers of each initial partition. If the i-th row of the feature matrix F belongs to the p-th partition, the network nodes represented by the i-th row are divided into the p-th partition.

[0091] Among them, steps (1) to (5) in step S4 are well-known technologies.

[0092] Step S5, locating the installation positions of the water meter and the valve on the boundary pipe section by using the Dijkstra algorithm, and modeling and simulating the DMA partitioning results and the valve closing conditions by using the pipe network hydraulic simulation software.

[0093] Based on the above steps, the spectral clustering algorithm is used to realize the DMA partitioning of each pipe network node in the urban water supply network, and then the Dijkstra algorithm is used to locate the installation position of the water meter and valve on the boundary pipe segment, specifically:

[0094] (1) Obtain the partition center of each DMA partition (i.e., the cluster center output by spectral clustering) and the location of each water source point in the urban water supply network;

[0095] (2) Using the inverse of the pipeline flow as the weight, obtain the flow weight undirected graph;

[0096] (3) Using the Dijkstra algorithm, the shortest path between the center of each DMA partition and the water source point is obtained as the main water supply path of the DMA partition;

[0097] (4) Based on the DMA zoning results of each pipe network node in the urban water supply network, obtain the boundary pipe section passing through the main water supply path as the water inlet pipe section of the DMA partition, and install valves on the remaining boundary pipe sections of the DMA partition and close them. The boundary pipe section is the pipe section in the urban water supply network that connects different DMA partitions;

[0098] (5) Use EPANET-MATLAB Toolkit to model and simulate the DMA partition results and valve closing conditions, and notify the administrator to perform intelligent regulation of each valve in the DMA partition.

[0099] Among them, steps (1) to (5) in step S5 are well-known technologies.

[0100] In summary, the embodiment of the present application obtains the hydraulic characteristic similarity of the network nodes by analyzing the consistency of the node hydraulic data of adjacent network nodes and the stability of the actual hydraulic conditions of the pipelines. The beneficial effect lies in that all pipelines in the entire water supply network topology diagram are considered to avoid the problem that the actual water supply conditions of the urban water supply network itself have obvious peak and valley characteristics, and the actual hydraulic condition stability of a single pipeline is relatively low, thereby improving the accuracy of evaluating the similarity of the hydraulic characteristics of the network nodes; based on the topological flexibility of important nodes in the ring network and amplifying the difference in the importance scores of each important node in the local water supply network area, a screening index is constructed to correct the importance scores of important nodes, thereby avoiding the problem of multiple initial partition centers in the local water supply network area, resulting in an unbalanced number of DMA partition nodes; through the topological structure and actual hydraulic characteristic analysis of the urban water supply network, the initial partition center of the spectral clustering algorithm is adaptively obtained, the stability of the spectral clustering algorithm for DMA partitioning is improved, and it is helpful for the urban water supply network to identify the leakage location.

[0101] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of the present application are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0102] The various embodiments in the present application are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0103] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A DMA modeling method for urban water supply network based on multidimensional data fusion, characterized in that: The method comprises the following steps: Obtain a water supply network topology diagram of the urban water supply network; obtain various pipeline hydraulic data of each pipeline at each time in the water supply network topology diagram and various node hydraulic data of each network node through the network hydraulic simulation software; Based on the degree of disorder of the pipeline hydraulic data of each pipeline in the water supply network topology diagram, each pipeline is evaluated in combination with an evaluation algorithm to obtain the hydraulic condition stability of each pipeline; based on the similarity of the node hydraulic data fluctuations of the network nodes at both ends of each pipeline in the water supply network topology diagram and the hydraulic condition stability, the hydraulic characteristic similarity between the network nodes at both ends of each pipeline is obtained; A weighted undirected graph of the pipe network is constructed based on the hydraulic characteristic similarity, and the importance score of each pipe network node in the weighted undirected graph of the pipe network is calculated to obtain the important nodes in the water supply pipe network topology graph; a cost undirected graph is constructed based on the inverse of the hydraulic characteristic similarity, and a screening index of each important node is calculated based on the minimum path length between important nodes in the cost undirected graph and the node degree of each important node in the water supply pipe network topology graph; the initial partition center is screened based on the screening index and the importance score to obtain the number of partitions; Based on the weighted undirected graph of the pipe network, the initial partition center and the number of partitions, combined with the spectral clustering algorithm, the DMA partition of the urban water supply pipe network is carried out; the installation positions of water meters and valves are located on the boundary pipe sections through the Dijkstra algorithm, and the DMA partition results and valve closing conditions are modeled and simulated through the pipe network hydraulic simulation software; The construction process of the cost undirected graph is: taking the inverse of the hydraulic characteristic similarity between the pipe network nodes at both ends of each pipe as the cost weight of the corresponding pipe in the water supply pipe network topology diagram to obtain the cost undirected graph.

2. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The various pipeline hydraulic data of each pipeline and the various node hydraulic data of each pipeline network node include: the flow rate, flow velocity, water head and water pressure of each pipeline, and the water demand, total water head and pressure of each pipeline network node.

3. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The process of obtaining the stability of the hydraulic working condition is as follows: Calculate the standard deviation of all the time data of each item of pipeline hydraulic data of each pipeline in the water supply network topology diagram, and record the matrix composed of the standard deviations of all items of pipeline hydraulic data of all pipelines as the hydraulic condition stability matrix; The hydraulic condition stability matrix is ​​used as the input of the evaluation algorithm, and the standard deviation of each pipeline hydraulic data is used as a very small index, and the output is a comprehensive evaluation score for each pipeline, which is recorded as the hydraulic condition stability of each pipeline.

4. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The process of obtaining the hydraulic characteristics similarity is as follows: The time series of each node hydraulic data of each pipe network node is recorded as the node hydraulic data sequence; for any pipe C, the pipe network nodes at both ends of pipe C are recorded as , , calculate the network nodes and The similarity between the hydraulic data sequences of the same item nodes is recorded as the first similarity; Pipeline network node and The average value of all the first similarities between the nodes is taken as the and The hydraulic demand consistency between ; Pipeline network nodes and The hydraulic characteristics similarity between , The expression is: , where It's a pipeline The hydraulic stability of is a normalization function.

5. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The construction process of the pipe network weighted undirected graph is: the hydraulic characteristic similarity between the pipe network nodes at both ends of each pipe is used as the weight of the corresponding pipe in the water supply pipe network topology diagram to obtain the pipe network weighted undirected graph.

6. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1, characterized in that: The process of obtaining the importance score is as follows: taking the weighted undirected graph of the pipe network as the input of the sorting algorithm, and outputting the importance score of each pipe network node.

7. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The important node acquisition process is as follows: the importance scores of all pipe network nodes are used as input of the classification algorithm, two types of pipe network nodes are output, and all pipe network nodes in the type of pipe network nodes with the highest average importance score are used as important nodes.

8. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1 is characterized in that: The process of obtaining the screening index is as follows: Obtain the length of the minimum cost path between any two important nodes in the cost undirected graph; construct an adjacency matrix of all important nodes, where the inverse of the minimum cost path length between any two important nodes is the value of the corresponding parameter in the adjacency matrix; use the adjacency matrix as the input of the clustering algorithm, and output each cluster; record the screening index of the ath important node in the kth cluster as , The expression is: , where is the difference between the node degrees of the ath and rth important nodes in the kth cluster, is the difference between the importance scores of the ath and rth important nodes in the kth cluster, is the total number of important nodes in the kth cluster, is a normalization function.

9. The urban water supply network DMA modeling method based on multidimensional data fusion according to claim 1, characterized in that: The process of screening the initial partition centers and obtaining the number of partitions is as follows: The probability of selecting the partition center of the important node x in the water supply network topology is recorded as , The expression is: , where and are the importance score and screening index of important node x in the water supply network topology graph, respectively; is the preset allocation weight coefficient; The partition center selection probability of all important nodes is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. The important nodes whose partition center selection probability is greater than the segmentation threshold are used as the initial partition center; the number of all initial partition centers is counted as the number of partitions.

Images