A pipeline flow detection and early warning system and method based on data acquisition

Abstract

The present invention relates to the technical field of pipeline detection, and specifically to a pipeline flow detection and early warning system and method based on data acquisition. First, a pipeline node is obtained according to pipeline layout data, and then a node flow index is obtained according to node flow data. If it exceeds a threshold, it is an abnormal node; then the pipeline connected to the abnormal node is obtained and processed to obtain a set of segmented points of the pipeline to be detected, and then data acquisition and processing are performed to obtain a flow data matrix; the pipeline similarity between the pipeline to be detected and the normal pipeline is calculated, and the normal pipeline is selected and processed according to the pipeline similarity to obtain a flow data matrix set of the normal pipeline; then cluster analysis is performed on the set to obtain a cluster center; the distance between the flow data matrix of the pipeline to be detected and the cluster center is calculated, and if the distance is greater than the threshold, it indicates that the pipeline to be detected is an abnormal pipeline. Through the detection method proposed by the present invention, the detection and early warning efficiency of long-distance water pipelines is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline detection, and in particular to a pipeline flow detection and early warning system and method based on data acquisition. Background Art

[0002] Irrigation projects are often built in mountainous areas, hilly areas, remote rural areas and other areas with complex terrain and weak infrastructure. They face multiple challenges in project layout and construction organization, among which the problem of water shortage is particularly prominent. The characteristics of this engineering environment make the construction of the project subject to greater constraints from factors such as geological conditions, construction environment and climate change, and especially put forward higher requirements on water resource scheduling and control capabilities.

[0003] When carrying out engineering construction in areas with scarce water resources, the source of water for the project usually needs to be planned in accordance with local natural conditions and project needs. Common water sources now include surface water resources, groundwater resources, precipitation resources and artificial water supply. After obtaining the water source, diverting water to the project through pipelines is a widely used water supply method, especially in areas with complex terrain and uneven distribution of water resources. It can achieve long-distance and large-flow water transportation to meet the construction needs of mountainous areas or irrigation areas. However, long-distance transportation of water resources has problems such as low safety and uneven water demand in different regions. It is necessary to detect and warn the flow of pipelines to optimize the utilization rate of water resources in engineering construction and further increase the speed of engineering construction.

[0004] Therefore, a pipeline flow detection and early warning system and method based on data acquisition are proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a pipeline flow detection and early warning system and method based on data acquisition. First, pipeline nodes are obtained according to pipeline layout data, and then node flow indicators are obtained according to the flow data of the nodes, and it is judged whether it is an abnormal node; then the pipeline connected to the abnormal node is obtained and data acquisition and processing are performed to obtain a flow data matrix; the pipeline similarity between the pipeline to be detected and the normal pipeline is calculated, and the normal pipeline is selected and processed according to the pipeline similarity to obtain a flow data matrix set of the normal pipeline; then cluster analysis is performed on the set to obtain a cluster center; the distance between the flow data matrix of the pipeline to be detected and the cluster center is calculated, and if the distance is greater than a threshold, it indicates that the pipeline to be detected is an abnormal pipeline.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A pipeline flow detection and early warning method based on data collection, comprising:

[0008] The pipeline layout data is obtained and processed to obtain pipeline nodes, wherein the pipeline nodes include main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes; the node flow data of the pipeline nodes is obtained and processed to obtain node flow indicators. If the node flow indicator exceeds a threshold, it is an abnormal node;

[0009] Furthermore, the node traffic index includes:

[0010] Acquire and process the data of the main water pipeline node to obtain a first node flow index. If the first node flow index is greater than a preset first node flow index threshold, the main water pipeline node is the abnormal node.

[0011] Acquire and process the data of the water distribution pipeline node to obtain a second node flow index, and if the second node flow index is greater than a preset second node flow index threshold, the water distribution pipeline node is the abnormal node;

[0012] The data of the pipeline turning node is obtained and processed to obtain a third node flow index. If the third node flow index is greater than a preset third node flow index threshold, the pipeline turning node is the abnormal node.

[0013] Further, if the abnormal nodes are adjacent nodes, the pipes between the adjacent abnormal nodes are marked as abnormal pipes.

[0014] Acquire the pipeline connected to the abnormal node to obtain a set of pipelines to be detected; acquire the pipeline to be detected and perform segmentation processing to obtain a set of segmentation points of the pipeline to be detected; acquire the set of segmentation points of the pipeline to be detected and perform data collection to obtain a flow data matrix;

[0015] Furthermore, the segmentation process includes:

[0016] The pipeline to be detected is obtained and uniformly segmented into N+2 segments to obtain N+1 pipeline segmentation points to be detected of the pipeline to be detected, and the N+1 pipeline segmentation points to be detected constitute the pipeline segmentation point set to be detected;

[0017] Furthermore, the set of segmented points of the pipeline to be detected is obtained and data collection is performed to obtain pipeline flow data.

[0018] Furthermore, the traffic data matrix includes:

[0019] Obtain pipeline flow data and perform standardization processing to obtain N+1 processed vectors with a vector length of M; obtain N+1 processed vectors and perform differential processing with adjacent processed vectors to obtain N differential flow data vectors with a vector length of M;

[0020] Get N traffic data vectors and perform vector concatenation to obtain the traffic data matrix. The matrix size is .

[0021] Calculate the pipeline similarity between the pipeline to be detected and the normal pipeline, the pipeline similarity is used to measure whether the pipeline to be detected and the normal pipeline are the same type of pipelines; obtain the normal pipeline whose pipeline similarity is greater than a preset similarity threshold and perform segmentation processing to obtain a normal pipeline segmentation point set; obtain the normal pipeline segmentation point set and perform data collection to obtain the flow data matrix;

[0022] Furthermore, the pipeline similarity includes:

[0023] The pipeline data is acquired and processed to obtain a pipeline function vector, wherein the pipeline data includes a pipeline length, a pipeline radius, and a pipeline position; the pipeline similarity is calculated according to the pipeline function vectors of the pipeline to be detected and the normal pipeline, and the normal pipeline whose pipeline similarity is greater than a preset pipeline similarity threshold is acquired to obtain a normal pipeline set.

[0024] Acquire all normal pipelines whose pipeline similarity is greater than a preset similarity threshold and process them to obtain a normal flow data matrix set, and process them according to the first cluster analysis method or the second cluster analysis method to obtain a cluster center;

[0025] Furthermore, the first cluster analysis method includes:

[0026] The square root of the sum of squares of the differences between corresponding elements of matrices is used as the distance metric, the normal traffic data matrix set data is input into the clustering algorithm, and the traffic data matrix is ​​processed to obtain the cluster center.

[0027] Furthermore, the second cluster analysis method includes:

[0028] Extracting statistical features of the flow data matrix to obtain a statistical feature vector, wherein the statistical features include mean, variance, eigenvalue, spectral radius and condition number;

[0029] The normal traffic data matrix set is acquired and processed to obtain a statistical feature vector set, and cluster analysis is performed on the statistical feature vector set to obtain a cluster center.

[0030] Calculating the distance between the flow data matrix of the pipeline to be detected and the cluster center, if the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline;

[0031] Get all pipelines to be detected for processing, get a set of abnormal pipelines, and issue an early warning.

[0032] Furthermore, the warning also includes:

[0033] The positions of the abnormal node and the abnormal pipeline connected to the abnormal node are obtained and an early warning is issued.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. First, obtain the data of pipeline layout and divide the pipeline into small sections through the main water supply pipeline node, water distribution pipeline node and pipeline turning node, which is conducive to subsequent detection and abnormal location; then design the first, second and third flow indicators for the main water supply pipeline node, water distribution pipeline node and pipeline turning node respectively, which can effectively identify the abnormal conditions of different nodes and ensure the accuracy of detection; if the abnormal node is an adjacent node, the pipeline between the nodes is directly marked as an abnormal pipeline, and the nodes and pipelines are combined for early warning, which improves the efficiency of detection and early warning.

[0036] 2. Divide the pipeline into sections and set flow data collection points at the section points to collect comprehensive flow data, which is conducive to a more detailed analysis of the pipeline later; by calculating the pipeline similarity, find the normal pipeline with similar functions to the abnormal pipeline for analysis to reduce the possibility of misjudgment; by processing and analyzing the collected flow data, the flow data matrix is ​​obtained, which can more comprehensively judge the operation status of the pipeline, improve the sensitivity to abnormal pipeline data, and facilitate more accurate distinction of abnormal pipelines.

[0037] 3. The present invention provides two feasible clustering analysis methods. One is to directly use the flow data matrix for clustering, which can capture the global change trend of pipeline flow and effectively improve the accuracy of anomaly detection; the other is clustering analysis based on statistical feature vectors, extracting statistical features such as the mean, variance and eigenvalue of the matrix, simplifying the calculation complexity of the high-dimensional matrix, while retaining the core information of the matrix, and then performing cluster analysis on the statistical feature vectors, which can significantly improve the robustness of abnormal pipeline judgment and improve the accuracy of abnormal pipeline identification; through the above two methods, an efficient detection based on cluster analysis is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flow chart of a pipeline flow detection and early warning method based on data acquisition provided in Example 1 of the present invention;

[0039] Figure 2 A flowchart of the second cluster analysis method provided in Example 1 of the present invention;

[0040] Figure 3 A schematic diagram of the structure of a pipeline flow detection and early warning system based on data acquisition provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Embodiment 1

[0043] Irrigation project construction is the construction of engineering systems for the development, transportation and rational distribution of water resources. Irrigation areas mainly serve agricultural irrigation, while taking into account flood control and drainage, ecological protection and comprehensive utilization functions. Irrigation areas usually cover large tracts of land, and irrigation projects need to cover these areas to ensure the rational distribution of water resources. Therefore, they are characterized by scattered points, long lines and wide areas, and irrigation project construction is usually carried out in areas with scarce water resources. During the construction of the irrigation project, a construction team set up a water diversion pipeline to introduce construction water. The water diversion pipeline includes a main water supply pipeline and a water distribution pipeline. The pipeline is laid along the irrigation project, and has the characteristics of narrow and long pipelines, wide coverage and difficulty in safety inspection. The construction team introduced a pipeline flow detection and early warning method based on data acquisition provided by the present invention to perform safety inspections on pipelines that transport water resources over long distances. The method flow is as follows Figure 1 As shown, the specific implementation is as follows:

[0044] First, pipeline layout data is obtained and processed to obtain pipeline nodes, which include main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes; node flow data in the pipeline nodes is obtained and processed to obtain node flow indicators. If the node flow indicator exceeds a threshold, it is an abnormal node;

[0045] Furthermore, the main water supply pipeline node refers to the starting point of the water supply pipeline, the water distribution pipeline node refers to the intersection of the pipeline diversion, and the pipeline turning node refers to the location where the pipeline flow direction is changed. Table 1 shows the number of nodes of the water diversion pipeline system constructed in this project.

[0046] Table 1. Number of nodes of various types

[0047]

[0048] Further, the node flow index includes a first node flow index, a second node flow index and a third node flow index;

[0049] Further, the data of the main water supply pipeline node is obtained and processed to obtain a first node flow index. If the first node flow index is greater than a preset first node flow index threshold, it is the abnormal node. The calculation formula of the first node flow index is:

[0050] ;

[0051] in, represents the first node traffic index, Represents the weight coefficient of flow volatility, Indicated in The flow rate within a time period, Indicates the maximum flow rate during this time period. Indicates the minimum flow rate during this time period. Indicates the average flow rate during this time period. represents the weight coefficient of flow deviation, Indicates the preset flow rate. represents the weight coefficient of pressure deviation, Indicated in The average pressure during the time period, Indicates the preset pressure.

[0052] Further, the data of the water distribution pipeline node is obtained and processed to obtain a second node flow index. If the second node flow index is greater than a preset second node flow index threshold, it is the abnormal node. The calculation formula of the second node flow index is:

[0053] ;

[0054] in, represents the second node traffic index, represents the weight coefficient of the flow deviation of the water distribution pipeline node, express The water flow rate during the time period, represents the demand water flow during this time period, represents the weight coefficient of the water distribution pipeline deviation, Indicates A water distribution pipeline The water flow obtained during the time period, Indicates the first The required water flow of each water distribution pipe is Indicates the number of water distribution pipes at this water distribution pipe node.

[0055] Further, the data of the pipeline turning node is obtained and processed to obtain a third node flow index. If the third node flow index is greater than a preset third node flow index threshold, it is the abnormal node. The calculation formula of the third node flow index is:

[0056] ;

[0057] in, represents the third node traffic index, The weight coefficient representing the pressure difference at the pipe bend, represents the pressure at the entrance of the turning node, represents the pressure at the exit of the turning node, Indicates the preset maximum pressure difference, The weight coefficient representing the flow difference at the pipe bend, Indicates that the entrance of the turning node is The flow rate within a time period, Indicates that the exit of the turning node is at The flow rate within a time period, represents the weight coefficient of flow stability, It represents the standard deviation of flow rate per unit time. It represents the mean flow rate per unit time. represents the drag loss ratio of the turning node, represents the fluid density, Indicates the flow velocity at the turning node. Table 2 shows the number of abnormal nodes after identification.

[0058] Table 2. Number of abnormal nodes

[0059]

[0060] By designing different flow index calculation formulas for the main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes, and conducting tests, abnormal nodes can be accurately identified, and accurate monitoring of different functional nodes in the pipeline system can be achieved. According to the functional characteristics of different nodes, such as flow fluctuations at the main water supply node, allocation requirements at the water distribution node and fluid loss at the turning node, targeted design of flow indicators is achieved, making the detection more comprehensive.

[0061] Further, if the abnormal nodes are adjacent nodes, the pipes between the abnormal nodes are marked as abnormal pipes.

[0062] Furthermore, if the abnormal node is an adjacent node, it reflects a deeper structural or functional fault in the pipeline, such as blockage, leakage, excessive loss or unstable connection in the pipeline. Marking it as an abnormal pipeline helps to focus on the problem area and quickly locate possible fault points. In the above data, there are two pairs of adjacent nodes that are both abnormal nodes, one of which is a water distribution pipeline node and a pipeline turning node.

[0063] Further alarms are issued for abnormal nodes and abnormal pipelines to remind staff of the locations of abnormal pipelines and abnormal nodes.

[0064] By marking abnormal pipelines, the inspection focus can be concentrated in key areas, and the abnormal pipelines can be further analyzed with relevant information such as flow data, pressure data, and pipeline material. This method not only improves the efficiency of fault diagnosis, but also reduces the time cost of large-scale inspection of the entire pipeline network.

[0065] Acquire the pipeline connected to the abnormal node to obtain a set of pipelines to be detected; acquire the pipeline to be detected and perform segmentation processing to obtain a set of segmentation points of the pipeline to be detected; acquire the set of segmentation points of the pipeline to be detected and perform data collection to obtain a flow data matrix;

[0066] Furthermore, the segmentation process includes:

[0067] Acquire the pipeline to be detected and perform uniform N+2 segmentation processing on it to obtain N+1 segmentation points of the pipeline to be detected;

[0068] The pipeline segmentation points to be detected are obtained and data is collected to obtain pipeline flow data, which includes flow velocity, flow per unit time, pipeline pressure, etc.

[0069] Through uniform segmentation processing, the long pipeline is divided into N+2 small sections. Flow data collection points are set between each section to obtain N+1 collection points, and the pipeline flow data of each collection point is collected, which helps to expand the monitoring range of flow data and improve the accuracy of data collection.

[0070] Furthermore, the traffic data matrix includes:

[0071] Obtain pipeline flow data and perform standardization processing to obtain N+1 processed vectors with a vector length of M; obtain N+1 processed vectors and perform differential processing with adjacent processed vectors to obtain N differential flow data vectors with a vector length of M;

[0072] Get N traffic data vectors and perform vector concatenation to obtain the traffic data matrix. The matrix size is .

[0073] Standardizing pipeline flow data can eliminate scale differences between different pipelines caused by slight differences in length, diameter or flow range, making the data comparable; by calculating the difference between adjacent flow data vectors, it can effectively capture the local change characteristics of pipeline flow, significantly amplify abnormal fluctuations, and enhance the sensitivity of anomaly detection.

[0074] Calculate the pipeline similarity between the pipeline to be detected and the normal pipeline, and the pipeline similarity is used to measure whether the pipeline to be detected and the normal pipeline are the same type of pipelines; obtain the normal pipeline whose pipeline similarity is greater than the preset similarity threshold and perform segmentation processing to obtain a normal pipeline segmentation point set; obtain the normal pipeline segmentation point set and perform data collection to obtain a flow data matrix; the normal pipeline refers to other pipelines in the pipeline system that are not identified as pipelines to be detected and normal pipelines in historical data.

[0075] Furthermore, the flow data matrix of the normal pipeline and the flow data matrix of the pipeline to be tested can be obtained, and the matrix size is .

[0076] Furthermore, the pipeline similarity includes:

[0077] The pipeline data is obtained and normalized to obtain the pipeline function vector, which includes the pipeline length, pipeline radius, and the type of nodes at both ends of the pipeline. Then, the cosine similarity formula is used to calculate the similarity between the two vectors. The calculation formula for pipeline similarity is:

[0078] ;

[0079] in, represents pipeline similarity, a pipeline capability vector representing the pipeline to be tested, a pipeline capability vector representing a normal pipeline, Represents the two-norm, which is used to calculate the vector modulus.

[0080] Furthermore, the pipeline similarity between the pipeline to be detected and all normal pipelines is calculated, and if the pipeline similarity is greater than a preset pipeline similarity threshold, a normal pipeline set is obtained.

[0081] After the pipeline data is standardized, the influence of the dimensional differences of different feature data on the calculation is eliminated, making the cosine similarity calculation results more objective and accurate; by calculating the pipeline similarity between the pipeline to be detected and the normal pipeline, and comparing it with the preset threshold, normal pipelines with similar functional types to the pipeline to be detected can be screened out, and the detection scope can be more accurately narrowed to a set of pipelines with similar characteristics, avoiding interference from irrelevant pipelines and improving the efficiency of subsequent detection and analysis.

[0082] Acquire and process all normal pipelines whose pipeline similarity is greater than a preset similarity threshold to obtain a normal flow data vector set, and perform cluster analysis to obtain a cluster center;

[0083] Calculating the distance between the flow data matrix to be detected and the cluster center, if the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline;

[0084] Furthermore, the first cluster analysis method includes:

[0085] The square root of the sum of squares of matrix element differences is used as a distance metric, and the matrix data is input into a clustering algorithm to analyze the traffic data matrix to obtain a cluster center.

[0086] Furthermore, the calculation formula of the metric distance is:

[0087] ;

[0088] in, Representation Matrix and matrix The metric distance between represents the Frobenius norm, Representation Matrix Middle Line The elements of the column, Display Matrix Middle Line Elements of a column.

[0089] Furthermore, the Frobenius norm, also called the Frobenius norm, is a norm of a matrix, which is used to measure the size of a matrix or the distance between matrices.

[0090] Furthermore, the K-means clustering algorithm is used, the number of clusters is set to 1, and then a matrix is ​​randomly selected as the initial cluster center. The distance from each matrix to the current cluster center is calculated, and each matrix is ​​assigned to the cluster center with the nearest distance. The average matrix of all current matrices is recalculated as the new cluster center. The assignment and update steps are repeated until the cluster centers converge.

[0091] Furthermore, we get the cluster center Then, the distance between the pipeline flow data matrix to be detected and the cluster center of the pipeline to be detected is calculated. If the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline.

[0092] Furthermore, the flow data matrix of the pipeline A1 to be tested is obtained , and calculate the matrix and cluster centers The distance is 0.92, which is greater than the preset distance threshold 0.5, that is, the pipeline A1 to be detected is an abnormal pipeline.

[0093] By solving the cluster center and then combining the matrix metric distance of the Frobenius norm to calculate the distance to the cluster center, abnormal pipelines can be identified quickly and efficiently, which not only reduces the complexity of the data, but also improves the accuracy and real-time performance of detection.

[0094] Furthermore, the process of the second cluster analysis method is as follows Figure 2 As shown, including:

[0095] Obtaining a flow data matrix and extracting statistical features of the matrix to obtain a statistical feature vector, wherein the statistical features include a mean, a variance, an eigenvalue, a spectral radius, and a condition number;

[0096] Obtain a set of statistical feature vectors and perform cluster analysis to obtain cluster centers.

[0097] Furthermore, after extracting the above statistical features of the flow data matrix, they are combined into a statistical feature vector. These statistical feature vectors can be used as representatives of the matrix data of the pipeline. The form of the statistical feature vector is:

[0098] ;

[0099] in, represents the statistical eigenvector, represents the mean, represents the variance, express eigenvalues, represents the spectral radius, Represents the condition number.

[0100] Furthermore, the statistical feature vectors of all pipelines are combined into a statistical feature vector set, and then cluster analysis is performed on the set, usually using K-means clustering or other clustering algorithms.

[0101] Furthermore, after obtaining the cluster center, the distance between the statistical feature vector of the pipeline to be detected and the cluster center is calculated. If the distance exceeds a preset distance threshold, the pipeline to be detected is an abnormal pipeline.

[0102] The second cluster analysis method extracts the statistical characteristics of the flow data matrix and performs cluster analysis, which not only reduces the complexity of data processing and improves the analysis efficiency, but also can judge abnormal pipelines through accurate cluster center calculation and effectively detect abnormal pipeline flow.

[0103] The pipelines to be detected in the pipeline set to be detected are obtained and processed to obtain an abnormal pipeline set, and an early warning is issued.

[0104] Furthermore, the warning also includes:

[0105] The positions of the abnormal node and the abnormal pipeline connected to the abnormal node are obtained and an early warning is issued.

[0106] Furthermore, once an abnormal pipeline is detected, the system will automatically trigger an early warning, which can be carried out through multiple channels, such as notifying relevant personnel through system alarms, text messages, emails, etc. The early warning content may include the identification of the abnormal pipeline, the location of the abnormal pipeline, and potential risk warnings.

[0107] The implementation of this early warning mechanism can not only quickly respond to potential pipeline failures, but also help staff identify possible chain risks in advance, and ensure that relevant personnel can take prompt measures through a multi-channel notification mechanism.

[0108] A pipeline flow detection and early warning method based on data acquisition provided by an embodiment of the present invention realizes accurate detection and rapid early warning of pipeline anomalies by designing targeted flow index calculation formulas for different nodes of the entire pipeline system, and combining segmented processing, flow data acquisition, data standardization, similarity calculation and cluster analysis methods. First, by conducting an in-depth analysis of the flow data of the main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes, and designing flow indicators, it is helpful to effectively identify abnormal nodes according to the functional characteristics of the nodes; at the same time, through the segmented acquisition of flow data and the processing of the flow data matrix, the sensitivity to local change characteristics is enhanced, and the accuracy of anomaly identification is improved. The two clustering analysis methods provided further cluster the flow data matrix or statistical feature vector, and judge the abnormal situation of the pipeline in combination with the distance, which greatly improves the detection efficiency and accuracy.

[0109] Embodiment 2

[0110] In this embodiment, a pipeline flow detection and early warning system based on data collection is provided. The system structure is as follows: Figure 3 As shown, the specific implementation is as follows:

[0111] The abnormal node acquisition module is used to acquire and process pipeline layout data to obtain pipeline nodes, which include main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes; acquire and process node flow data of the pipeline nodes to obtain node flow indicators. If the node flow indicator exceeds a threshold, it is an abnormal node;

[0112] Further, the node flow index includes a first node flow index, a second node flow index and a third node flow index;

[0113] Further, the data of the main water supply pipeline node is obtained and processed to obtain a first node flow index. If the first node flow index is greater than a preset first node flow index threshold, it is the abnormal node. The calculation formula of the first node flow index is:

[0114] ;

[0115] in, represents the first node traffic index, Represents the weight coefficient of flow volatility, Indicated in The flow rate within a time period, Indicates the maximum flow rate during this time period. Indicates the minimum flow rate during this time period. Indicates the average flow rate during this time period. represents the weight coefficient of flow deviation, Indicates the preset flow rate. represents the weight coefficient of pressure deviation, Indicated in The average pressure during the time period, Indicates the preset pressure.

[0116] Further, the data of the water distribution pipeline node is obtained and processed to obtain a second node flow index. If the second node flow index is greater than a preset second node flow index threshold, it is the abnormal node. The calculation formula of the second node flow index is:

[0117] ;

[0118] in, represents the second node traffic index, represents the weight coefficient of the flow deviation of the water distribution pipeline node, express The water flow rate during the time period, represents the demand water flow during this time period, represents the weight coefficient of the water distribution pipeline deviation, Indicates A water distribution pipeline The water flow obtained during the time period, Indicates the first The required water flow of each water distribution pipe is Indicates the number of water distribution pipes at this water distribution pipe node.

[0119] Further, the data of the pipeline turning node is obtained and processed to obtain a third node flow index. If the third node flow index is greater than a preset third node flow index threshold, it is the abnormal node. The calculation formula of the third node flow index is:

[0120] ;

[0121] in, represents the third node traffic index, The weight coefficient representing the pressure difference at the pipe bend, represents the pressure at the entrance of the turning node, represents the pressure at the exit of the turning node, Indicates the preset maximum pressure difference, The weight coefficient representing the flow difference at the pipe bend, Indicates that the entrance of the turning node is The flow rate within a time period, Indicates that the exit of the turning node is at The flow rate within a time period, represents the weight coefficient of flow stability, It represents the standard deviation of flow rate per unit time. It represents the mean flow rate per unit time. represents the drag loss ratio of the turning node, represents the fluid density, Indicates the flow velocity at the turning node. Table 3 shows the abnormal nodes after identification.

[0122] Table 3. Number of abnormal nodes

[0123] ;

[0124] The pipeline analysis module to be detected is used to obtain the pipeline connected to the abnormal node to obtain a set of pipelines to be detected; obtain the pipeline to be detected and perform segmentation processing to obtain a set of segmentation points of the pipeline to be detected; obtain the set of segmentation points of the pipeline to be detected and perform data collection to obtain a flow data matrix;

[0125] Furthermore, in this embodiment, there are 130 pipelines to be inspected, constituting a pipeline set to be inspected.

[0126] The normal pipeline analysis module calculates the pipeline similarity between the pipeline to be detected and the normal pipeline, and the pipeline similarity is used to measure whether the pipeline to be detected and the normal pipeline are the same type of pipelines; obtains the normal pipelines whose pipeline similarity is greater than a preset similarity threshold and performs segmentation processing to obtain a normal pipeline segmentation point set; obtains the normal pipeline segmentation point set and performs data collection to obtain the flow data matrix;

[0127] Furthermore, the pipeline similarity includes:

[0128] The pipeline data is obtained and normalized to obtain the pipeline function vector, which includes the pipeline length, pipeline radius, and the type of nodes at both ends of the pipeline. Then, the cosine similarity formula is used to calculate the similarity between the two vectors. The calculation formula for pipeline similarity is:

[0129] ;

[0130] in, represents pipeline similarity, a pipeline capability vector representing the pipeline to be tested, A vector of pipeline capabilities representing a normal pipeline.

[0131] Furthermore, the pipeline similarity between the pipeline to be detected and all normal pipelines is calculated, and if the pipeline similarity is greater than a preset pipeline similarity threshold, a normal pipeline set is obtained.

[0132] A cluster center calculation module is used to obtain and process all normal pipelines whose pipeline similarity is greater than a preset similarity threshold to obtain a normal flow data matrix set, and process them according to a second cluster analysis method to obtain a cluster center;

[0133] Furthermore, the second cluster analysis method includes:

[0134] Obtaining a flow data matrix and extracting statistical features of the matrix to obtain a statistical feature vector, wherein the statistical features include a mean, a variance, an eigenvalue, a spectral radius, and a condition number;

[0135] Obtain a set of statistical feature vectors and perform cluster analysis to obtain cluster centers.

[0136] Furthermore, after extracting the above statistical features of the flow data matrix, they are combined into a statistical feature vector. These statistical feature vectors can be used as representatives of the matrix data of the pipeline. The form of the statistical feature vector is:

[0137] ;

[0138] in, represents the statistical eigenvector, represents the mean, represents the variance, express eigenvalues, represents the spectral radius, Represents the condition number.

[0139] Furthermore, the statistical feature vectors of all pipelines are combined into a statistical feature vector set, and then cluster analysis is performed on the set, usually using K-means clustering or other clustering algorithms.

[0140] Furthermore, after obtaining the cluster center, the distance between the statistical feature vector of the pipeline to be detected and the cluster center is calculated. If the distance exceeds a preset distance threshold, the pipeline to be detected is an abnormal pipeline.

[0141] An abnormal pipeline judgment module is used to calculate the distance between the flow data matrix of the pipeline to be detected and the cluster center. If the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline;

[0142] Furthermore, 130 pipelines were judged. Table 4 shows some pipeline data. After calculation, the distance between the flow data matrix and the cluster center of all pipelines is less than the preset threshold, indicating that there is no abnormal pipeline.

[0143] Table 4. Metric distances of some pipelines to be inspected

[0144]

[0145] The early warning module is used to obtain all the abnormal pipelines, obtain an abnormal pipeline set, and issue an early warning in combination with abnormal nodes.

[0146] Furthermore, during the pipeline inspection process, no abnormal pipelines were found, only abnormal nodes were warned, and the system notified the staff of the location of the abnormal nodes through alarms and other means.

[0147] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A pipeline flow detection and early warning method based on data acquisition, characterized in that: include: The pipeline layout data is obtained and processed to obtain pipeline nodes, wherein the pipeline nodes include main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes; the node flow data of the pipeline nodes is obtained and processed to obtain node flow indicators. If the node flow indicator exceeds a threshold, it is an abnormal node; Acquire the pipeline connected to the abnormal node to obtain a set of pipelines to be detected; acquire the pipeline to be detected and perform segmentation processing to obtain a set of segmentation points of the pipeline to be detected; acquire the set of segmentation points of the pipeline to be detected and perform data collection to obtain a flow data matrix of the pipeline to be detected; Calculate the pipeline similarity between the pipeline to be detected and the normal pipeline, the pipeline similarity is used to measure whether the pipeline to be detected and the normal pipeline are the same type of pipelines; obtain the normal pipeline whose pipeline similarity is greater than a preset similarity threshold and perform segmentation processing to obtain a normal pipeline segmentation point set; obtain the normal pipeline segmentation point set and perform data collection to obtain the flow data matrix of the normal pipeline; Acquire all normal pipelines whose pipeline similarity is greater than a preset similarity threshold and process them to obtain a normal flow data matrix set, and process them according to the first cluster analysis method or the second cluster analysis method to obtain a cluster center; Calculating the distance between the flow data matrix of the pipeline to be detected and the cluster center, if the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline; Get all pipelines to be detected for processing, get a set of abnormal pipelines, and issue an early warning.

2. According to the data acquisition-based pipeline flow detection and early warning method of claim 1, it is characterized in that: The node traffic indicators include: Acquire and process the data of the main water pipeline node to obtain a first node flow index. If the first node flow index is greater than a preset first node flow index threshold, the main water pipeline node is the abnormal node. Acquire and process the data of the water distribution pipeline node to obtain a second node flow index, and if the second node flow index is greater than a preset second node flow index threshold, the water distribution pipeline node is the abnormal node; The data of the pipeline turning node is obtained and processed to obtain a third node flow index. If the third node flow index is greater than a preset third node flow index threshold, the pipeline turning node is the abnormal node.

3. According to the data acquisition-based pipeline flow detection and early warning method of claim 2, it is characterized in that: Also includes: If the abnormal nodes are adjacent nodes, the pipes between the adjacent abnormal nodes are marked as abnormal pipes.

4. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The segmentation process includes: Acquire the pipeline to be detected and perform uniform N+2 segmentation processing to obtain N+1 pipeline segmentation points to be detected of the pipeline to be detected, and the N+1 pipeline segmentation points to be detected constitute the pipeline segmentation point set to be detected; The pipeline segmentation point set to be detected is obtained and data is collected to obtain pipeline flow data.

5. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The traffic data matrix includes: Obtain pipeline flow data and perform standardization processing to obtain N+1 processed vectors with a vector length of M; obtain N+1 processed vectors and perform differential processing with adjacent processed vectors to obtain N differential flow data vectors with a vector length of M; Get N traffic data vectors and perform vector concatenation to obtain the traffic data matrix. The matrix size is .

6. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The pipeline similarity includes: The pipeline data is acquired and processed to obtain a pipeline function vector, wherein the pipeline data includes a pipeline length, a pipeline radius, and a pipeline position; the pipeline similarity is calculated according to the pipeline function vectors of the pipeline to be detected and the normal pipeline, and the normal pipeline whose pipeline similarity is greater than a preset pipeline similarity threshold is acquired to obtain a normal pipeline set.

7. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The first cluster analysis method comprises: The square root of the sum of squares of the differences between corresponding elements of matrices is used as the distance metric, the normal traffic data matrix set data is input into the clustering algorithm, and the traffic data matrix is ​​processed to obtain the cluster center.

8. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The second cluster analysis method comprises: Extracting statistical features of the flow data matrix to obtain a statistical feature vector, wherein the statistical features include mean, variance, eigenvalue, spectral radius and condition number; The normal traffic data matrix set is acquired and processed to obtain a statistical feature vector set, and cluster analysis is performed on the statistical feature vector set to obtain a cluster center.

9. The pipeline flow detection and early warning method based on data acquisition according to claim 1 is characterized in that: The warnings include: The location of the abnormal node and the location of the abnormal pipeline in the abnormal pipeline set connected to the abnormal node are obtained, and an early warning is issued according to the location.

10. A pipeline flow detection and early warning system based on data acquisition, characterized in that: include: The abnormal node acquisition module is used to acquire and process pipeline layout data to obtain pipeline nodes, which include main water supply pipeline nodes, water distribution pipeline nodes and pipeline turning nodes; acquire and process node flow data of the pipeline nodes to obtain node flow indicators. If the node flow indicator exceeds a threshold, it is an abnormal node; The pipeline analysis module to be detected is used to obtain the pipeline connected to the abnormal node to obtain a set of pipelines to be detected; obtain the pipeline to be detected and perform segmentation processing to obtain a set of segmentation points of the pipeline to be detected; obtain the set of segmentation points of the pipeline to be detected and perform data collection to obtain a flow data matrix; The normal pipeline analysis module calculates the pipeline similarity between the pipeline to be detected and the normal pipeline, and the pipeline similarity is used to measure whether the pipeline to be detected and the normal pipeline are the same type of pipelines; obtains the normal pipelines whose pipeline similarity is greater than a preset similarity threshold and performs segmentation processing to obtain a normal pipeline segmentation point set; obtains the normal pipeline segmentation point set and performs data collection to obtain the flow data matrix; A cluster center calculation module is used to obtain and process all normal pipelines whose pipeline similarity is greater than a preset similarity threshold to obtain a normal flow data matrix set, and process them according to the first cluster analysis method or the second cluster analysis method to obtain a cluster center; An abnormal pipeline judgment module is used to calculate the distance between the flow data matrix of the pipeline to be detected and the cluster center. If the distance is greater than a preset distance threshold, it indicates that the pipeline to be detected is an abnormal pipeline; The early warning module is used to obtain all the abnormal pipelines, obtain an abnormal pipeline set, and issue an early warning in combination with the abnormal nodes.

Images