Water supply network leakage detection method and system

Abstract

The invention provides a water supply network leakage detection method and system, and relates to the technical field of data processing, and the method comprises the steps: carrying out the exception processing of an original flow data set and an original pressure data set, so as to obtain a standby flow data set and a standby pressure data set; obtaining predicted flow data through the flow prediction model, and judging whether the pipeline node has a large amount of leakage based on the real-time flow data and the predicted flow data; combining every two of the plurality of pipeline nodes into a plurality of node pairs, obtaining the fluctuation correlation degree of the node pairs, and judging whether the node pairs are abnormal pairs or not according to the fluctuation correlation degree; and selecting a plurality of to-be-identified nodes from all the pipeline nodes based on the pipeline nodes in the abnormal pairs, obtaining residual values corresponding to the to-be-identified nodes, and selecting leakage nodes from the plurality of to-be-identified nodes through the residual values so as to complete small-amount leakage detection. Compared with traditional manual detection, the detection efficiency is improved, meanwhile, the detection precision is improved, and accurate detection of a small amount of leakage is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for detecting leakage in a water supply network. Background Art

[0002] The water supply network is an indispensable infrastructure for modern urban construction, and plays a vital role in urban development and people's lives. Its essence is the pipeline system for water transmission and distribution to users in water supply projects.

[0003] However, with the continuous expansion of the scale of urban water supply networks and the increase in their service time, the leakage problem of large water supply networks has become a difficult problem that needs to be solved urgently in leakage control. If leakage occurs in the water supply network, it will not only waste water resources, but also increase the safety risk of water supply.

[0004] At present, leakage detection of water supply networks mostly still uses traditional passive manual inspection methods. Inspectors use tools such as listening sticks to detect and locate leakage in water supply networks. This method is not only inefficient, but also difficult to accurately identify whether there is a leakage when the leakage amount is small. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a water supply network leakage detection method and system, aiming to solve the technical problem that the passive manual inspection method in the prior art detects and locates leakage in the water supply network, which not only has low detection efficiency but also is difficult to accurately identify whether there is a leakage when the leakage amount is small.

[0006] In order to achieve the above objectives, in a first aspect, an embodiment of the present application provides a water supply network leakage detection method, comprising the following steps: Acquire an original flow data set and an original pressure data set of a pipeline node in a water supply network, and perform exception processing on the original flow data set and the original pressure data set to acquire a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data, and the standby pressure data set includes a plurality of standby pressure data; Acquire predicted flow data through a flow prediction model, and determine whether there is a large amount of leakage at the pipeline node based on the real-time flow data and the predicted flow data; Performing equalization processing on the standby pressure data set to obtain an updated pressure data set, wherein the updated pressure data set includes a plurality of updated pressure data, combining a plurality of pipeline nodes into a plurality of node pairs, obtaining a fluctuation correlation degree of the node pairs, and determining whether the node pairs are abnormal pairs according to the fluctuation correlation degree; Based on the pipeline nodes in the abnormal pair, several nodes to be identified are selected from all the pipeline nodes, the node pairs corresponding to the nodes to be identified are all selected as reference pairs, and the fluctuation correlation corresponding to the reference pairs is selected as the reference correlation, the predicted correlation of the reference pairs is obtained based on the fluctuation prediction model, the residual value corresponding to the node to be identified is obtained based on the reference correlation and the predicted correlation, and the leakage node is selected from the several nodes to be identified through the residual value to complete a small amount of leakage detection.

[0007] Further, the original flow data set includes a plurality of flow data in a continuous time frame, the original pressure data set includes a plurality of pressure data in a continuous time frame, and the step of performing exception processing on the original flow data set and the original pressure data set to obtain a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data and the standby pressure data set includes a plurality of standby pressure data comprises: Obtaining a flow average of a plurality of the flow data, and obtaining a pressure average of a plurality of the pressure data, respectively obtaining a flow calibration value and a pressure calibration value based on the flow average and the pressure average, obtaining a flow threshold range through the flow average and the flow calibration value, and obtaining a pressure threshold range through the pressure average and the pressure calibration value; Compare a plurality of the flow data with the flow threshold range respectively, select the flow data outside the flow threshold range as abnormal flow data, select the flow data within the flow threshold range as retained flow data, remove the abnormal flow data, and combine a plurality of the retained flow data into a retained flow data set; Comparing a plurality of the pressure data with the pressure threshold range respectively, selecting the pressure data outside the pressure threshold range as abnormal pressure data, selecting the pressure data within the pressure threshold range as retained pressure data, eliminating the abnormal pressure data, and combining a plurality of the retained pressure data into a retained pressure data set; Based on the time frame, it is determined whether there is a flow missing area in the reserved flow data set; if there is a flow missing area in the reserved flow data set, flow filling data is generated based on the reserved flow data adjacent to the flow missing area, the flow missing area is filled by the flow filling data, and both the reserved flow data and the flow filling data are selected as standby flow data to form a standby flow data set; Based on the time frame, it is determined whether there is a pressure missing area in the retained pressure data set. If there is a pressure missing area in the retained pressure data set, pressure filling data is generated based on the retained pressure data adjacent to the pressure missing area, and the pressure missing area is filled with the pressure filling data. Both the retained pressure data and the pressure filling data are selected as stand-by pressure data to form a stand-by pressure data set.

[0008] Furthermore, the formula for obtaining the flow calibration value is: , in, represents the flow calibration value of the i-th pipeline node, represents the total number of time frames, represents the flow data of the i-th pipeline node in the t-th time frame, represents the mean flow rate of the i-th pipeline node.

[0009] Furthermore, the formula for obtaining the updated pressure data is: , in, represents the updated pressure data of the jth pipeline node in the tth time frame, represents the standby pressure data of the jth pipeline node in the tth time frame, represents the mean pressure of the jth pipeline node, Represents the pressure calibration value of the jth pipeline node.

[0010] Furthermore, the step of obtaining the fluctuation correlation of the node pair and determining whether the node pair is an abnormal pair according to the fluctuation correlation comprises: Based on a time window, divide the plurality of updated pressure data into a plurality of window data groups, and obtain the window average pressure of the window data groups; Acquire a plurality of fluctuation correlations corresponding to the node pair through the updated pressure data and the window average pressure; Obtain the fluctuation correlation difference between adjacent fluctuation correlations, compare the fluctuation correlation difference with a fluctuation threshold, and if the fluctuation correlation difference is greater than the fluctuation threshold, determine that the node pair is an abnormal pair.

[0011] Furthermore, the formula for obtaining the fluctuation correlation is: , in, It represents the fluctuation correlation degree in the mth time window when the jth pipeline node and the kth pipeline node are a node pair. represents the length of the m-th time window, represents the q-th time frame in the m-th time window, represents the updated pressure data of the j-th pipeline node at the q-th time frame in the m-th time window, represents the window average pressure of the j-th pipeline node under the m-th time window, represents the updated pressure data of the k-th pipeline node at the q-th time frame in the m-th time window, represents the window average pressure of the k-th pipeline node under the m-th time window.

[0012] Furthermore, the step of selecting a plurality of nodes to be identified from all the pipeline nodes based on the pipeline nodes in the anomaly pair includes: Select both pipeline nodes in the anomaly pair as anomaly nodes, and select all node pairs with the anomaly nodes as pairs to be used; Select all pipeline nodes in the anomaly pair and the pairs to be used as nodes to be identified.

[0013] Furthermore, the step of obtaining a residual value corresponding to the node to be identified based on the reference correlation degree and the prediction correlation degree includes: Obtain the correlation residual between the reference correlation degree and the prediction correlation degree under the same time window; Superimpose a plurality of the correlation residuals into a sub-residual corresponding to the reference pair; Superimpose a plurality of the sub-residuals into a residual value corresponding to the node to be identified.

[0014] Furthermore, the step of selecting a leakage node from a plurality of the nodes to be identified through the residual value includes: Obtain a residual calibration value through a plurality of the residual values, and respectively obtain the calibration difference between the residual value and the residual calibration value; Compare the calibration difference with a residual threshold, and select the node to be identified corresponding to the calibration difference greater than the residual threshold as the leakage node.

[0015] In a second aspect, an embodiment of the present application provides a water supply network leakage detection system, which is applied to the water supply network leakage detection method as described in the first aspect above. The system includes: An acquisition module, configured to obtain an original flow data set and an original pressure data set of pipeline nodes in a water supply network, perform anomaly processing on the original flow data set and the original pressure data set to obtain a data set of available flows and a data set of available pressures, the data set of available flows includes a plurality of available flow data, and the data set of available pressures includes a plurality of available pressure data; A first detection module is used to obtain predicted flow data through a flow prediction model, and determine whether there is a large amount of leakage at the pipeline node based on the real-time flow data and the predicted flow data; an identification module, configured to perform equalization processing on the standby pressure data set to obtain an updated pressure data set, wherein the updated pressure data set includes a plurality of updated pressure data, and to combine a plurality of pipeline nodes into a plurality of node pairs in pairs, to obtain a fluctuation correlation degree of the node pairs, and to determine whether the node pairs are abnormal pairs according to the fluctuation correlation degree; The second detection module is used to select several nodes to be identified from all the pipeline nodes based on the pipeline nodes in the abnormal pair, select all node pairs corresponding to the nodes to be identified as reference pairs, and select the fluctuation correlation corresponding to the reference pairs as reference correlation, obtain the predicted correlation of the reference pairs based on the fluctuation prediction model, obtain the residual value corresponding to the node to be identified based on the reference correlation and the predicted correlation, and select the leakage node from the several nodes to be identified through the residual value to complete a small amount of leakage detection.

[0016] In a third aspect, an embodiment of the present application provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the water supply network leakage detection method as described in the first aspect above is implemented.

[0017] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the water supply network leakage detection method as described in the first aspect above is implemented.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: after obtaining the original flow data set and the original pressure data set, by performing exception processing on both, the wrong collection and missed collection caused by the error of the collection sensor during the data collection process can be avoided, the accuracy of the data can be improved, and an accurate data basis can be provided for subsequent leakage detection; the difference between the actual flow data and the predicted flow data can be used to quickly complete the identification of pipeline nodes with a large number of leakages; by setting the node pair and introducing the fluctuation correlation, a small amount of leakage can be identified by the pressure synergy of the two nodes in the time window. When the influence of a small amount of leakage on the change of pressure or flow is small, the detection of a small amount of leakage can still be accurately completed by the sudden change of the hydraulic relationship between the nodes, and the influence of water fluctuation or noise on the detection of a small amount of leakage can be avoided; by performing the equalization processing, the comparability of data between different nodes is ensured; a large amount of leakage is identified by flow, and a small amount of leakage is identified by pressure. Compared with traditional manual detection, the detection efficiency is improved while the detection accuracy is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Flow chart of the water supply network leakage detection method in the first embodiment of the present invention; Figure 2 It is a structural block diagram of a water supply network leakage detection system in a second embodiment of the present invention; The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0020] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0021] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0023] See also Figure 1 The first embodiment of the present invention provides a water supply network leakage detection method, comprising the following steps: S10: obtaining an original flow data set and an original pressure data set of a pipeline node in a water supply network, and performing exception processing on the original flow data set and the original pressure data set to obtain a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data, and the standby pressure data set includes a plurality of standby pressure data; It can be understood that a flow sensor and a pressure sensor are arranged on the pipeline node to obtain the original flow data set and the original pressure data set. The original flow data set includes a plurality of flow data in a continuous time frame, and the original pressure data set includes a plurality of pressure data in a continuous time frame.

[0024] The step S10 comprises: S110: Obtaining a flow average of a plurality of flow data, and obtaining a pressure average of a plurality of pressure data, respectively obtaining a flow calibration value and a pressure calibration value based on the flow average and the pressure average, obtaining a flow threshold range through the flow average and the flow calibration value, and obtaining a pressure threshold range through the pressure average and the pressure calibration value; The formula for obtaining the flow calibration value is: , in, represents the flow calibration value of the i-th pipeline node, represents the total number of time frames, represents the flow data of the i-th pipeline node in the t-th time frame, The method for obtaining the pressure calibration value is the same as the method for obtaining the flow calibration value, which will not be described in detail here.

[0025] S120: Compare the plurality of flow data with the flow threshold range respectively, select the flow data outside the flow threshold range as abnormal flow data, select the flow data within the flow threshold range as retained flow data, remove the abnormal flow data, and combine the plurality of retained flow data into a retained flow data set; The first endpoint value=the flow average value-3*the flow calibration value, the second endpoint value=the flow average value+3*the flow calibration value, and the first endpoint value and the second endpoint value constitute the flow threshold range.

[0026] S130: comparing the plurality of pressure data with the pressure threshold range respectively, selecting the pressure data outside the pressure threshold range as abnormal pressure data, selecting the pressure data within the pressure threshold range as retained pressure data, eliminating the abnormal pressure data, and combining the plurality of retained pressure data into a retained pressure data set; The pressure threshold range is obtained in the same manner as the flow threshold range, which will not be described in detail here.

[0027] S140: judging whether there is a flow missing area in the reserved flow data set based on the time frame; if there is a flow missing area in the reserved flow data set, generating flow filling data based on the reserved flow data adjacent to the flow missing area, filling the flow missing area with the flow filling data, and selecting both the reserved flow data and the flow filling data as standby flow data to form a standby flow data set; In this embodiment, the average value of the reserved flow data adjacent to the flow missing area is the flow filling data.

[0028] S150: Determine whether there is a pressure missing area in the reserved pressure dataset based on the time frame. If there is a pressure missing area in the reserved pressure dataset, generate pressure filling data based on the reserved pressure data adjacent to the pressure missing area, fill the pressure missing area with the pressure filling data, and select both the reserved pressure data and the pressure filling data as the pressure data to be used to form a pressure dataset to be used.

[0029] S20: Obtain predicted flow data through a flow prediction model, and determine whether there is a large amount of leakage at the pipeline node based on the real-time flow data and the predicted flow data; It should be noted that the flow prediction model is a trained flow prediction neural network model, and its training process is as follows: Obtain the previous normal flow dataset corresponding to the pipeline node, partition the previous normal flow dataset into an input dataset and an output dataset, use the input dataset as the input value of the flow prediction neural network model to obtain training predicted flow data through the flow prediction neural network model, construct a loss function based on the training predicted flow data and the output dataset, and then complete the training of the flow prediction neural network model to obtain the predicted flow data through it. When there is a large difference between the real-time flow data and the predicted flow data, it can be determined that there is a large amount of leakage at the pipeline node, resulting in a change in the flow. The application of neural network models for data prediction has been relatively common, and will not be elaborated here. Use the dataset of flow to be used as the input value of the flow prediction model to obtain predicted flow data.

[0030] S30: Perform equalization processing on the pressure dataset to be used to obtain an updated pressure dataset. The updated pressure dataset includes several updated pressure data. Combine several pipeline nodes in pairs into several node pairs, obtain the fluctuation correlation degree of the node pairs, and determine whether the node pairs are abnormal pairs through the fluctuation correlation degree; The formula for obtaining the updated pressure data is: , where, represents the updated pressure data of the j-th pipeline node at the t-th time frame, represents the pressure data to be used of the j-th pipeline node at the t-th time frame, represents the average pressure of the j-th pipeline node, represents the pressure calibration value of the j-th pipeline node. Assume there are four nodes A, B, C, and D, then there are node pairs (A, B), (A, C), (A, D), (B, C), (B, D), (C, D).

[0031] Step S30 includes: S310: Partition a number of the updated pressure data into a number of window data groups based on a time window, and obtain the window average pressure of the window data groups; Assume that the updated pressure data set includes the updated pressure data at 24 time frames, that is, the updated pressure data is collected once an hour, and the length of the time window is 3h, then 8 window data groups are generated. Perform averaging processing on 3 updated pressure data within the window data group to obtain the window average pressure.

[0032] S320: Obtain a number of fluctuation correlation degrees corresponding to the node pair through the updated pressure data and the window average pressure; The formula for obtaining the fluctuation correlation degree is: , where represents the fluctuation correlation degree at the m-th time window when the j-th pipeline node and the k-th pipeline node are a node pair, represents the length of the m-th time window, represents the q-th time frame in the m-th time window, represents the updated pressure data of the j-th pipeline node at the q-th time frame in the m-th time window, represents the window average pressure of the j-th pipeline node at the m-th time window, represents the updated pressure data of the k-th pipeline node at the q-th time frame in the m-th time window, represents the window average pressure of the k-th pipeline node at the m-th time window. If there are 8 time windows, there are 8 corresponding fluctuation correlation degrees.

[0033] S330: Obtain the fluctuation correlation difference between adjacent fluctuation correlation degrees, compare the fluctuation correlation difference with a fluctuation threshold, and if the fluctuation correlation difference is greater than the fluctuation threshold, determine that the node pair is an abnormal pair; A significant change in the fluctuation correlation difference characterizes a change in the coordination between two nodes under the time window, indicating that there may be a small leakage situation between the two pipeline nodes corresponding to the node pair.

[0034] S40: Select several nodes to be identified from all the pipeline nodes based on the pipeline nodes in the anomaly pair. Select all the node pairs corresponding to the nodes to be identified as reference pairs, and select the fluctuation correlation degree corresponding to the reference pairs as the reference correlation degree. Obtain the predicted correlation degree of the reference pairs based on the fluctuation prediction model. Obtain the residual value corresponding to the nodes to be identified based on the reference correlation degree and the predicted correlation degree. Select the leakage nodes from several nodes to be identified through the residual value, so as to complete the detection of a small amount of leakage; The step S40 includes: S410: Select both pipeline nodes in the anomaly pair as anomaly nodes, and select all the node pairs with the anomaly nodes as standby pairs; Assume that the node pair (A, B) is an anomaly pair, then both A and B are selected as anomaly nodes, and the node pairs (A, B), (A, C), (A, D), (B, C), (B, D) are all standby pairs.

[0035] S420: Select all the pipeline nodes in the anomaly pair and the standby pairs as nodes to be identified; Further, nodes A, B, C, and D are all nodes to be identified. It should be noted that the anomaly nodes are included in several nodes to be identified, only with different names.

[0036] Similarly, the reference pairs corresponding to node A are (A, B), (A, C), (A, D). The fluctuation prediction model and the flow prediction model are prediction neural network models of the same nature, which will not be elaborated here. Use the fluctuation correlation degree with an earlier time window as the input of the fluctuation prediction model to obtain the predicted correlation degree. By selecting the nodes to be identified, all the pipeline nodes related to the anomaly nodes are included in the detection range to avoid missed detection.

[0037] S430: Obtain the correlation residual between the reference correlation degree and the predicted correlation degree in the same time window; The difference between the reference correlation degree and the predicted correlation degree is the correlation residual.

[0038] S440: Stack several correlation residuals into a sub-residual corresponding to the reference pair; It can be understood that because there are several time windows, there are several correlation residuals, which form the sub-residual in a stacked manner.

[0039] S450: Stack several sub-residuals into a residual value corresponding to the nodes to be identified; Taking node A as an example, it has 3 reference pairs, so there are 3 sub-residuals. Stack the 3 sub-residuals into a residual value corresponding to node A.

[0040] S460: Obtain a residual calibration value from a plurality of the residual values, and respectively obtain a calibration difference between the residual value and the residual calibration value; After obtaining the residual values of the nodes to be identified, the mean value of the plurality of the residual values is the residual calibration value.

[0041] S470: Compare the calibration difference with a residual threshold, and select the node to be identified corresponding to the calibration difference greater than the residual threshold as a leakage node.

[0042] It can be understood that this leakage node is the pipeline node with a small amount of leakage. After obtaining the predicted correlation degree, compare the predicted correlation degree with a reference correlation degree, generate the residual value, and perform superposition on it. On the basis of the collaborative fluctuation between pipeline nodes, amplify the abnormal signal, capture the tiny collaborative anomaly, and complete the precise positioning of the leakage node.

[0043] After obtaining the original flow rate data set and the original pressure data set, by performing anomaly processing on both of them, it is possible to avoid mis-collection and missed collection caused by the error of the collection sensor during the data collection process, improve the accuracy of the data, and provide an accurate data basis for subsequent leakage detection; identify the pipeline nodes with a large amount of leakage quickly through the difference between the actual flow rate data and the predicted flow rate data; by setting the node pair and introducing the fluctuation correlation degree, identify a small amount of leakage through the pressure coordination between two nodes. Even when the change influence degree of a small amount of leakage on the pressure or flow rate is small, it is still possible to accurately complete the detection of a small amount of leakage through the sudden change of the hydraulic relationship between nodes, and at the same time avoid the influence of water fluctuation or noise on the detection of a small amount of leakage; by performing the equalization processing, ensure the data comparability between different nodes; identify a large amount of leakage through the flow rate and identify a small amount of leakage through the pressure. Compared with the traditional manual detection, it improves the detection efficiency and the detection accuracy at the same time.

[0044] Please refer to Figure 2 , the second embodiment of the present invention provides a water supply network leakage detection system, which is applied to the water supply network leakage detection method in the above embodiment, and the parts that have been described will not be repeated. As used hereinafter, terms such as "module", "unit", "sub-unit", etc. may be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0045] The system includes: The acquisition module 10 is used to obtain the original flow rate data set and the original pressure data set of the pipeline nodes in the water supply network, perform anomaly processing on the original flow rate data set and the original pressure data set to obtain the flow rate data set to be used and the pressure data set to be used. The flow rate data set to be used includes several flow rate data to be used, and the pressure data set to be used includes several pressure data to be used; The acquisition module 10 includes: The first unit is used to obtain the flow rate mean value of several of the flow rate data, and obtain the pressure mean value of several of the pressure data, obtain the flow rate calibration value and the pressure calibration value based on the flow rate mean value and the pressure mean value respectively, obtain the flow rate threshold range through the flow rate mean value and the flow rate calibration value, and obtain the pressure threshold range through the pressure mean value and the pressure calibration value; The second unit is used to compare several of the flow rate data with the flow rate threshold range respectively, select the flow rate data outside the flow rate threshold range as abnormal flow rate data, select the flow rate data within the flow rate threshold range as reserved flow rate data, eliminate the abnormal flow rate data, and combine several of the reserved flow rate data into a reserved flow rate data set; The third unit is used to compare several of the pressure data with the pressure threshold range respectively, select the pressure data outside the pressure threshold range as abnormal pressure data, select the pressure data within the pressure threshold range as reserved pressure data, eliminate the abnormal pressure data, and combine several of the reserved pressure data into a reserved pressure data set; The fourth unit is used to judge whether there is a flow rate missing area in the reserved flow rate data set based on the time frame. If there is a flow rate missing area in the reserved flow rate data set, generate flow rate filling data based on the reserved flow rate data adjacent to the flow rate missing area, fill the flow rate missing area with the flow rate filling data, and select both the reserved flow rate data and the flow rate filling data as the flow rate data to be used to form the flow rate data set to be used; The fifth unit is used to judge whether there is a pressure missing area in the reserved pressure data set based on the time frame. If there is a pressure missing area in the reserved pressure data set, generate pressure filling data based on the reserved pressure data adjacent to the pressure missing area, fill the pressure missing area with the pressure filling data, and select both the reserved pressure data and the pressure filling data as the pressure data to be used to form the pressure data set to be used; The first detection module 20 is used to obtain the predicted flow rate data through the flow rate prediction model, and judge whether there is a large amount of leakage in the pipeline node based on the real-time flow rate data and the predicted flow rate data; An identification module 30, configured to perform equalization processing on the to-be-used pressure data set to obtain an updated pressure data set, where the updated pressure data set includes a number of updated pressure data, combine the pipeline nodes in pairs into a number of node pairs, obtain the fluctuation correlation degree of the node pairs, and determine whether the node pairs are abnormal pairs through the fluctuation correlation degree; The identification module 30 includes: A sixth unit, configured to partition a number of the updated pressure data into a number of window data groups based on a time window, and obtain the window average pressure of the window data groups; A seventh unit, configured to obtain a number of fluctuation correlation degrees corresponding to the node pairs through the updated pressure data and the window average pressure; An eighth unit, configured to obtain the fluctuation correlation difference between adjacent fluctuation correlation degrees, compare the fluctuation correlation difference with a fluctuation threshold, and if the fluctuation correlation difference is greater than the fluctuation threshold, determine that the node pair is an abnormal pair; A second detection module 40, configured to select a number of to-be-identified nodes from all the pipeline nodes based on the pipeline nodes in the abnormal pairs, select all the node pairs corresponding to the to-be-identified nodes as reference pairs, and select the fluctuation correlation degrees corresponding to the reference pairs as reference correlation degrees, obtain the predicted correlation degrees of the reference pairs based on a fluctuation prediction model, obtain the residual values corresponding to the to-be-identified nodes based on the reference correlation degrees and the predicted correlation degrees, and select the leakage nodes from the number of to-be-identified nodes through the residual values to complete a small amount of leakage detection; The second detection module 40 includes: A ninth unit, configured to select both pipeline nodes in the abnormal pair as abnormal nodes, and select all node pairs with the abnormal nodes as to-be-used pairs; A tenth unit, configured to select all pipeline nodes in the abnormal pairs and the to-be-used pairs as to-be-identified nodes; An eleventh unit, configured to obtain the correlation residuals of the reference correlation degrees and the predicted correlation degrees under the same time window; A twelfth unit, configured to stack a number of the correlation residuals into a sub-residual value corresponding to the reference pair; A thirteenth unit, configured to stack a number of the sub-residual values into a residual value corresponding to the to-be-identified node; A fourteenth unit, configured to obtain a residual calibration value through a number of the residual values, and respectively obtain the calibration differences between the residual values and the residual calibration value; A fifteenth unit, configured to compare the calibration differences with a residual threshold, and select the to-be-identified nodes corresponding to the calibration differences greater than the residual threshold as leakage nodes.

[0046] The present invention also provides a computer, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for detecting water supply network leakage as described in the above technical solution is implemented.

[0047] The present invention also provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for detecting water supply network leakage as described in the above technical solution is implemented.

[0048] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0049] The above-described embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the appended claims.

Claims

1. A method for detecting water leakage in a water supply network, characterized in that, The following steps are involved: Acquire an original flow data set and an original pressure data set of a pipeline node in a water supply network, and perform exception processing on the original flow data set and the original pressure data set to acquire a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data, and the standby pressure data set includes a plurality of standby pressure data; Acquire predicted flow data through a flow prediction model, and determine whether there is a large amount of leakage at the pipeline node based on the real-time flow data and the predicted flow data; Performing equalization processing on the standby pressure data set to obtain an updated pressure data set, wherein the updated pressure data set includes a plurality of updated pressure data, combining a plurality of pipeline nodes into a plurality of node pairs, obtaining a fluctuation correlation degree of the node pairs, and determining whether the node pairs are abnormal pairs according to the fluctuation correlation degree; Based on the pipeline nodes in the abnormal pair, several nodes to be identified are selected from all the pipeline nodes, the node pairs corresponding to the nodes to be identified are all selected as reference pairs, and the fluctuation correlation corresponding to the reference pairs is selected as the reference correlation, the predicted correlation of the reference pairs is obtained based on the fluctuation prediction model, the residual value corresponding to the node to be identified is obtained based on the reference correlation and the predicted correlation, and the leakage node is selected from the several nodes to be identified through the residual value to complete a small amount of leakage detection.

2. The method for detecting water leakage in a water supply pipe network according to claim 1, wherein The original flow data set includes a plurality of flow data in a continuous time frame, the original pressure data set includes a plurality of pressure data in a continuous time frame, and the step of performing abnormal processing on the original flow data set and the original pressure data set to obtain a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data and the standby pressure data set includes a plurality of standby pressure data comprises: Obtaining a flow average of a plurality of the flow data, and obtaining a pressure average of a plurality of the pressure data, respectively obtaining a flow calibration value and a pressure calibration value based on the flow average and the pressure average, obtaining a flow threshold range through the flow average and the flow calibration value, and obtaining a pressure threshold range through the pressure average and the pressure calibration value; Compare a plurality of the flow data with the flow threshold range respectively, select the flow data outside the flow threshold range as abnormal flow data, select the flow data within the flow threshold range as retained flow data, remove the abnormal flow data, and combine a plurality of the retained flow data into a retained flow data set; Comparing a plurality of the pressure data with the pressure threshold range respectively, selecting the pressure data outside the pressure threshold range as abnormal pressure data, selecting the pressure data within the pressure threshold range as retained pressure data, eliminating the abnormal pressure data, and combining a plurality of the retained pressure data into a retained pressure data set; Based on the time frame, it is determined whether there is a flow missing area in the reserved flow data set; if there is a flow missing area in the reserved flow data set, flow filling data is generated based on the reserved flow data adjacent to the flow missing area, the flow missing area is filled by the flow filling data, and both the reserved flow data and the flow filling data are selected as standby flow data to form a standby flow data set; Based on the time frame, it is determined whether there is a pressure missing area in the retained pressure data set. If there is a pressure missing area in the retained pressure data set, pressure filling data is generated based on the retained pressure data adjacent to the pressure missing area, and the pressure missing area is filled with the pressure filling data. Both the retained pressure data and the pressure filling data are selected as stand-by pressure data to form a stand-by pressure data set.

3. The water supply network leakage detection method according to claim 2, characterized in that, The formula for obtaining the flow calibration value is: ， Among them, represents the flow calibration value of the i-th pipeline node, represents the total number of time frames, represents the flow data of the i-th pipeline node in the t-th time frame, represents the average flow of the i-th pipeline node.

4. The water supply network leakage detection method according to claim 1, characterized in that The formula for obtaining the updated pressure data is: ， Among them, represents the updated pressure data of the j-th pipeline node at the t-th time frame, represents the standby pressure data of the j-th pipeline node at the t-th time frame, represents the average pressure of the j-th pipeline node, represents the calibrated pressure value of the j-th pipeline node.

5. The water supply network leakage detection method according to claim 1, characterized in that The step of obtaining the fluctuation correlation of the node pair and determining whether the node pair is an abnormal pair according to the fluctuation correlation comprises: Based on a time window, divide the plurality of updated pressure data into a plurality of window data groups, and obtain the window average pressure of the window data groups; Acquire a plurality of fluctuation correlations corresponding to the node pair through the updated pressure data and the window average pressure; Obtain the fluctuation correlation difference between adjacent fluctuation correlations, compare the fluctuation correlation difference with a fluctuation threshold, and if the fluctuation correlation difference is greater than the fluctuation threshold, determine that the node pair is an abnormal pair.

6. The water supply network leakage detection method according to claim 5, wherein The formula for obtaining the fluctuation correlation is: ， wherein, represents the fluctuation correlation degree under the m-th time window when the j-th pipeline node and the k-th pipeline node are a node pair; represents the length of the m-th time window; represents the q-th time frame in the m-th time window; represents the updated pressure data of the j-th pipeline node under the q-th time frame in the m-th time window; represents the window average pressure of the j-th pipeline node under the m-th time window; represents the updated pressure data of the k-th pipeline node under the q-th time frame in the m-th time window; represents the window average pressure of the k-th pipeline node under the m-th time window.

7. The water supply network leakage detection method according to claim 1, characterized in that, The step of selecting a plurality of nodes to be identified from all the pipeline nodes based on the pipeline nodes in the abnormal pair comprises: Selecting both pipeline nodes in the abnormal pair as abnormal nodes, and selecting both node pairs with the abnormal nodes as standby pairs; The pipeline nodes in the abnormal pair and the standby pair are all selected as nodes to be identified.

8. The water supply network leakage detection method according to claim 5, wherein, The step of acquiring the residual value corresponding to the node to be identified based on the reference association degree and the predicted association degree comprises: Obtaining the correlation residual of the reference correlation and the predicted correlation in the same time window; superimposing a plurality of the associated residuals into a sub-residual corresponding to the reference pair; A plurality of the sub-residuals are superimposed to form a residual value corresponding to the node to be identified.

9. The water supply network leakage detection method according to claim 1, characterized in that The step of selecting a leaky node from a plurality of nodes to be identified by using the residual value comprises: Obtaining residual calibration values through a plurality of the residual values, and respectively obtaining calibration differences between the residual values and the residual calibration values; The calibration difference is compared with a residual threshold, and a to-be-identified node corresponding to a calibration difference greater than the residual threshold is selected as a leakage node.

10. A water supply pipe network leakage detection system, which is applied to the water supply pipe network leakage detection method described in any one of claims 1 to 9, and is characterized in that The system comprises: A collection module, used to obtain an original flow data set and an original pressure data set of a pipeline node in a water supply network, and perform exception processing on the original flow data set and the original pressure data set to obtain a standby flow data set and a standby pressure data set, wherein the standby flow data set includes a plurality of standby flow data, and the standby pressure data set includes a plurality of standby pressure data; A first detection module is used to obtain predicted flow data through a flow prediction model, and determine whether there is a large amount of leakage at the pipeline node based on the real-time flow data and the predicted flow data; an identification module, configured to perform equalization processing on the standby pressure data set to obtain an updated pressure data set, wherein the updated pressure data set includes a plurality of updated pressure data, and to combine a plurality of pipeline nodes into a plurality of node pairs in pairs, to obtain a fluctuation correlation degree of the node pairs, and to determine whether the node pairs are abnormal pairs according to the fluctuation correlation degree; The second detection module is used to select several nodes to be identified from all the pipeline nodes based on the pipeline nodes in the abnormal pair, select all node pairs corresponding to the nodes to be identified as reference pairs, and select the fluctuation correlation corresponding to the reference pairs as reference correlation, obtain the predicted correlation of the reference pairs based on the fluctuation prediction model, obtain the residual value corresponding to the node to be identified based on the reference correlation and the predicted correlation, and select the leakage node from the several nodes to be identified through the residual value to complete a small amount of leakage detection.

Images