Method and system for locating leakage points in sewer networks based on multi-source sensing signal analysis

Abstract

This invention discloses a method and system for locating leak points in sewage pipe networks based on multi-source sensor signal analysis, relating to the field of pipe network leak monitoring. The method includes: acquiring the pipe network topology of the target area; deploying acoustic sensor arrays at key nodes; collecting the first vibration signal sequence in response to a leak event; and constructing a sensor response time-series map based on geographic coordinates. It also includes: receiving the first arrival time difference of the first vibration signal, calculating the initial location interval of the leak source, and converging the initial location interval based on the energy attenuation rate; extracting fluid pressure transient signals from adjacent nodes of candidate location pipe segments, comparing waveform distortion, and identifying the location correction based on the reflected wave characteristics of the leak; and within the refined location interval, performing fusion positioning based on the synchronicity between the first vibration signal sequence and the fluid pressure transient signal, and outputting the final location coordinates. This solves the problems of severe monitoring lag, extremely poor positioning accuracy, and complete failure to detect deep leaks without surface manifestations in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of pipeline leakage monitoring, specifically to a method and system for locating leakage points in sewage pipelines based on multi-source sensor signal analysis. Background Technology

[0002] Due to factors such as pipeline aging, construction defects, and geological subsidence, sewage pipe network leakage problems are becoming increasingly prominent. In particular, deep-buried gravity flow main pipes, due to their considerable burial depth, exhibit extremely strong concealment of sewage seepage. In the early stages of leakage, there are no signs on the surface, making them difficult to detect in a timely manner and easily leading to sudden road collapses or serious groundwater pollution.

[0003] Traditional methods rely on manual inspections along pipelines, relying on observation of superficial signs such as damp ground and abnormally lush vegetation to make judgments based on experience. This approach suffers from serious regulatory lag, extremely poor location accuracy, and complete failure to detect deep leaks that have no surface manifestations. Summary of the Invention

[0004] This application provides a method and system for locating leakage points in sewage pipe networks based on multi-source sensor signal analysis, which addresses the problems of serious regulatory lag, extremely poor positioning accuracy, and complete failure to detect deep leakage without surface manifestations in existing technologies.

[0005] In view of the above problems, this application provides a method and system for locating leakage points in sewage pipe networks based on multi-source sensor signal analysis.

[0006] In a first aspect, this application provides a method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis, the method comprising:

[0007] The pipeline topology of the target area is obtained. The first vibration signal sequence output in response to the leakage event is collected by the acoustic sensor array deployed at key nodes. Combined with the geographical coordinates of each sensor node, a sensor response time series map is constructed.

[0008] Based on the sensor response time series spectrum, the initial location interval of the leakage source in the pipeline network topology is calculated by using the difference in the first arrival time of the first vibration signal received by each sensor node. The initial location interval is then converged according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate location pipe segments.

[0009] Fluid pressure transient signals are extracted from adjacent nodes of the candidate positioning pipe segment, and waveform distortion is compared with the pressure fluctuation reference curve to identify the reflected wave characteristics caused by leakage. The position of the candidate positioning pipe segment is corrected based on the reflected wave characteristics to obtain a refined positioning range.

[0010] Within the refined positioning range, multi-source sensor fusion positioning is performed based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, and the final positioning coordinates of the leak point are output.

[0011] Secondly, the present invention provides a sewage pipe network leakage point location system based on multi-source sensor signal analysis, the system comprising:

[0012] The time-series map construction module is used to obtain the pipeline topology of the target area. By deploying acoustic sensor arrays at key nodes, it collects the first vibration signal sequence output in response to leakage events and combines it with the geographical coordinates of each sensor node to construct a sensor response time-series map.

[0013] The candidate positioning pipe segment generation module is used to calculate the initial positioning interval of the leakage source in the pipe network topology based on the sensor response time series spectrum and the difference in the first arrival time of the first vibration signal received by each sensor node, and to converge the initial positioning interval according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate positioning pipe segments.

[0014] The position correction module is used to extract fluid pressure transient signals from adjacent nodes of the candidate positioning pipe segment, compare the waveform distortion with the pressure fluctuation reference curve, identify the reflected wave characteristics caused by leakage, and correct the position of the candidate positioning pipe segment according to the reflected wave characteristics to obtain a refined positioning range.

[0015] The positioning fusion module is used to perform multi-source sensor fusion positioning within the refined positioning range based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, and output the final positioning coordinates of the leak point.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] This application first acquires the pipeline topology of the target area and deploys an acoustic sensor array to construct a sensor response time series map. Second, based on the sensor response time series map, it uses the difference in the first arrival time of the first vibration signal received by each sensor node to calculate the initial positioning interval, and then converges the initial positioning interval to generate candidate positioning pipe segments, effectively eliminating false candidate pipe segments and improving the accuracy of the initial positioning. Third, it compares the fluid pressure transient signal with the pressure fluctuation baseline curve to identify the characteristics of reflected waves caused by leakage, and corrects the position of the candidate positioning pipe segments, achieving precise convergence from pipe segments to specific intervals. Finally, within the refined positioning interval, based on the synchronization between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, it performs multi-source sensor fusion to achieve dynamic weighted fusion of signal quality, improving the accuracy of the positioning results and providing timely and accurate decision-making basis for pipeline network operation and maintenance units, effectively preventing secondary disasters such as road collapse and groundwater pollution caused by leakage. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the sewage network leakage point location method based on multi-source sensor signal analysis proposed in this application.

[0019] Figure 2 This is a schematic diagram of the sewage pipe network leakage point location system based on multi-source sensor signal analysis, as described in this application.

[0020] In the attached diagram, the components represented by each number are as follows:

[0021] The time-series map construction module 11, the candidate positioning pipe segment generation module 12, the position correction module 13, and the positioning fusion module 14 are all included. Detailed Implementation

[0022] This application provides a method for locating leakage points in sewage pipe networks based on multi-source sensor signal analysis, which specifically solves the problems of serious regulatory lag, extremely poor positioning accuracy, and complete failure to address deep leakage without surface manifestations in existing technologies.

[0023] The present invention will now be described in detail with reference to the accompanying drawings.

[0024] Example 1, as Figure 1 As shown, this application provides a method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis, characterized in that the method includes:

[0025] S10: Obtain the pipeline topology of the target area, collect the first vibration signal sequence output in response to the leakage event through the acoustic sensor array deployed at key nodes, and construct the sensor response time series map by combining the geographical coordinates of each sensor node.

[0026] In this embodiment, the pipe network topology is an abstract mathematical model of the interconnection between pipe segments and nodes in the sewage pipe network system within the target area. Nodes represent key locations, and edges represent pipe segments. Key nodes are locations in the pipe network topology that are sensitive to changes in hydraulic state or are easy to monitor. The acoustic sensor array is a collection of multiple acoustic sensors deployed at a single key node. The first vibration signal sequence is time-series data formed by elastic waves excited by leakage events and propagating in the pipe wall or water body, collected by the acoustic sensor array during continuous monitoring. The sensor response time-series graph is a multi-dimensional data visualization and structured representation method.

[0027] Specifically, based on geographic information system (GIS) data of the target area, the precise topology of the sewage pipe network is obtained, clarifying the pipe segment directions, lengths, diameters, and node connections. Subsequently, based on the sewage pipe network topology, acoustic sensor arrays are deployed at key nodes. When leakage occurs in the pipe network, the fluid at the leak point is ejected under pressure, exciting the pipe wall to vibrate and propagate as elastic waves along the pipe wall and the surrounding medium. The acoustic sensor arrays deployed at the key nodes convert this vibration into electrical signals, generating the first vibration signal sequence.

[0028] Subsequently, the arrival time of the first vibration signal sequence collected by each sensor node was extracted. Finally, using the pipeline topology as a spatial framework, the geographical coordinates, arrival times, and signal propagation paths between nodes were correlated and integrated to construct a sensor response time series map, which characterizes the attenuation path of the leakage signal along the pipeline network.

[0029] Step S10 in the method provided in this embodiment of the invention includes:

[0030] Based on geographic information system data of the target area, obtain the pipeline network topology;

[0031] The key nodes are determined based on the pipe segment intersections, pipe diameter change points, and direction inflection points in the pipeline network topology.

[0032] Acoustic sensor arrays are deployed at each key node, wherein the acoustic sensor arrays continuously collect acoustic signals and output a first vibration signal sequence;

[0033] Extract the arrival time of the first vibration signal received by each sensor node, combine it with the geographical coordinates of each sensor node, determine the propagation direction of the leakage signal in the pipeline topology, and establish the temporal correlation between sensor nodes.

[0034] Based on the aforementioned temporal correlation and the spatial distance between sensor nodes in the pipeline network topology, a sensor response time series map is generated with each sensor node as the reference and the first arrival time as the time label.

[0035] Based on the propagation direction, the energy attenuation rate of the leakage signal along the propagation path between adjacent sensor nodes is calculated, and the energy attenuation rate is marked on the corresponding propagation path in the sensing response time series graph.

[0036] In this embodiment of the application, the geographic information system data is a set of digital data stored in the form of a geospatial database, containing the spatial location, attribute information and interrelationships of underground pipeline facilities within the target area; the pipe segment junction is the location where two or more pipes in the pipeline network are connected, such as at a tee or cross; the pipe diameter change point is the location where the pipe diameter changes, usually located at the connection of pipe segments of different specifications; the direction turning point is the location where the direction of the pipeline changes significantly, such as at a bend.

[0037] First, based on the Geographic Information System (GIS) data of the target area, the pipeline survey data archived by the municipal pipeline management department of the target area is retrieved to extract the spatial coordinates, direction, connection relationship, pipe diameter, and node attribute information of the pipe segments, forming the basic data source for the pipeline network topology. Then, each manhole or pipe segment connection point is defined as a node, and the pipes between adjacent nodes are defined as edges. Based on the sewage flow direction, directional attributes are assigned to the edges, thereby constructing the complete pipeline network topology.

[0038] Secondly, sensors should be deployed at points where the physical structure of the pipeline network changes significantly. This is because acoustic signals are significantly reflected, refracted, or attenuated at pipe junctions, abrupt changes in pipe diameter, or turning points in flow direction. These locations are natural anomalies on the propagation path of leakage signals, and deploying sensors at these locations can maximize the capture of signal changes.

[0039] Specifically, after obtaining the complete pipeline network topology, the intersections, diameter change points, and directional inflection points of the pipeline network topology are analyzed according to preset screening rules. First, the intersections of pipeline segments where signals from different directions converge or branch are identified; deploying sensors can cover multiple pipeline segments. Second, all diameter change points are identified; since sudden changes in pipe diameter cause signal reflection, deploying sensors helps to capture the characteristics of reflected waves. Third, all directional inflection points where the signal propagation characteristics differ from those of straight pipeline segments are identified.

[0040] Nodes that meet the criteria are extracted and marked as key nodes, serving as the basis for subsequent deployment of acoustic sensor arrays. For example, in a pipeline network containing 50 inspection wells, 15 key nodes are identified, and sensors only need to be deployed at the intersections of pipe segments and points where the pipe diameter changes.

[0041] Furthermore, after identifying the key nodes, acoustic sensor arrays are installed at these nodes. Each array contains multiple sensors arranged in predetermined geometric positions to ensure different response characteristics to waves arriving from different directions. After installation, the sensor arrays are driven to continuously acquire vibration signals from the pipe wall or water body at a high sampling rate, convert them into digital signals, and organize them into a first vibration signal sequence in chronological order. This first vibration signal sequence includes environmental background noise and abnormal vibration signals that may be caused by leakage events.

[0042] Furthermore, the first arrival time is the absolute time point at which the vibration wave generated by the leakage event first arrives at the sensor node, identified from the first vibration signal sequence; the propagation direction is the main path of the vibration wave generated by the leakage source in the pipeline network, usually from the sensor that first receives the signal to the sensor that receives the signal later.

[0043] Specifically, the first vibration signal sequence output by each sensor node is analyzed in real time to calculate the current ambient background noise level and set a trigger threshold. Then, the geographic coordinates of all nodes are spatiotemporally matched with their first arrival times, identifying the node with the smallest first arrival time as the first response node and the node with the second smallest as the second response node. Based on the connection relationship and geographic coordinates of these two nodes in the pipeline topology, it is determined that the leakage signal propagates from the first response node to the second response node. Then, the propagation velocity between the two nodes is calculated. Finally, the coordinates of the first and second response nodes, the propagation time difference, and the propagation velocity are recorded as correlation data to form a temporal correlation relationship between the nodes.

[0044] Furthermore, after establishing temporal correlations, a sensor response time-series map is constructed. First, the pipeline topology is used as the skeleton of the map. Then, the geographic coordinates of each sensor node are mapped onto this skeleton to form the spatial reference of the map. Next, the first arrival time extracted from each sensor node is used as the time label for that node and marked next to the node. For node pairs with established temporal correlations, the propagation direction, the propagation time difference between nodes, and the energy attenuation rate are marked on the connecting pipeline path to generate the sensor response time-series map.

[0045] Finally, based on the propagation direction, the energy attenuation rate of the leakage signal along the propagation path between adjacent sensor nodes is calculated, and the energy attenuation rate is marked on the corresponding propagation path in the sensor response time series diagram. Here, the propagation path is the physical channel through which the leakage signal propagates from the leakage source to the sensor node in the pipeline network; the energy attenuation rate is the degree of energy loss of the vibration signal after it propagates a certain distance along the pipe segment. The energy attenuation rate can be obtained by calculating the attenuation ratio of the vibration amplitude of the first vibration signal received by the later node to the vibration amplitude of the first vibration signal received by the previous node in the adjacent sensor nodes.

[0046] Specifically, after determining the propagation direction, energy analysis is performed on the first vibration signal sequence between adjacent sensor nodes, such as the first response node and the second response node.

[0047] The energy of the signal at the first response node is extracted; for example, the root mean square value of the signal within a specific time window is calculated as the initial energy. The energy of the signal at the second response node is extracted as the received energy. Simultaneously, the spatial distance between these two nodes is obtained from the network topology. Then, the energy attenuation rate is calculated according to a formula. After calculation, the energy attenuation rate is used as an attribute value and marked on the propagation path connecting the two nodes in the sensor response time-series graph.

[0048] In step S10 of the method provided in this embodiment of the invention, the arrival time of the first vibration signal received by each sensor node is extracted, and the propagation direction of the leakage signal in the pipeline topology is determined by combining the geographical coordinates of each sensor node, and the temporal correlation relationship between the sensor nodes is established, including:

[0049] From the first vibration signal sequence output by each sensor node, the moment when the vibration amplitude first exceeds the environmental noise threshold is extracted as the first arrival moment when each sensor node receives the first vibration signal.

[0050] The geographic coordinates of each sensor node are spatiotemporally matched with the first arrival time, and the sensor node with the smallest first arrival time is identified as the first response node, and the sensor node with the second smallest first arrival time is identified as the second response node.

[0051] The direction of leakage signal propagation is determined by the direction from the first response node to the second response node, combined with the pipe segment orientation between the first response node and the second response node in the pipeline network topology.

[0052] According to the propagation direction, the difference between the first arrival time of the first response node and the first arrival time of the second response node is taken as the propagation time difference between nodes;

[0053] The ratio of the spatial distance between the first response node and the second response node in the pipeline topology to the propagation time difference between the nodes is taken as the propagation speed between nodes.

[0054] The geographic coordinates of the first response node, the geographic coordinates of the second response node, the propagation time difference between the nodes, and the propagation speed between the nodes are associated and recorded to establish a temporal correlation relationship between the sensor nodes.

[0055] In this embodiment, vibration amplitude data is first collected for each sensor node over a fixed period of time. The root mean square of the vibration amplitude data is calculated using the statistical characteristics of the data within the statistical analysis window, and this root mean square is used as the ambient background noise level at the location. An appropriate ambient noise threshold is then set. This threshold is not a fixed value but is dynamically adjusted based on factors such as day-night variations to ensure a balance between sensitivity to leakage events and the ability to resist false alarms.

[0056] The amplitude of each sampling point in the first vibration signal sequence is compared with an environmental noise threshold. When the signal amplitude is detected to exceed the environmental noise threshold for the first time, the timestamp corresponding to the sampling point is recorded as the first arrival time of the first vibration signal received by the corresponding sensor node.

[0057] Secondly, after extracting the first arrival times of all sensor nodes, for each responding sensor node, its geographical coordinates are associated and integrated with the corresponding first arrival time. Then, all responding nodes are sorted in ascending order using the first arrival time as the key.

[0058] After sorting, the node at the first position in the sequence, i.e., the node with the smallest first arrival time, is identified as the first response node; the node at the second position in the sequence, i.e., the node with the second smallest first arrival time, is identified as the second response node. For example, in a leakage event, sensor nodes A, B, and C are all triggered, with their first arrival times of 0.5 seconds, 1.2 seconds, and 2.1 seconds, respectively. In this case, node A is identified as the first response node, and node B is identified as the second response node.

[0059] Furthermore, the vector pointing from the first response node to the second response node indicates the geometric direction of the leakage signal propagation. However, since the vibration wave propagates along the pipe, its actual path must conform to the constraints of the pipe network topology. Therefore, the pipe network topology is queried to obtain the physical pipe segments and their directions that exist between the first and second response nodes.

[0060] If there is a directly connected pipe segment between two nodes, the propagation direction is the direction of that pipe segment; if there is a path consisting of multiple pipe segments between two nodes, the path from the first response node, through a series of pipe segments, and finally to the second response node is extracted and used as the final direction of leakage signal propagation.

[0061] Furthermore, since the signal arrives at the first response node first and then the second response node, the difference between the arrival times of the first and second response nodes is taken as the propagation time difference between the nodes, representing the time it takes for the vibration wave to propagate along the pipe from the first response node to the second response node. For example, if the arrival time of the first response node A is tA = 0.5s and the arrival time of the second response node B is tB = 1.2s, then the propagation time difference between the nodes Δt = tB - tA = 0.7s.

[0062] Furthermore, the geographical coordinates of the first response node and the second response node in the pipeline network topology are obtained. Based on the pipeline segment connection relationship between the two nodes recorded in the pipeline network topology, the actual path length along the pipeline segment between the first response node and the second response node is extracted. This actual path length is used as the spatial distance. The spatial distance is obtained by accumulating the lengths of all pipeline segments connecting the two nodes to ensure that the distance value is completely consistent with the actual propagation path of the signal.

[0063] Then, using the propagation time difference between nodes as the denominator and the spatial distance as the numerator, the ratio of the two is calculated to obtain the actual speed of the vibration wave propagating in the pipe section between the first and second response nodes. For example, if the spatial distance between nodes A and B is 350 meters and the propagation time difference between nodes is 0.7 seconds, then the calculated propagation speed between nodes is 500 meters per second.

[0064] Finally, the geographical coordinates of the first response node, the geographical coordinates of the second response node, the propagation time difference between nodes, and the propagation speed between nodes are correlated to form a temporal correlation record, establishing the temporal correlation relationship between sensor nodes. For example, for a certain leakage event, the temporal correlation record established by the system is as follows: first node (coordinates X1, Y1), second node (coordinates X2, Y2), propagation time difference 0.7s, propagation speed 500m / s.

[0065] In this embodiment, the pipeline topology is acquired based on geographic information system data, and key nodes are determined according to pipe segment intersections, diameter change points, and directional turning points to enable the selection of sensor deployment locations and reduce construction complexity. The first arrival time is extracted by combining a first vibration signal sequence with an environmental noise threshold, and the first and second response nodes are identified to achieve automated identification of the leakage event response order. The propagation direction is determined by the pipeline topology, and the propagation time difference and propagation speed between nodes are calculated to establish a structured temporal correlation, enabling adaptive real-time calculation of wave velocity. Furthermore, the energy attenuation rate along the propagation path is marked in the sensor response time-series map to construct a unified spatiotemporal data structure, providing an efficient data interface for subsequent positioning algorithms.

[0066] S20: Based on the sensing response time series spectrum, the initial location interval of the leakage source in the pipeline network topology is calculated by using the difference in the first arrival time of the first vibration signal received by each sensor node, and the initial location interval is converged according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate location pipe segments.

[0067] In this embodiment, the inversion calculation is based on known observation results and combined with known physical models to deduce unknown source parameters in reverse; the leakage source is the source of the vibration signal, that is, the physical location where leakage actually occurs in the sewage pipe network; the initial positioning interval is a region containing the leakage point obtained through inversion calculation; the energy attenuation rate is the rate at which the energy of the vibration signal decreases as the propagation distance increases when it propagates in the pipe section; the candidate positioning pipe section is the single pipe section within the initial positioning interval that is considered most likely to leak after being screened by the energy attenuation rate.

[0068] Specifically, after obtaining the sensor response time series map, the at least two sensor nodes that first received the signal are first extracted, and the time difference is calculated based on the difference in the first arrival times of the nodes. A hyperbolic trajectory is constructed on a two-dimensional plane or pipeline topology. The hyperbolic trajectory is spatially overlaid with the pipeline topology to construct an initial positioning interval. To further converge the positioning results, the energy attenuation rate of signal propagation on each candidate pipeline segment is extracted from the sensor response time series map. By comparing the energy attenuation rates of each pipeline segment within the initial positioning interval, the pipeline segment with the smallest attenuation rate is selected as the candidate positioning pipeline segment.

[0069] Step S20 in the method provided in this embodiment of the invention includes:

[0070] The first response node is extracted from the sensing response time sequence graph as the first reference node, and the second response node is extracted as the second reference node.

[0071] Obtain the inter-node propagation speed between the first reference node and the second reference node, and multiply the difference between the first arrival time of the first reference node and the first arrival time of the second reference node by the inter-node propagation speed.

[0072] The difference between the first distance between the first reference node and the leakage wave source and the second distance between the second reference node and the leakage wave source is calculated as the first distance difference;

[0073] A hyperbolic trajectory is established based on the first reference node and the second reference node;

[0074] The hyperbolic trajectory is spatially overlaid with the pipeline network topology, and the set of pipe segments where the hyperbolic trajectory intersects with the pipeline network segments is extracted as the initial positioning interval.

[0075] Extract each pipe segment from the initial positioning interval and obtain the sensor nodes connected to both ends of each pipe segment;

[0076] Extract the energy attenuation rate marked on the propagation path between the sensor nodes at both ends of each pipe segment from the sensing response time sequence graph.

[0077] Compare the energy attenuation rates of each pipe segment and select the pipe segment with the smallest energy attenuation rate as the candidate positioning pipe segment.

[0078] In this embodiment, the response node information for the current leakage event is first extracted from the constructed sensor response time-series map. Based on the first arrival time, the node with the smallest first arrival time is extracted and defined as the first reference node. Simultaneously, the node with the second smallest first arrival time is extracted and defined as the second reference node. Subsequently, the first and second response nodes are selected as reference nodes. The reference nodes are the first to sense the leakage event, have the highest signal-to-noise ratio, and their first arrival time extraction is the most reliable, providing the most accurate input parameters for subsequent positioning.

[0079] For example, suppose that in a leakage event, sensor nodes A, B, C, and D are triggered in sequence. The system automatically extracts node A as the first reference node and node B as the second reference node, without selecting node C or node D, even though they may also provide useful information. However, the combination of A and B can give the most accurate time difference constraint.

[0080] Secondly, the inter-node propagation velocity between the first and second reference nodes is obtained from the temporal correlation. Then, the difference between the arrival times of the first and second reference nodes is calculated to obtain the inter-node propagation time difference. Multiplying the inter-node propagation velocity by the inter-node propagation time difference yields a product that is numerically equal to the spatial distance between the first and second reference nodes. For example, if the inter-node propagation velocity is 500 m / s and the propagation time difference is 0.7 s, the result is 350 m.

[0081] Furthermore, according to the physical principles of vibration wave propagation, the vibration waves generated by the leakage source will propagate in all directions at the same speed. Therefore, the difference in the arrival times of the vibration waves at the first and second reference nodes directly reflects the difference in the path length from the wave source to the two nodes.

[0082] Specifically, the difference between the distance from the leakage wave source to the second reference node and the distance from the leakage wave source to the first reference node is taken as the first distance difference. This first distance difference is equal to the wave velocity multiplied by the difference in the first arrival times of the two nodes. For example, if the propagation speed between the nodes is 500 m / s and the first arrival time difference is 0.7 s, then the first distance difference is 350 m.

[0083] Furthermore, the geographic coordinates of the first reference node are taken as the first focus of the hyperbola, and the geographic coordinates of the second reference node are taken as the second focus of the hyperbola. The first distance difference is taken as the difference in distance from any point on the hyperbola to the two focuses. The set of points that satisfy the conditions is used to form a hyperbola, and the length of the real axis, the length of the imaginary axis, and the coordinates of the center point of the hyperbola are determined to construct the hyperbola trajectory.

[0084] Furthermore, the hyperbolic trajectory is spatially overlaid with the pipeline network topology, and the set of pipe segments where the hyperbolic trajectory intersects with the pipeline network segments is extracted as the initial positioning interval.

[0085] Specifically, the hyperbola trajectory is used as one spatial layer and the pipeline network topology is used as another spatial layer. Under a unified geographic coordinate system, the constructed hyperbola trajectory is superimposed on the pipeline network topology map to identify the pipe segments traversed by the hyperbola trajectory, extract all pipe segments that intersect with the hyperbola trajectory, and form the initial positioning interval.

[0086] Furthermore, after determining the initial positioning interval, for each candidate pipe segment within the initial positioning interval, the node information connected to both ends is queried from the pipeline network topology. Since sensors are deployed at key nodes, the upstream and downstream nodes of each pipe segment are recorded through an acoustic sensor array to obtain the signal propagation characteristics between these nodes later. For example, for candidate pipe segment L1, its two ends are connected to nodes A and B, respectively; for candidate pipe segment L2, its two ends are connected to nodes C and D, respectively, establishing a clear index relationship for subsequent energy attenuation rate comparison.

[0087] Furthermore, for each candidate pipe segment, the propagation path connecting the sensor nodes at both ends is located in the sensor response time series graph based on the sensor nodes connected at both ends. The total energy attenuation rate from one end node to the other can be obtained by accumulating the attenuation rates of each segment along the path. If the two ends of the candidate pipe segment are adjacent nodes, the energy attenuation rate marked on the propagation path in the graph is directly extracted; if there are other nodes between the two ends, the attenuation rates of each segment along the path need to be accumulated.

[0088] Finally, the energy attenuation rates of the extracted candidate pipe segments are sorted and compared. Since the energy of the vibration signal attenuates with increasing distance during propagation, if the leak point is located on a certain pipe segment, the energy attenuation rate along the propagation path between the sensor nodes at both ends of the segment should theoretically be less than that of other pipe segments. Therefore, the pipe segment with the smallest energy attenuation rate is selected as the candidate positioning pipe segment. For example, if the initial positioning interval includes three pipe segments L1, L2, and L3, with energy attenuation rates of 0.5dB / m, 1.2dB / m, and 1.8dB / m respectively, the system selects the L1 pipe segment with the smallest energy attenuation rate as the candidate positioning pipe segment.

[0089] In step S20 of the method provided in this embodiment of the invention, establishing a hyperbolic trajectory based on the first reference node and the second reference node includes:

[0090] The geographic coordinates of the first reference node and the geographic coordinates of the second reference node are respectively used as the first focus and the second focus of the hyperbola;

[0091] The spatial distance between the first reference node and the second reference node in the pipeline topology is taken as the focal length of the hyperbola.

[0092] The first distance difference is taken as the difference between the distance from any point on the hyperbola to the first focus and the distance to the second focus;

[0093] Based on the difference between the focal length and the distance, the real axis length and imaginary axis length of the hyperbola are determined, and the hyperbola trajectory is constructed.

[0094] In this embodiment, the geographic coordinates of the first and second reference nodes are first obtained from the sensing response time-series graph. The geographic coordinates of the first reference node are set as the first focus of the hyperbola, and the geographic coordinates of the second reference node are set as the second focus of the hyperbola. The relative positions of the first and second foci in space determine the orientation and basic shape of the hyperbola. Since the vibration signal generated by the leakage source is first sensed by the first reference node and then by the second reference node, the leakage point is located on the hyperbola with the two nodes as foci, and is located on the branch closer to the first reference node. For example, if the geographic coordinates of the first reference node A are (0,0) and the geographic coordinates of the second reference node B are (350,0), then the system sets these two points as the first focus F1 (0,0) and the second focus F2 (350,0) of the hyperbola, respectively.

[0095] Secondly, the spatial distance between the first and second reference nodes is extracted from the pipeline topology, and the actual path length of this segment along the pipeline is used as the focal length of the hyperbola. For example, the straight-line distance between node A and node B may be only 300 meters, but the actual pipeline length is 350 meters, so 350 meters is used as the focal length.

[0096] Furthermore, according to the physical principle of vibration wave propagation, the vibration waves generated by the leakage wave source propagate in all directions at the same speed. Therefore, the difference between the distance from the wave source to the first reference node and the distance to the second reference node is equal to the wave speed multiplied by the difference in the arrival times of the two nodes, i.e., the first distance difference. For example, if the first distance difference is 350 meters, then the difference in distance from any point on the hyperbola to F1 and F2 is always 350 meters.

[0097] Finally, after determining the coordinates, focal lengths, and the first distance difference between the two foci, the hyperbolic trajectory is constructed. First, the length of the real semi-axis, a = first distance difference / 2, is calculated. Then, based on the geometric relationship of the hyperbola, c² = a² + b², the length of the imaginary semi-axis, b, is calculated. Next, the coordinates of the hyperbola's center point are determined. The direction of the hyperbola's real axis is determined based on the direction of the line connecting the foci. Finally, in the center point coordinate system, a complete hyperbolic trajectory is constructed, typically containing two branches. Based on the signal propagation direction, the branch closer to the first reference node is selected as the valid trajectory.

[0098] In this embodiment, the first and second response nodes are extracted as reference nodes, and the first distance difference is determined based on the difference between the propagation speed between nodes and the first arrival time, avoiding the positioning error introduced by using a fixed theoretical wave speed. By constructing a hyperbolic trajectory, precise geometric constraints are provided for subsequent spatial overlay analysis. By spatially overlaying the hyperbolic trajectory with the pipeline topology, an initial positioning interval is obtained, achieving convergence of physical pipe segments and narrowing the search range for leak points. Furthermore, sensor nodes connected to both ends of each pipe segment are extracted from the initial positioning interval, the energy attenuation rate corresponding to each pipe segment is extracted, and the pipe segment with the smallest energy attenuation rate is selected as the candidate positioning pipe segment, achieving convergence from multiple pipe segments to a single candidate pipe segment, significantly enhancing the reliability of positioning.

[0099] S30: Extract fluid pressure transient signals from adjacent nodes of the candidate positioning pipe segment, compare the waveform distortion with the pressure fluctuation reference curve, identify the reflected wave characteristics caused by leakage, and correct the position of the candidate positioning pipe segment based on the reflected wave characteristics to obtain a refined positioning range.

[0100] In this embodiment, the fluid pressure transient signal is a dynamic signal that causes the fluid pressure in the pipe to fluctuate violently in a very short time due to events such as valve opening and closing, pump start-up and shutdown, or sudden leakage during the operation of the pipeline network; the pressure fluctuation reference curve is used to characterize the ideal fluctuation shape of pressure over time under normal operation of the pipeline network without any external disturbances; waveform distortion comparison is the process of comparing and analyzing the actual collected fluid pressure transient signal with the pressure fluctuation reference curve; the reflected wave feature is a specific waveform distortion originating from the pressure wave generated by the leakage point; the refined positioning interval is based on the candidate positioning pipe segment, and by using the reflected wave feature and calculating the propagation time of the reflected wave, the search range is refined from the entire pipe segment to a certain interval on the pipe segment.

[0101] Specifically, after identifying candidate positioning pipe sections, high-precision fluid pressure transient signals are extracted from adjacent nodes. A pre-established pressure fluctuation reference curve is retrieved. The actual collected pressure signal is then compared with this reference curve using waveform overlay and distortion analysis. In the absence of leakage, the pressure signal should closely match the reference curve. When leakage occurs, the reflected wave is superimposed on the original pressure waveform, forming a reflected wave characteristic. The starting time of the reflected wave is then extracted, and the distance between the sensor node and the leakage point is calculated to determine the refined positioning interval.

[0102] Step S30 in the method provided in this embodiment of the invention includes:

[0103] Fluid pressure transient signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment to obtain upstream pressure signals and downstream pressure signals.

[0104] The upstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the upstream pressure signal relative to the pressure fluctuation reference curve.

[0105] The pressure fluctuation reference curve is used to characterize the pressure fluctuation pattern over time under leak-free conditions. It is obtained by statistical fitting of the historical pressure signal of the target area pipeline network under normal operating conditions. The fluctuation boundary of the reference curve is defined by the amplitude of the environmental pressure fluctuation, and the pressure wave propagation period is defined by the time interval between adjacent peaks.

[0106] When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the upstream pressure signal is marked as having reflected wave characteristics.

[0107] The downstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the downstream pressure signal relative to the pressure fluctuation reference curve.

[0108] When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the downstream pressure signal is marked as having reflected wave characteristics.

[0109] In this embodiment, after determining the candidate location pipe segment, the sensor nodes connected to both ends of the segment are identified based on the pipeline network topology. Since acoustic sensors and pressure sensors are installed on the key nodes where sensors are deployed, the fluid pressure transient signal corresponding to the time window of the current leakage event is extracted from the pressure sensor at the upstream node as the upstream pressure signal. Similarly, the corresponding signal is extracted from the pressure sensor at the downstream node as the downstream pressure signal.

[0110] Secondly, the upstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the upstream pressure signal relative to the pressure fluctuation reference curve.

[0111] Among them, waveform distortion amplitude is calculated on the same time axis after waveform overlay, the amplitude difference between the upstream pressure signal curve and the pressure fluctuation reference curve is calculated, and the maximum value in the difference sequence is taken as waveform distortion amplitude, which characterizes the maximum deviation of the reflected wave caused by leakage from the normal pressure fluctuation.

[0112] The distortion duration is calculated from the moment when the upstream pressure signal curve first deviates from the pressure fluctuation reference curve and exceeds the amplitude of the environmental pressure fluctuation, until the moment when the signal curve returns to the fluctuation boundary of the reference curve. The time between the start and end times is used as the distortion duration, which characterizes the duration of the impact of the reflected wave on the pressure waveform.

[0113] Specifically, a pre-established pressure fluctuation reference curve is retrieved, and the upstream pressure signal waveform is overlaid on the same time axis as the pressure fluctuation reference curve. Under normal operating conditions without leakage, the actual pressure signal should closely match the reference curve; when leakage occurs, the negative pressure wave generated at the leakage point and its reflected wave will be superimposed on the original waveform, forming obvious waveform distortion, thereby determining whether there is a pressure anomaly caused by leakage.

[0114] By comparing points one by one or performing waveform correlation analysis, the waveform distortion amplitude of the actual pressure signal relative to the reference curve is calculated, that is, the maximum amplitude difference between the actual waveform and the reference curve. Simultaneously, the duration of distortion is calculated, that is, the time elapsed from when the waveform begins to deviate from the reference curve until it reverts to the reference curve.

[0115] For example, if the upstream pressure signal shows a peak with an amplitude of 0.15 MPa at t=10.2 seconds, while the fluctuation amplitude of the baseline curve at that moment is only 0.02 MPa, then the distortion amplitude is 0.13 MPa and the distortion duration is 0.3 seconds.

[0116] The pressure fluctuation baseline curve is used to characterize the pressure fluctuation pattern over time under leak-free conditions. It is obtained by statistical fitting of the historical pressure signal of the target area pipeline network under normal operating conditions. The fluctuation boundary of the baseline curve is defined by the amplitude of the environmental pressure fluctuation, and the pressure wave propagation period is defined by the time interval between adjacent peaks.

[0117] Specifically, the ambient pressure fluctuation amplitude is the normal range of pressure fluctuations caused by factors such as pump start-up and shutdown, valve adjustment, etc., under normal operating conditions without leakage. It is determined by statistical analysis of historical pressure signals, such as calculating the standard deviation or percentiles, and serves as a threshold for judging whether abnormal fluctuations have occurred. The pressure wave propagation period is the characteristic period of the periodic fluctuations formed by the propagation and reflection of pressure waves in the pipeline network under normal operating conditions. It is obtained by spectral analysis or peak detection of historical pressure signals and is usually equal to the time required for the pressure wave to complete one round trip propagation in the pipeline network.

[0118] The mean and standard deviation of the pressure signal sequence are calculated. The mean plus or minus three times the standard deviation is used as the normal fluctuation boundary of the pressure signal, and three times the standard deviation is taken as the ambient pressure fluctuation amplitude. The pressure wave propagation period is obtained by the time interval between adjacent peaks. For example, by analyzing 30 consecutive days of normal operating pressure data from a pipeline network, the ambient pressure fluctuation amplitude is found to be ±0.03 MPa, and the pressure wave propagation period is 2.5 seconds.

[0119] Furthermore, after calculating the waveform distortion amplitude and duration, the presence of reflected wave characteristics in the upstream pressure signal is determined according to preset judgment rules. The criteria include whether the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and whether the distortion duration is less than the pressure wave propagation period. If both conditions are met, the upstream pressure signal is determined to contain reflected wave characteristics caused by leakage, and this characteristic is marked with information including the time of reflection, distortion amplitude, and duration. If either condition is not met, the reflected wave characteristic is determined to be absent.

[0120] For example, a distortion amplitude of 0.12 MPa was detected in the upstream pressure signal, which exceeds the environmental fluctuation amplitude of 0.03 MPa. The distortion duration was 0.3 seconds, which is less than the pressure wave propagation period of 2.5 seconds. It was determined that there were reflected wave characteristics, and the start time of the reflected wave was marked as t=10.2 seconds.

[0121] Furthermore, using the same method, the downstream pressure signal is overlaid with the pressure fluctuation reference curve. Through point-by-point comparison or waveform correlation analysis, the waveform distortion amplitude of the downstream pressure signal relative to the reference curve is calculated, i.e., the maximum amplitude difference between the actual waveform and the reference curve. Simultaneously, the distortion duration is calculated, i.e., the time elapsed from the start of the waveform deviating from the reference curve to its return to the reference curve, providing a data basis for subsequent identification of reflected wave characteristics. For example, a distortion amplitude of 0.11 MPa was detected in the downstream pressure signal at t=10.8 seconds, with a distortion duration of 0.35 seconds.

[0122] Furthermore, based on the waveform distortion amplitude and duration, the system determines whether reflected wave characteristics exist in the downstream pressure signal. The system determines that the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period. When both conditions are met simultaneously, the system determines that reflected wave characteristics exist in the downstream pressure signal and marks these characteristics. For example, if the downstream pressure signal meets the determination criteria, the system marks the start time of the reflected wave as t=10.8 seconds.

[0123] If any condition is not met, it is determined that there are no reflected wave characteristics. Furthermore, reflected wave characteristics may exist simultaneously in the upstream and downstream sections, or they may exist only in one end, depending on the specific location of the leak point on the pipe section and the deployment of the sensors.

[0124] Step S30 in the method provided in this embodiment of the invention further includes:

[0125] When there are reflected wave characteristics in the upstream pressure signal, the start time of the reflected wave is extracted from the upstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the upstream node is taken as the upstream reflection time difference. The upstream reflection time difference is multiplied by the propagation speed between nodes to obtain the upstream reflection distance between the upstream sensor node and the leakage point.

[0126] When there are reflected wave characteristics in the downstream pressure signal, the start time of the reflected wave is extracted from the downstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the downstream node is taken as the downstream reflection time difference. The downstream reflection time difference is multiplied by the propagation speed between nodes to obtain the downstream reflection distance between the downstream sensor node and the leakage point.

[0127] Starting from the upstream node of the candidate positioning pipe segment, the upstream reflection distance is extended along the pipe segment direction into the interior of the candidate positioning pipe segment to obtain the upstream positioning boundary;

[0128] Taking the downstream node of the candidate positioning pipe segment as the endpoint, the downstream reflection distance is extended in the opposite direction to the interior of the candidate positioning pipe segment along the pipe segment to obtain the downstream positioning boundary;

[0129] The pipe section between the upstream positioning boundary and the downstream positioning boundary is used as the refined positioning interval.

[0130] In this embodiment, when there are reflected wave characteristics in the upstream pressure signal, the start time of the reflected wave is extracted from the upstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the upstream node is taken as the upstream reflection time difference. The upstream reflection time difference is multiplied by the propagation speed between nodes to obtain the upstream reflection distance between the upstream sensor node and the leakage point.

[0131] Specifically, firstly, the normal arrival time of the pressure wave refers to the theoretical arrival time of the pressure wave from the leak point along the pipe segment to the corresponding sensor node. Starting from the leak source and ending at the sensor node, the propagation path of the pressure wave is determined along the pipe segment in the pipeline topology. The ratio of the spatial distance of the propagation path to the propagation speed between nodes is obtained, and this ratio is used as the pressure wave propagation time. The normal arrival time of the pressure wave is then obtained by adding the time of the leak event to the pressure wave propagation time.

[0132] Subsequently, by analyzing the rate of change of the first or second derivative of the waveform, the point where the waveform first significantly deviates from the baseline curve is identified, thus obtaining the start time of the reflected wave from the waveform data of the upstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave is calculated to obtain the upstream reflection time difference. According to the principle of wave propagation, this time difference is equal to the time required for the reflected wave to propagate from the leak point to the upstream sensor node. Therefore, multiplying the upstream reflection time difference by the propagation velocity between nodes yields the upstream reflection distance between the upstream sensor node and the leak point, which serves as a parameter for subsequently determining the upstream positioning boundary.

[0133] For example, if the start time of the reflected wave in the upstream pressure signal is t=10.2 seconds and the normal arrival time of the pressure wave is t=9.8 seconds, then the upstream reflection time difference is 0.4 seconds. Multiplying this by the propagation speed between nodes, which is 500m / s, we get the upstream reflection distance as 200 meters, meaning the leak point is located 200 meters downstream of the upstream sensor node.

[0134] Similarly, when reflected wave characteristics exist in the downstream pressure signal, the same method is used to analyze the rate of change of the first or second derivative of the waveform, identifying the point where the waveform first significantly deviates from the reference curve, thus obtaining the start time of the reflected wave extracted from the downstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the downstream node is used as the downstream reflection time difference. Multiplying the downstream reflection time difference by the propagation velocity between nodes yields the downstream reflection distance between the downstream sensor node and the leak point, where the downstream reflection distance represents the distance the leak point is upstream of the downstream sensor node.

[0135] For example, if the start time of the reflected wave in the downstream pressure signal is t=10.8 seconds and the normal arrival time of the pressure wave is t=10.5 seconds, then the downstream reflection time difference is 0.3 seconds. Multiplying this by the propagation speed between nodes, which is 500m / s, we get the downstream reflection distance as 150 meters, meaning the leak point is located 150 meters upstream of the downstream sensor node.

[0136] Secondly, after obtaining the upstream reflection distance, the upstream boundary of the refined positioning interval is constructed. First, the upstream node position of the candidate positioning pipe segment is determined by obtaining the geographic coordinates of the upstream node position from the geographic information system data. Then, the pipe segment direction is determined according to the pipeline network topology. Starting from the upstream node, the length of the upstream reflection distance is measured along the pipe segment direction, and the position point at the end of the upstream reflection distance is marked as the upstream positioning boundary.

[0137] According to the principles of wave propagation, the time it takes for the reflected wave from the leak point to travel to the upstream node is the upstream reflection time difference. Therefore, the leak point must be located downstream of the upstream node at the upstream reflection distance. The leak point cannot be located between the upstream node and the upstream positioning boundary; the upstream positioning boundary is the upstream limit of the possible locations of the leak point. For example, if the candidate positioning pipe segment is 500 meters long, the upstream node coordinates are (0,0), the pipe segment direction is the positive x-axis, and the upstream reflection distance is 200 meters, then the upstream positioning boundary coordinates are (200,0), and the leak point is downstream of this point.

[0138] Next, after obtaining the downstream reflection distance, the downstream boundary of the refined positioning interval is also constructed. This determines the downstream node position of the candidate positioning pipe segment. Then, using the downstream node as the endpoint, the length of the downstream reflection distance is measured along the opposite direction of the pipe segment, and the endpoint is marked as the downstream positioning boundary.

[0139] According to the principle of wave propagation, the time it takes for the reflected wave generated at the leak point to travel from the leak point to the downstream node is the downstream reflection time difference. Therefore, the leak point must be located at the downstream reflection distance upstream of the downstream node. The leak point cannot be located between the downstream positioning boundary and the downstream node; the downstream positioning boundary is the downstream limit of the possible locations of the leak point. For example, if the downstream node coordinates of the candidate positioning pipe segment are (500,0) and the downstream reflection distance is 150 meters, extending 150 meters backward from the downstream node, the downstream positioning boundary coordinates are (350,0), and the leak point is located upstream of the node.

[0140] Ultimately, the pipe section between the upstream and downstream positioning boundaries is used as the refined positioning range.

[0141] Specifically, the refined positioning range begins at the upstream positioning boundary and ends at the downstream positioning boundary. The length of the pipe segment between the two boundaries is equal to the total length of the candidate positioning pipe segment minus the upstream reflection distance and then the downstream reflection distance.

[0142] If the sum of the upstream and downstream reflection distances is less than the total length of the candidate positioning pipe segment, the refined positioning interval is a non-empty pipe segment interval; if the sum equals the total length, the refined positioning interval shrinks to a single point; if the sum is greater than the total length, it indicates a potential contradiction in the identification of upstream and downstream reflected wave characteristics, triggering a verification mechanism. For example, if the total length of the candidate positioning pipe segment is 500 meters, the upstream reflection distance is 200 meters, and the downstream reflection distance is 150 meters, then the refined positioning interval is a 150-meter pipe segment interval from 200 meters to 350 meters from the upstream node.

[0143] In this embodiment, transient fluid pressure signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment. These signals are then overlaid with a pressure fluctuation reference curve to calculate the waveform distortion amplitude and duration, thus establishing the pressure fluctuation reference curve. Reflected wave characteristics are identified through dual judgment conditions to eliminate persistent pressure fluctuation interference, improving the specificity and accuracy of reflected wave identification. Furthermore, the upstream and downstream reflection distances are calculated to convert reflected wave characteristics into spatial distance. Starting from the upstream node, the upstream positioning boundary is obtained; conversely, starting from the downstream node, the downstream positioning boundary is obtained. The pipe segment interval between these two points is used as a refined positioning interval, achieving convergence from the candidate positioning pipe segment to the specific interval.

[0144] S40: Within the refined positioning range, based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, multi-source sensor fusion positioning is performed, and the final positioning coordinates of the leak point are output.

[0145] In this embodiment, the spectral centroid offset is the change in the spectral centroid relative to the initial state or the previous moment; the fluctuation period is the time interval between adjacent peaks in the fluid pressure transient signal; synchronicity is the inherent physical relationship between the rate of change of the spectral centroid of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal; multi-source sensor fusion positioning is a method of comprehensively processing information from different types of sensors to collaboratively determine the final location; the final positioning coordinates are the precise location of the leak point output after initial positioning, candidate pipe segment screening, and refined interval positioning.

[0146] Specifically, after obtaining the refined positioning range, the first vibration signal sequence collected by the acoustic sensor array and the fluid pressure transient signal collected by the pressure sensor are extracted from the upstream and downstream sensor nodes corresponding to the refined positioning range, respectively. Then, the vibration and pressure signals from the upstream and downstream are analyzed. The ratio of the rate of change of the spectral centroid to the pressure fluctuation period is calculated to obtain the upstream and downstream synchronization parameters. Finally, using the geographical coordinates of the upstream and downstream sensors as a reference, the fused coordinate point is calculated as the final output of the leak point coordinates.

[0147] Step S40 in the method provided in this embodiment of the invention includes:

[0148] The first vibration signal sequence is extracted from the upstream sensor node and the downstream sensor node corresponding to the refined positioning interval, respectively, to obtain the upstream vibration signal sequence and the downstream vibration signal sequence;

[0149] The rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate;

[0150] The fluctuation period is extracted from the upstream pressure signal to obtain the upstream pressure fluctuation period; the fluctuation period is extracted from the downstream pressure signal to obtain the downstream pressure fluctuation period.

[0151] Calculate the ratio of the upstream spectral centroid change rate to the upstream pressure fluctuation period to obtain the upstream synchronicity parameter; and calculate the ratio of the downstream spectral centroid change rate to the downstream pressure fluctuation period to obtain the downstream synchronicity parameter.

[0152] The ratio of the upstream synchronization parameter to the downstream synchronization parameter is used as the fusion weight;

[0153] Obtain the geographic coordinates of the upstream sensor node and the geographic coordinates of the downstream sensor node;

[0154] The fusion weight is used as the contribution coefficient of the upstream sensor node. The product of the geographical coordinates of the upstream sensor node and the contribution coefficient is calculated to obtain the upstream weighted coordinates.

[0155] Calculate the difference between 1 and the fusion weight as the contribution coefficient of the downstream sensor node, and calculate the product of the geographical coordinates of the downstream sensor node and the downstream contribution coefficient to obtain the downstream weighted coordinates.

[0156] The upstream weighted coordinates and the downstream weighted coordinates are added together to obtain a fused coordinate point, which is then used as the final location coordinates of the leak point.

[0157] In this embodiment, after determining the refined positioning interval, the candidate positioning pipe segment where the interval is located is identified according to the pipeline network topology, and the sensor nodes connected to both ends of the pipe segment are obtained. From the acoustic sensor array at the upstream node, a first vibration signal sequence corresponding to the time window of the current leakage event is extracted as the upstream vibration signal sequence. Similarly, the corresponding signal is extracted from the downstream node as the downstream vibration signal sequence. The extracted time window is typically centered on the arrival time of the first response node, extending forward and backward by a certain time range to ensure complete capture of the vibration signals caused by the leakage event.

[0158] Secondly, the centroid of the spectrum is the central position of the signal energy distribution in the frequency domain; the rate of change of the centroid of the spectrum is the rate of change of the centroid offset over time, that is, the amount of change of the centroid of the spectrum per unit time.

[0159] Specifically, the vibration signal sequence is divided into multiple consecutive subsequences according to a fixed time window. Then, the spectral centroid frequency of each subsequence is calculated, and the spectral centroid offset is obtained. Dividing the spectral centroid offset by the time window length yields the spectral centroid change rate, which reflects the rate of change of the vibration signal frequency characteristics over time. By calculating the spectral centroid change rates upstream and downstream respectively, quantitative indicators characterizing the dynamic characteristics of the vibration signals at both ends are obtained.

[0160] Furthermore, waveform analysis is performed on the upstream pressure signal to identify peaks and troughs, and the time interval between adjacent peaks or troughs is calculated. This time interval is taken as the upstream pressure fluctuation period. The same method is used for the downstream pressure signal to identify peaks and troughs and calculate the time interval between adjacent peaks or troughs, which is also taken as the downstream pressure fluctuation period. For example, upstream pressure signal analysis shows that the time interval between adjacent peaks is stable at approximately 0.2 seconds, meaning the pressure fluctuation period is 0.2 seconds; downstream pressure signal analysis shows that the pressure fluctuation period is 0.21 seconds.

[0161] Furthermore, dividing the upstream spectral centroid change rate by the upstream pressure fluctuation period yields the upstream synchronicity parameter. Similarly, dividing the downstream spectral centroid change rate by the downstream pressure fluctuation period yields the downstream synchronicity parameter. Under stable leakage conditions, there is an inherent physical relationship between the two; the jet at the leakage outlet is both a vibration source and a pressure wave source. Therefore, the change in vibration frequency and the period of the pressure wave should be synchronized.

[0162] For example, the upstream synchronicity parameter is 100 / 0.2 = 500; the downstream synchronicity parameter is 80 / 0.21 ≈ 381. The upstream synchronicity parameter is greater than the downstream one, indicating that the upstream signal is more synchronized with the leakage event.

[0163] Furthermore, the upstream synchronization parameter is divided by the downstream synchronization parameter to obtain the fusion weight.

[0164] Furthermore, the geographic coordinates of the upstream sensor node and the downstream sensor node are obtained from the geographic information system database. For example, the geographic coordinates of upstream node A are (100, 200), and the geographic coordinates of downstream node B are (500, 200).

[0165] Furthermore, the fusion weight is used as the contribution coefficient of the upstream sensor nodes. Then, the geographical coordinates of the upstream nodes are multiplied by the contribution coefficient to obtain the upstream weighted coordinates. The larger the weight, the greater the influence of the sensor node in the subsequent summation.

[0166] Further, the difference between 1 and the fusion weight is calculated as the contribution coefficient of the downstream sensor node. Then, the geographical coordinates of the downstream node are multiplied by the downstream contribution coefficient to obtain the downstream weighted coordinates. The downstream normalized contribution coefficient = |1 - fusion weight|.

[0167] Finally, the upstream weighted coordinates and the downstream weighted coordinates are vector-added to obtain the merged coordinate point. This merged coordinate point is the final location coordinate of the leak point. For example, adding the upstream weighted coordinates (131, 262) and the downstream weighted coordinates (155, 62) yields the merged coordinate point (386, 328).

[0168] In step S40 of the method provided in this embodiment of the invention, the rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate, including:

[0169] The upstream vibration signal sequence and the downstream vibration signal sequence are divided into multiple upstream subsequences and multiple downstream subsequences according to the same time window, wherein the length of the time window is determined according to the pressure wave propagation period and is less than the pressure wave propagation period;

[0170] Perform spectral analysis on each upstream subsequence, calculate the spectral centroid frequency of each upstream subsequence, and obtain the upstream centroid frequency sequence;

[0171] Calculate the ratio of the difference in frequency between adjacent upstream centroids to the length of the time window to obtain the rate of change of the upstream spectral centroid.

[0172] Perform spectral analysis on each downstream subsequence, calculate the spectral centroid frequency of each downstream subsequence, and obtain the downstream centroid frequency sequence;

[0173] The ratio of the difference between adjacent downstream centroid frequencies to the length of the time window is calculated to obtain the rate of change of the downstream spectral centroid.

[0174] In this embodiment, the upstream and downstream vibration signal sequences are first subjected to the same framing process. The length of the time window is determined based on the pressure wave propagation period. The length of the time window is typically taken as 1 / 3 to 1 / 2 of the pressure wave propagation period to ensure that the signal segments contained within each time window can reflect the details of the vibration frequency changes caused by the leakage event.

[0175] The upstream vibration signal sequence is divided into multiple consecutive subsequences according to the time window length. These subsequences can be non-overlapping or partially overlapping. Similarly, the downstream vibration signal sequence is divided into the same number of downstream subsequences, with the time interval of each downstream subsequence strictly aligned with the corresponding upstream subsequence. This results in multiple upstream and multiple downstream subsequences. For example, if the pressure wave propagation period is 0.2 seconds, the time window length is set to 0.08 seconds, and the sampling rate is 1000Hz. Each window contains 80 sampling points, dividing the 5-second vibration signal sequence into approximately 62 subsequences.

[0176] Secondly, spectral analysis is performed on each upstream subsequence. Specifically, for each subsequence, the time-domain signal is windowed by multiplying each sampling point in the subsequence with the corresponding window function coefficient to suppress spectral leakage caused by signal truncation. Then, a Fast Fourier Transform (FFT) is performed to convert the time-domain signal into a complex sequence in the frequency domain, obtaining the frequency domain representation of the subsequence. The transform result includes the real and imaginary parts, and the amplitude and phase spectra at each frequency point are calculated accordingly. Next, the power spectral density is calculated based on the frequency domain transform result, i.e., the amplitude at each frequency point is squared to obtain the energy of the frequency component.

[0177] According to the definition of the centroid of the spectrum, calculate the weighted average value of each frequency point: multiply each frequency value by its corresponding power spectral density, sum them and divide by the sum of the total power spectral densities to obtain the centroid frequency of the spectrum of the subsequence.

[0178] Repeat the above calculation for all subsequences to obtain a series of spectral centroid frequency values ​​arranged in chronological order, which constitute the upstream centroid frequency sequence, reflecting the change law of the frequency centroid of the upstream vibration signal over time. For example, spectral analysis of 62 subsequences of the upstream vibration signal yields the upstream centroid frequency sequence [200Hz, 202Hz, 205Hz, 210Hz, 218Hz, 225Hz, 232Hz, 240Hz].

[0179] Next, for each pair of adjacent centroid frequency values ​​in the upstream subsequence, calculate the difference between the latter and the former value to obtain the spectral centroid offset. Then divide by the time window length to obtain the spectral centroid change rate within the corresponding time window. Repeat the calculation for all adjacent window pairs to obtain the spectral centroid change rate values. If the centroid frequency sequence has N values, the change rate sequence has N-1 values.

[0180] Similarly, following the method of spectral analysis of upstream subsequences, spectral analysis is performed on each downstream subsequence to calculate the spectral centroid frequency of each downstream subsequence, thus obtaining the downstream centroid frequency sequence. For example, spectral analysis of 62 subsequences of the downstream vibration signal yields the downstream centroid frequency sequence [195Hz, 196Hz, 198Hz, 201Hz, 206Hz, 210Hz, 215Hz, 219Hz].

[0181] Finally, for each pair of adjacent centroid frequency values ​​in the downstream subsequence, the difference between adjacent downstream centroid frequencies is calculated to obtain the spectral centroid offset. This is then divided by the time window length to obtain the spectral centroid change rate within the corresponding time window. This calculation is repeated for all adjacent window pairs to obtain the downstream spectral centroid change rate. For example, if the difference between adjacent downstream centroid frequencies is 5 Hz and the time window length is 0.08 seconds, then the downstream spectral centroid change rate is 62.5 Hz / s, while the upstream change rate is 95 Hz / s, which is greater than the downstream change rate.

[0182] In this embodiment, vibration and pressure signal sequences are extracted from upstream and downstream sensor nodes corresponding to a refined positioning interval, respectively. The time window length is determined based on the pressure wave propagation period, achieving adaptive matching between the time window length and the dynamic characteristics of the leakage event. Spectral analysis is performed on each sub-sequence to calculate the spectral centroid frequency, and then the spectral centroid change rate is calculated to extract the dynamic characteristics of the vibration signal frequency characteristics evolving over time. Subsequently, the fluctuation period is extracted from the pressure signal, and synchronization parameters are calculated. The ratio of the upstream and downstream synchronization parameters is used as the fusion weight, providing a clear physical basis for weight allocation. Finally, the weights are multiplied by the normalized weights and summed to obtain the fused coordinate point as the final positioning coordinate of the leakage point. This achieves deep fusion positioning of vibration and pressure signals, improving the accuracy and engineering practicality of the fusion positioning.

[0183] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0184] In this embodiment, the pipeline topology is first obtained based on geographic information system data. Key nodes are then determined according to pipe segment intersections, diameter change points, and directional turning points to facilitate sensor deployment and reduce construction complexity. The first arrival time is extracted by combining a first vibration signal sequence with an environmental noise threshold, and the first and second response nodes are identified, enabling automated identification of the leakage event response order. The propagation direction is determined using the pipeline topology, and the propagation time difference and propagation speed between nodes are calculated to establish a structured temporal correlation, enabling adaptive real-time calculation of wave velocity. Furthermore, the energy attenuation rate along the propagation path is marked in the sensor response time-series map, constructing a unified spatiotemporal data structure to provide an efficient data interface for subsequent positioning algorithms.

[0185] Secondly, the first and second response nodes are extracted as reference nodes, and the first distance difference is determined based on the difference between the propagation speed between nodes and the first arrival time, avoiding the positioning error introduced by using a fixed theoretical wave speed. By constructing a hyperbolic trajectory, precise geometric constraints are provided for subsequent spatial overlay analysis. By spatially overlaying the hyperbolic trajectory with the pipeline network topology, an initial positioning interval is obtained, achieving convergence of physical pipe segments and narrowing the search range for leak points. Furthermore, sensor nodes connected to both ends of each pipe segment are extracted from the initial positioning interval, the energy attenuation rate corresponding to each pipe segment is extracted, and the pipe segment with the smallest energy attenuation rate is selected as the candidate positioning pipe segment, achieving convergence from multiple pipe segments to a single candidate pipe segment, significantly enhancing the reliability of positioning.

[0186] Next, transient fluid pressure signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment. These signals are then overlaid with the pressure fluctuation baseline curve to calculate the waveform distortion amplitude and duration, thus establishing the pressure fluctuation baseline curve. Reflected wave characteristics are identified through dual judgment conditions to eliminate persistent pressure fluctuation interference, improving the specificity and accuracy of reflected wave identification. Furthermore, the upstream and downstream reflection distances are calculated to convert reflected wave characteristics into spatial distance. Starting from the upstream node, the upstream positioning boundary is obtained; conversely, starting from the downstream node, the downstream positioning boundary is obtained. The pipe segment interval between these two points is used as the refined positioning interval, achieving convergence from the candidate positioning pipe segment to the specific interval.

[0187] Finally, by extracting vibration and pressure signal sequences from upstream and downstream sensor nodes corresponding to the refined positioning interval, and determining the time window length based on the pressure wave propagation period, an adaptive match between the time window length and the dynamic characteristics of the leakage event is achieved. Spectral analysis is performed on each sub-sequence to calculate the spectral centroid frequency, and then the spectral centroid change rate is calculated to extract the dynamic characteristics of the vibration signal frequency characteristics evolving over time. Subsequently, the fluctuation period is extracted from the pressure signal, and the synchronization parameter is calculated. The ratio of the upstream and downstream synchronization parameters is used as the fusion weight, giving the weight allocation a clear physical basis. Finally, the weights are multiplied by the normalized weights and summed to obtain the fused coordinate point as the final positioning coordinate of the leakage point. This achieves deep fusion positioning of vibration and pressure signals, improving the accuracy and engineering practicality of the fusion positioning.

[0188] Example 2, as Figure 2 As shown, based on the same inventive concept as the sewage network leakage point location method based on multi-source sensor signal analysis provided in Embodiment 1, this embodiment of the invention also provides a sewage network leakage point location system based on multi-source sensor signal analysis, the system comprising:

[0189] The time-series map construction module 11 is used to obtain the pipeline topology of the target area. Through the acoustic sensor array deployed at key nodes, it collects the first vibration signal sequence output in response to the leakage event, and combines the geographical coordinates of each sensor node to construct the sensor response time-series map.

[0190] The candidate positioning pipe segment generation module 12 is used to calculate the initial positioning interval of the leakage source in the pipe network topology based on the sensing response time sequence map and the difference in the first arrival time of the first vibration signal received by each sensor node, and to converge the initial positioning interval according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate positioning pipe segments.

[0191] The position correction module 13 is used to extract fluid pressure transient signals from adjacent nodes of the candidate positioning pipe segment, compare the waveform distortion with the pressure fluctuation reference curve, identify the reflected wave characteristics caused by leakage, and correct the position of the candidate positioning pipe segment according to the reflected wave characteristics to obtain a refined positioning range.

[0192] The positioning fusion module 14 is used to perform multi-source sensor fusion positioning within the refined positioning range based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, and output the final positioning coordinates of the leak point.

[0193] In one embodiment, the time series graph construction module 11 is used for:

[0194] Based on geographic information system data of the target area, obtain the pipeline network topology;

[0195] The key nodes are determined based on the pipe segment intersections, pipe diameter change points, and direction inflection points in the pipeline network topology.

[0196] Acoustic sensor arrays are deployed at each key node, wherein the acoustic sensor arrays continuously collect acoustic signals and output a first vibration signal sequence;

[0197] Extract the arrival time of the first vibration signal received by each sensor node, combine it with the geographical coordinates of each sensor node, determine the propagation direction of the leakage signal in the pipeline topology, and establish the temporal correlation between sensor nodes.

[0198] Based on the aforementioned temporal correlation and the spatial distance between sensor nodes in the pipeline network topology, a sensor response time series map is generated with each sensor node as the reference and the first arrival time as the time label.

[0199] Based on the propagation direction, the energy attenuation rate of the leakage signal along the propagation path between adjacent sensor nodes is calculated, and the energy attenuation rate is marked on the corresponding propagation path in the sensing response time series graph.

[0200] This involves extracting the arrival time of the first vibration signal received by each sensor node, combining this with the geographical coordinates of each sensor node, determining the propagation direction of the leakage signal in the pipeline topology, and establishing the temporal correlation between the sensor nodes, including:

[0201] From the first vibration signal sequence output by each sensor node, the moment when the vibration amplitude first exceeds the environmental noise threshold is extracted as the first arrival moment when each sensor node receives the first vibration signal.

[0202] The geographic coordinates of each sensor node are spatiotemporally matched with the first arrival time, and the sensor node with the smallest first arrival time is identified as the first response node, and the sensor node with the second smallest first arrival time is identified as the second response node.

[0203] The direction of leakage signal propagation is determined by the direction from the first response node to the second response node, combined with the pipe segment orientation between the first response node and the second response node in the pipeline network topology.

[0204] According to the propagation direction, the difference between the first arrival time of the first response node and the first arrival time of the second response node is taken as the propagation time difference between nodes;

[0205] The ratio of the spatial distance between the first response node and the second response node in the pipeline topology to the propagation time difference between the nodes is taken as the propagation speed between nodes.

[0206] The geographic coordinates of the first response node, the geographic coordinates of the second response node, the propagation time difference between the nodes, and the propagation speed between the nodes are associated and recorded to establish a temporal correlation relationship between the sensor nodes.

[0207] In one embodiment, the candidate positioning pipe segment generation module 12 is used for:

[0208] The first response node is extracted from the sensing response time sequence graph as the first reference node, and the second response node is extracted as the second reference node.

[0209] Obtain the inter-node propagation speed between the first reference node and the second reference node, and multiply the difference between the first arrival time of the first reference node and the first arrival time of the second reference node by the inter-node propagation speed.

[0210] The difference between the first distance between the first reference node and the leakage wave source and the second distance between the second reference node and the leakage wave source is calculated as the first distance difference;

[0211] A hyperbolic trajectory is established based on the first reference node and the second reference node;

[0212] The hyperbolic trajectory is spatially overlaid with the pipeline network topology, and the set of pipe segments where the hyperbolic trajectory intersects with the pipeline network segments is extracted as the initial positioning interval.

[0213] Extract each pipe segment from the initial positioning interval and obtain the sensor nodes connected to both ends of each pipe segment;

[0214] Extract the energy attenuation rate marked on the propagation path between the sensor nodes at both ends of each pipe segment from the sensing response time sequence graph.

[0215] Compare the energy attenuation rates of each pipe segment and select the pipe segment with the smallest energy attenuation rate as the candidate positioning pipe segment.

[0216] The process of establishing a hyperbolic trajectory based on the first reference node and the second reference node includes:

[0217] The geographic coordinates of the first reference node and the geographic coordinates of the second reference node are respectively used as the first focus and the second focus of the hyperbola;

[0218] The spatial distance between the first reference node and the second reference node in the pipeline topology is taken as the focal length of the hyperbola.

[0219] The first distance difference is taken as the difference between the distance from any point on the hyperbola to the first focus and the distance to the second focus;

[0220] Based on the difference between the focal length and the distance, the real axis length and imaginary axis length of the hyperbola are determined, and the hyperbola trajectory is constructed.

[0221] In one embodiment, the position correction module 13 is used for:

[0222] Fluid pressure transient signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment to obtain upstream pressure signals and downstream pressure signals.

[0223] The upstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the upstream pressure signal relative to the pressure fluctuation reference curve.

[0224] The pressure fluctuation reference curve is used to characterize the pressure fluctuation pattern over time under leak-free conditions. It is obtained by statistical fitting of the historical pressure signal of the target area pipeline network under normal operating conditions. The fluctuation boundary of the reference curve is defined by the amplitude of the environmental pressure fluctuation, and the pressure wave propagation period is defined by the time interval between adjacent peaks.

[0225] When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the upstream pressure signal is marked as having reflected wave characteristics.

[0226] The downstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the downstream pressure signal relative to the pressure fluctuation reference curve.

[0227] When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the downstream pressure signal is marked as having reflected wave characteristics.

[0228] In one embodiment, the position correction module 13 is further configured to:

[0229] When there are reflected wave characteristics in the upstream pressure signal, the start time of the reflected wave is extracted from the upstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the upstream node is taken as the upstream reflection time difference. The upstream reflection time difference is multiplied by the propagation speed between nodes to obtain the upstream reflection distance between the upstream sensor node and the leakage point.

[0230] When there are reflected wave characteristics in the downstream pressure signal, the start time of the reflected wave is extracted from the downstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the downstream node is taken as the downstream reflection time difference. The downstream reflection time difference is multiplied by the propagation speed between nodes to obtain the downstream reflection distance between the downstream sensor node and the leakage point.

[0231] Starting from the upstream node of the candidate positioning pipe segment, the upstream reflection distance is extended along the pipe segment direction into the interior of the candidate positioning pipe segment to obtain the upstream positioning boundary;

[0232] Taking the downstream node of the candidate positioning pipe segment as the endpoint, the downstream reflection distance is extended in the opposite direction to the interior of the candidate positioning pipe segment along the pipe segment to obtain the downstream positioning boundary;

[0233] The pipe section between the upstream positioning boundary and the downstream positioning boundary is used as the refined positioning interval.

[0234] In one embodiment, the positioning fusion module 14 is used for:

[0235] The first vibration signal sequence is extracted from the upstream sensor node and the downstream sensor node corresponding to the refined positioning interval, respectively, to obtain the upstream vibration signal sequence and the downstream vibration signal sequence;

[0236] The rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate;

[0237] The fluctuation period is extracted from the upstream pressure signal to obtain the upstream pressure fluctuation period; the fluctuation period is extracted from the downstream pressure signal to obtain the downstream pressure fluctuation period.

[0238] Calculate the ratio of the upstream spectral centroid change rate to the upstream pressure fluctuation period to obtain the upstream synchronicity parameter; and calculate the ratio of the downstream spectral centroid change rate to the downstream pressure fluctuation period to obtain the downstream synchronicity parameter.

[0239] The ratio of the upstream synchronization parameter to the downstream synchronization parameter is used as the fusion weight;

[0240] Obtain the geographic coordinates of the upstream sensor node and the geographic coordinates of the downstream sensor node;

[0241] The fusion weight is used as the contribution coefficient of the upstream sensor node. The product of the geographical coordinates of the upstream sensor node and the contribution coefficient is calculated to obtain the upstream weighted coordinates.

[0242] Calculate the difference between 1 and the fusion weight as the contribution coefficient of the downstream sensor node, and calculate the product of the geographical coordinates of the downstream sensor node and the downstream contribution coefficient to obtain the downstream weighted coordinates.

[0243] The upstream weighted coordinates and the downstream weighted coordinates are added together to obtain a fused coordinate point, which is then used as the final location coordinates of the leak point.

[0244] Specifically, the rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate, including:

[0245] The upstream vibration signal sequence and the downstream vibration signal sequence are divided into multiple upstream subsequences and multiple downstream subsequences according to the same time window, wherein the length of the time window is determined according to the pressure wave propagation period and is less than the pressure wave propagation period;

[0246] Perform spectral analysis on each upstream subsequence, calculate the spectral centroid frequency of each upstream subsequence, and obtain the upstream centroid frequency sequence;

[0247] Calculate the ratio of the difference in frequency between adjacent upstream centroids to the length of the time window to obtain the rate of change of the upstream spectral centroid.

[0248] Perform spectral analysis on each downstream subsequence, calculate the spectral centroid frequency of each downstream subsequence, and obtain the downstream centroid frequency sequence;

[0249] The ratio of the difference between adjacent downstream centroid frequencies to the length of the time window is calculated to obtain the rate of change of the downstream spectral centroid.

[0250] Compared to existing technologies, this application's embodiments first acquire the pipeline network topology based on geographic information system data, and determine key nodes according to pipe segment intersections, diameter change points, and directional turning points to enable sensor deployment location selection and reduce construction complexity. By combining a first vibration signal sequence with an environmental noise threshold, the first arrival time is extracted, and the first and second response nodes are identified, achieving automated identification of the leakage event response order. The propagation direction is determined through the pipeline network topology, and the propagation time difference and propagation speed between nodes are calculated to establish a structured temporal correlation, enabling adaptive real-time calculation of wave velocity. Furthermore, the energy attenuation rate along the propagation path is marked in the sensor response time-series map, constructing a spatiotemporally unified data structure to provide an efficient data interface for subsequent positioning algorithms.

[0251] Secondly, the first and second response nodes are extracted as reference nodes, and the first distance difference is determined based on the difference between the propagation speed between nodes and the first arrival time, avoiding the positioning error introduced by using a fixed theoretical wave speed. By constructing a hyperbolic trajectory, precise geometric constraints are provided for subsequent spatial overlay analysis. By spatially overlaying the hyperbolic trajectory with the pipeline network topology, an initial positioning interval is obtained, achieving convergence of physical pipe segments and narrowing the search range for leak points. Furthermore, sensor nodes connected to both ends of each pipe segment are extracted from the initial positioning interval, the energy attenuation rate corresponding to each pipe segment is extracted, and the pipe segment with the smallest energy attenuation rate is selected as the candidate positioning pipe segment, achieving convergence from multiple pipe segments to a single candidate pipe segment, significantly enhancing the reliability of positioning.

[0252] Next, transient fluid pressure signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment. These signals are then overlaid with the pressure fluctuation baseline curve to calculate the waveform distortion amplitude and duration, thus establishing the pressure fluctuation baseline curve. Reflected wave characteristics are identified through dual judgment conditions to eliminate persistent pressure fluctuation interference, improving the specificity and accuracy of reflected wave identification. Furthermore, the upstream and downstream reflection distances are calculated to convert reflected wave characteristics into spatial distance. Starting from the upstream node, the upstream positioning boundary is obtained; conversely, starting from the downstream node, the downstream positioning boundary is obtained. The pipe segment interval between these two points is used as the refined positioning interval, achieving convergence from the candidate positioning pipe segment to the specific interval.

[0253] Finally, by extracting vibration and pressure signal sequences from upstream and downstream sensor nodes corresponding to the refined positioning interval, and determining the time window length based on the pressure wave propagation period, an adaptive match between the time window length and the dynamic characteristics of the leakage event is achieved. Spectral analysis is performed on each sub-sequence to calculate the spectral centroid frequency, and then the spectral centroid change rate is calculated to extract the dynamic characteristics of the vibration signal frequency characteristics evolving over time. Subsequently, the fluctuation period is extracted from the pressure signal, and the synchronization parameter is calculated. The ratio of the upstream and downstream synchronization parameters is used as the fusion weight, giving the weight allocation a clear physical basis. Finally, the weights are multiplied by the normalized weights and summed to obtain the fused coordinate point as the final positioning coordinate of the leakage point. This achieves deep fusion positioning of vibration and pressure signals, improving the accuracy and engineering practicality of the fusion positioning.

Claims

1. A method for locating a leak in a sewer network based on multi-source sensor signal analysis, characterized in that, The method includes: The pipeline topology of the target area is obtained. The first vibration signal sequence output in response to the leakage event is collected by the acoustic sensor array deployed at key nodes. Combined with the geographical coordinates of each sensor node, a sensor response time series map is constructed. Based on the sensor response time series spectrum, the initial location interval of the leakage source in the pipeline network topology is calculated by using the difference in the first arrival time of the first vibration signal received by each sensor node. The initial location interval is then converged according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate location pipe segments. Fluid pressure transient signals are extracted from adjacent nodes of the candidate positioning pipe segment, and waveform distortion is compared with the pressure fluctuation reference curve to identify the reflected wave characteristics caused by leakage. The position of the candidate positioning pipe segment is corrected based on the reflected wave characteristics to obtain a refined positioning range. Within the refined positioning range, based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, multi-source sensor fusion positioning is performed to output the final positioning coordinates of the leak point, including: The first vibration signal sequence is extracted from the upstream sensor node and the downstream sensor node corresponding to the refined positioning interval, respectively, to obtain the upstream vibration signal sequence and the downstream vibration signal sequence; The rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate; The fluctuation period is extracted from the upstream pressure signal to obtain the upstream pressure fluctuation period; the fluctuation period is extracted from the downstream pressure signal to obtain the downstream pressure fluctuation period. Calculate the ratio of the upstream spectral centroid change rate to the upstream pressure fluctuation period to obtain the upstream synchronicity parameter; and calculate the ratio of the downstream spectral centroid change rate to the downstream pressure fluctuation period to obtain the downstream synchronicity parameter. The ratio of the upstream synchronization parameter to the downstream synchronization parameter is used as the fusion weight; Obtain the geographic coordinates of the upstream sensor node and the geographic coordinates of the downstream sensor node; The fusion weight is used as the contribution coefficient of the upstream sensor node. The product of the geographical coordinates of the upstream sensor node and the contribution coefficient is calculated to obtain the upstream weighted coordinates. Calculate the difference between 1 and the fusion weight as the contribution coefficient of the downstream sensor node, and calculate the product of the geographical coordinates of the downstream sensor node and the downstream contribution coefficient to obtain the downstream weighted coordinates. The upstream weighted coordinates and the downstream weighted coordinates are added together to obtain a fused coordinate point, which is then used as the final location coordinates of the leak point.

2. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 1, characterized in that, The pipeline topology of the target area is obtained. Acoustic sensor arrays deployed at key nodes are used to collect the first vibration signal sequence output in response to leakage events. Combined with the geographical coordinates of each sensor node, a sensor response time-series map is constructed, including: Based on geographic information system data of the target area, obtain the pipeline network topology; The key nodes are determined based on the pipe segment intersections, pipe diameter change points, and direction inflection points in the pipeline network topology. Acoustic sensor arrays are deployed at each key node, wherein the acoustic sensor arrays continuously collect acoustic signals and output a first vibration signal sequence; Extract the arrival time of the first vibration signal received by each sensor node, combine it with the geographical coordinates of each sensor node, determine the propagation direction of the leakage signal in the pipeline topology, and establish the temporal correlation between sensor nodes. Based on the aforementioned temporal correlation and the spatial distance between sensor nodes in the pipeline network topology, a sensor response time series map is generated with each sensor node as the reference and the first arrival time as the time label. Based on the propagation direction, the energy attenuation rate of the leakage signal along the propagation path between adjacent sensor nodes is calculated, and the energy attenuation rate is marked on the corresponding propagation path in the sensing response time series graph.

3. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 2, characterized in that, Extract the arrival time of the first vibration signal received by each sensor node, and combine it with the geographical coordinates of each sensor node to determine the propagation direction of the leakage signal in the pipeline topology, and establish the temporal correlation between the sensor nodes, including: From the first vibration signal sequence output by each sensor node, the moment when the vibration amplitude first exceeds the environmental noise threshold is extracted as the first arrival moment when each sensor node receives the first vibration signal. The geographic coordinates of each sensor node are spatiotemporally matched with the first arrival time, and the sensor node with the smallest first arrival time is identified as the first response node, and the sensor node with the second smallest first arrival time is identified as the second response node. The direction of leakage signal propagation is determined by the direction from the first response node to the second response node, combined with the pipe segment orientation between the first response node and the second response node in the pipeline network topology. According to the propagation direction, the difference between the first arrival time of the first response node and the first arrival time of the second response node is taken as the propagation time difference between nodes; The ratio of the spatial distance between the first response node and the second response node in the pipeline topology to the propagation time difference between the nodes is taken as the propagation speed between nodes. The geographic coordinates of the first response node, the geographic coordinates of the second response node, the propagation time difference between the nodes, and the propagation speed between the nodes are associated and recorded to establish a temporal correlation relationship between the sensor nodes.

4. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 1, characterized in that, Based on the sensor response time-series map, the initial location interval of the leakage source in the pipeline network topology is calculated by utilizing the difference in the first arrival time of the first vibration signal received by each sensor node. The initial location interval is then converged based on the energy attenuation rate of the first vibration signal sequence along its propagation path to generate candidate location pipe segments, including: The first response node is extracted from the sensing response time sequence graph as the first reference node, and the second response node is extracted as the second reference node. Obtain the inter-node propagation speed between the first reference node and the second reference node, and multiply the difference between the first arrival time of the first reference node and the first arrival time of the second reference node by the inter-node propagation speed. The difference between the first distance between the first reference node and the leakage wave source and the second distance between the second reference node and the leakage wave source is calculated as the first distance difference; A hyperbolic trajectory is established based on the first reference node and the second reference node; The hyperbolic trajectory is spatially overlaid with the pipeline network topology, and the set of pipe segments where the hyperbolic trajectory intersects with the pipeline network segments is extracted as the initial positioning interval. Extract each pipe segment from the initial positioning interval and obtain the sensor nodes connected to both ends of each pipe segment; Extract the energy attenuation rate marked on the propagation path between the sensor nodes at both ends of each pipe segment from the sensing response time sequence graph. Compare the energy attenuation rates of each pipe segment and select the pipe segment with the smallest energy attenuation rate as the candidate positioning pipe segment.

5. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 4, characterized in that, Based on the first reference node and the second reference node, a hyperbolic trajectory is established, including: The geographic coordinates of the first reference node and the geographic coordinates of the second reference node are respectively used as the first focus and the second focus of the hyperbola; The spatial distance between the first reference node and the second reference node in the pipeline topology is taken as the focal length of the hyperbola. The first distance difference is taken as the difference between the distance from any point on the hyperbola to the first focus and the distance to the second focus; Based on the difference between the focal length and the distance, the real axis length and imaginary axis length of the hyperbola are determined, and the hyperbola trajectory is constructed.

6. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 1, characterized in that, Fluid pressure transient signals are extracted from adjacent nodes of the candidate positioning pipe segment, and waveform distortion is compared with the pressure fluctuation reference curve to identify the characteristics of reflected waves caused by leakage, including: Fluid pressure transient signals are extracted from the upstream and downstream nodes of the candidate positioning pipe segment to obtain upstream pressure signals and downstream pressure signals. The upstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the upstream pressure signal relative to the pressure fluctuation reference curve. The pressure fluctuation reference curve is used to characterize the pressure fluctuation pattern over time under leak-free conditions. It is obtained by statistical fitting of the historical pressure signal of the target area pipeline network under normal operating conditions. The fluctuation boundary of the reference curve is defined by the amplitude of the environmental pressure fluctuation, and the pressure wave propagation period is defined by the time interval between adjacent peaks. When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the upstream pressure signal is marked as having reflected wave characteristics. The downstream pressure signal is overlaid with the pressure fluctuation reference curve to calculate the waveform distortion amplitude and distortion duration of the downstream pressure signal relative to the pressure fluctuation reference curve. When the waveform distortion amplitude exceeds the ambient pressure fluctuation amplitude and the distortion duration is less than the pressure wave propagation period, the downstream pressure signal is marked as having reflected wave characteristics.

7. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 1, characterized in that, Based on the reflected wave characteristics, the candidate positioning pipe segment is positionally corrected to obtain a refined positioning range, including: When there are reflected wave characteristics in the upstream pressure signal, the start time of the reflected wave is extracted from the upstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the upstream node is taken as the upstream reflection time difference. The upstream reflection time difference is multiplied by the propagation speed between nodes to obtain the upstream reflection distance between the upstream sensor node and the leakage point. When there are reflected wave characteristics in the downstream pressure signal, the start time of the reflected wave is extracted from the downstream pressure signal. The difference between the start time of the reflected wave and the normal arrival time of the pressure wave at the downstream node is taken as the downstream reflection time difference. The downstream reflection time difference is multiplied by the propagation speed between nodes to obtain the downstream reflection distance between the downstream sensor node and the leakage point. Starting from the upstream node of the candidate positioning pipe segment, the upstream reflection distance is extended along the pipe segment direction into the interior of the candidate positioning pipe segment to obtain the upstream positioning boundary; Taking the downstream node of the candidate positioning pipe segment as the endpoint, the downstream reflection distance is extended in the opposite direction to the interior of the candidate positioning pipe segment along the pipe segment to obtain the downstream positioning boundary; The pipe section between the upstream positioning boundary and the downstream positioning boundary is used as the refined positioning interval.

8. The method for locating leak points in sewage pipe networks based on multi-source sensor signal analysis according to claim 1, characterized in that, From the upstream vibration signal sequence and the downstream vibration signal sequence, the rate of change of the spectral centroid offset over time is extracted respectively to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate, including: The upstream vibration signal sequence and the downstream vibration signal sequence are divided into multiple upstream subsequences and multiple downstream subsequences according to the same time window, wherein the length of the time window is determined according to the pressure wave propagation period and is less than the pressure wave propagation period; Perform spectral analysis on each upstream subsequence, calculate the spectral centroid frequency of each upstream subsequence, and obtain the upstream centroid frequency sequence; Calculate the ratio of the difference in frequency between adjacent upstream centroids to the length of the time window to obtain the rate of change of the upstream spectral centroid. Perform spectral analysis on each downstream subsequence, calculate the spectral centroid frequency of each downstream subsequence, and obtain the downstream centroid frequency sequence; The ratio of the difference between adjacent downstream centroid frequencies to the length of the time window is calculated to obtain the rate of change of the downstream spectral centroid.

9. A sewage pipe network leakage point location system based on multi-source sensor signal analysis, characterized in that, The system is used to implement the sewage network leakage point location method based on multi-source sensor signal analysis according to any one of claims 1-8, the system comprising: The time-series map construction module is used to obtain the pipeline topology of the target area. By deploying acoustic sensor arrays at key nodes, it collects the first vibration signal sequence output in response to leakage events and combines it with the geographical coordinates of each sensor node to construct a sensor response time-series map. The candidate positioning pipe segment generation module is used to calculate the initial positioning interval of the leakage source in the pipe network topology based on the sensor response time series spectrum and the difference in the first arrival time of the first vibration signal received by each sensor node, and to converge the initial positioning interval according to the energy attenuation rate of the first vibration signal sequence on the propagation path to generate candidate positioning pipe segments. The position correction module is used to extract fluid pressure transient signals from adjacent nodes of the candidate positioning pipe segment, compare the waveform distortion with the pressure fluctuation reference curve, identify the reflected wave characteristics caused by leakage, and correct the position of the candidate positioning pipe segment according to the reflected wave characteristics to obtain a refined positioning range. The positioning fusion module is used to perform multi-source sensor fusion positioning within the refined positioning range based on the synchronicity between the spectral centroid offset of the first vibration signal sequence and the fluctuation period of the fluid pressure transient signal, and output the final positioning coordinates of the leak point, including: The first vibration signal sequence is extracted from the upstream sensor node and the downstream sensor node corresponding to the refined positioning interval, respectively, to obtain the upstream vibration signal sequence and the downstream vibration signal sequence; The rate of change of the spectral centroid offset over time is extracted from the upstream vibration signal sequence and the downstream vibration signal sequence, respectively, to obtain the upstream spectral centroid change rate and the downstream spectral centroid change rate; The fluctuation period is extracted from the upstream pressure signal to obtain the upstream pressure fluctuation period; the fluctuation period is extracted from the downstream pressure signal to obtain the downstream pressure fluctuation period. Calculate the ratio of the upstream spectral centroid change rate to the upstream pressure fluctuation period to obtain the upstream synchronicity parameter; and calculate the ratio of the downstream spectral centroid change rate to the downstream pressure fluctuation period to obtain the downstream synchronicity parameter. The ratio of the upstream synchronization parameter to the downstream synchronization parameter is used as the fusion weight; Obtain the geographic coordinates of the upstream sensor node and the geographic coordinates of the downstream sensor node; The fusion weight is used as the contribution coefficient of the upstream sensor node. The product of the geographical coordinates of the upstream sensor node and the contribution coefficient is calculated to obtain the upstream weighted coordinates. Calculate the difference between 1 and the fusion weight as the contribution coefficient of the downstream sensor node, and calculate the product of the geographical coordinates of the downstream sensor node and the downstream contribution coefficient to obtain the downstream weighted coordinates. The upstream weighted coordinates and the downstream weighted coordinates are added together to obtain a fused coordinate point, which is then used as the final location coordinates of the leak point.

Images