Real-time fault positioning method for drainage system

Abstract

The invention belongs to the technical field of computers, and particularly relates to a real-time fault positioning method for a drainage system, which comprises the following steps of: constructing a directed network model and a node-pipe section incidence matrix which reflect the complex topology of a pipe network according to a pipe network geographic information system and sensor data; calculating the flow unbalance degree of each node based on the real-time flow data and the incidence matrix, and dynamically checking and updating the network direction by using the real-time pressure or flow data; reversely tracing an upstream suspicious pipe section area in the network by taking the unbalance degree over-limit node as a starting point, and performing downstream influence propagation analysis; and finally, combining the suspicious region, the influence range and the whole network unbalance degree space distribution mode, and cooperatively judging the fault type and the accurate position. According to the method, the problems of false alarm and inaccurate positioning caused by model simplification are solved, and accurate and rapid traceability and diagnosis of complex drainage pipe network faults are realized.

Description

Technical Field

[0001] This invention belongs to the field of computer technology, and in particular relates to a method for real-time fault location in drainage systems. Background Technology

[0002] Urban drainage networks are critical infrastructure for ensuring urban water security and the water environment. Due to their large size, complex structure, and deep underground location, quickly and accurately locating the fault point in the event of blockages or leaks has always been a technical challenge in operation and maintenance.

[0003] Existing fault location technologies can be mainly divided into three categories: The first category is direct detection technologies that rely on physical contact or internal exploration, such as closed-circuit television (CCTV) detection. While the detection results are intuitive, they are post-incident inspections, inefficient, and unable to provide real-time early warnings. The second category is prediction technologies based on machine learning models. Their performance heavily depends on the scale and quality of historical data, and the models have poor interpretability, making it difficult to provide a clear path to fault location.

[0004] The third category is analysis methods based on hydraulic models and real-time monitoring data, which is also the most relevant prior art to this invention. For example, Chinese patent document CN118229463A discloses a method to assist in judging blockages by comparing the differences in flow data between upstream and downstream of a drainage pipe. This type of method embodies a typical technical approach, which relies on a pre-set, simplified system flow model (such as a tree-like branch or linear structure). However, the actual urban drainage network is a complex network system containing loops and cross-connections. Such simplified models cannot accurately describe the real flow distribution and transmission paths, leading to distorted analysis premises, high false alarm rates when applied in complex pipe networks, and an inability to systematically reveal the exact propagation chain and source of faults in the complete network. In addition, some prior art focuses on optimizing the construction of monitoring systems, such as the monitoring point layout method disclosed in Chinese patent document CN120217610A, which aims to improve the coverage and economy of monitoring. This is a planning stage technology and does not solve the core problem of subsequent operation and maintenance: how to perform real-time dynamic fault location based on existing monitoring systems.

[0005] Therefore, there is an urgent need to propose an integrated real-time location method that can overcome the above-mentioned defects, especially adapt to the complex network topology of drainage pipe networks, and achieve accurate fault source tracing and impact range analysis. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time fault location method for drainage systems, so as to solve the problem of the integrated real-time location method proposed in the background art that cannot achieve accurate fault source tracing and impact range analysis.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a real-time fault location method for a drainage system, comprising the following steps:

[0008] Step S1: Based on the geographic information system data of the drainage pipe network and the location information of the deployed flow sensors, construct a directed topology network model of the drainage system.

[0009] Step S2: Based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system, construct the node-pipe segment association matrix;

[0010] Step S3: Based on the real-time monitoring data of all flow sensors and the node-pipe segment correlation matrix, calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system.

[0011] Step S4: Based on the real-time pressure or flow rate symbol data of key pipe sections, perform dynamic verification and update on the preset pipe section water flow transmission direction in the directed topology network model of the drainage system.

[0012] Step S5: Based on the specific node whose real-time flow imbalance exceeds a preset threshold, in the directed topology network model of the drainage system, upstream reverse tracing is performed along all inflow edge directions of the specific node to locate the set of potentially suspicious pipe segment areas.

[0013] Step S6: Based on the set of potentially suspicious pipe segment areas, perform downstream impact analysis along the outflow edge direction of all end nodes in the set in the directed topology network model of the drainage system to determine the downstream impact range.

[0014] Step S7: Based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of all network nodes, perform collaborative analysis and comprehensive judgment to determine the specific type and precise geographical location of the fault.

[0015] Preferably, in step S1, the initial water flow conduction direction of each edge in the network is determined based on the pipeline design parameters and historical water flow data.

[0016] Preferably, in step S2, the node-pipe segment association matrix is ​​expressed in matrix form, where the rows of the matrix correspond to the nodes in the network, the columns of the matrix correspond to the pipe segments in the network, and values ​​are assigned to the corresponding positions in the matrix according to the inflow and outflow relationship between the pipe segments and the nodes.

[0017] Preferably, in step S3, when calculating the real-time flow imbalance, for each node, the absolute value of the difference between the total inflow and the total outflow calculated based on the node-pipeline correlation matrix and real-time monitoring data is used as the real-time flow imbalance of that node.

[0018] Preferably, the difference is defined as the total inflow minus the total outflow, with a positive difference indicating a flow surplus and a negative difference indicating a flow deficit.

[0019] Preferably, in step S3, the preset threshold is a dynamic threshold whose size is adaptively adjusted according to the historical flow data and real-time flow change rate of the nodes in the directed topology network model of the drainage system, so as to eliminate the systematic flow deviation under normal operating conditions.

[0020] Preferably, in step S4, when dynamically verifying and updating the network direction, the direction of water flow transmission is determined by comparing the real-time pressure difference signs of upstream and downstream nodes of the same pipe segment, or by analyzing the real-time flow signs and magnitude relationships between different inflow pipe segments of the same node.

[0021] Preferably, in steps S5 and S6, when performing upstream reverse tracing or downstream impact analysis, a graph search algorithm is used for traversal, wherein the graph search algorithm is a breadth-first search algorithm or a depth-first search algorithm.

[0022] Preferably, when the potentially suspicious pipe segment area set is concentrated upstream of a certain pipe segment, and the nodes within its downstream influence range show a continuous negative flow imbalance, it is determined that the pipe segment has a blockage fault.

[0023] When the distribution of the potentially suspicious pipe section area is discrete and its downstream influence range is not significant, and the imbalance of the abnormal node is a continuous low-amplitude positive value, it is determined to be an instrument measurement drift fault.

[0024] A real-time fault location system for a drainage system, comprising:

[0025] The network construction module is used to construct a directed topology network model of the drainage system based on the geographic information system data of the drainage pipe network and the location information of the deployed flow sensors.

[0026] The matrix construction module is used to construct a node-pipe segment association matrix based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system.

[0027] The imbalance calculation module is used to calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system based on the real-time monitoring data of all flow sensors and the node-pipe segment correlation matrix.

[0028] The direction verification and update module is used to perform dynamic verification and update of the water flow conduction direction of the pipe segments in the directed topology network model of the drainage system based on the real-time pressure or flow symbol data of the key pipe segments.

[0029] The upstream tracing module is used to perform upstream reverse tracing along all inflow edges of the specific node in the directed topology network model of the drainage system, based on the specific node whose real-time flow imbalance exceeds a preset threshold, to locate a set of potentially suspicious pipe segment areas.

[0030] The downstream analysis module is used to perform downstream impact analysis along the outflow edges of all end nodes in the directed topology network model of the drainage system, based on the set of potentially suspicious pipe segment areas, to determine the downstream impact range.

[0031] The fault determination module is used to perform collaborative analysis and comprehensive judgment based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of all network nodes, to determine the specific type and precise geographical location of the fault.

[0032] Compared with existing technologies, the advantages of this real-time fault location method for drainage systems are:

[0033] 1. Fundamentally improved model accuracy: By constructing a directed topological network model that accurately reflects the complex connection relationships of pipeline loops, junctions, etc., the problem of "model distortion" caused by existing technologies relying on simplified tree or linear models is fundamentally overcome, providing a correct physical and mathematical basis for all subsequent analyses.

[0034] 2. Significantly enhanced system robustness: A dynamic direction verification and update mechanism based on real-time pressure or flow data is introduced, enabling the system to automatically adapt to changes in pipeline flow direction caused by blockages, scheduling, etc. This solves the problem of static models completely failing under such conditions, significantly improving the practicality and reliability of the method.

[0035] 3. Breakthroughs in fault location capabilities in two aspects:

[0036] Precise source tracing: Based on graph theory algorithms, it realizes systematic reverse tracing from abnormal nodes upstream, which can clearly depict the potential propagation path and source area of ​​the fault, and expand the alarm "point" into a suspicious "area".

[0037] Impact prediction: It can analyze the scope of the impact of the fault on the downstream pipeline network, realizing the leap from "fault diagnosis" to "risk warning", and supporting the formulation of forward-looking emergency dispatch plans.

[0038] 4. Intelligent closed-loop diagnostic process: By comprehensively analyzing multi-source information such as "suspicious areas", "affected areas" and "network-wide traffic imbalance mode", it can automatically match fault characteristics (such as congestion and instrument drift) and output specific fault types and location judgments with confidence. It transforms complex data analysis into decision-making suggestions that can directly guide operation and maintenance actions, greatly reducing the technical threshold.

[0039] 5. A complete and indivisible technical solution: The aforementioned benefits do not exist in isolation, but rather stem from the strong logical closed loop and technological synergy formed by the seven steps. From accurate modeling and intelligent perception to bidirectional analysis and comprehensive judgment, each step strictly relies on the preceding outputs and works together to systematically solve all the core defects pointed out in the background technology, constituting a complete invention with outstanding substantive features and significant progress. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating a real-time fault location method for a drainage system according to an embodiment of the present invention.

[0041] Figure 2 This is a schematic diagram of the structure of a real-time fault location system for a drainage system according to an embodiment of the present invention;

[0042] Figure 3 This is a flowchart of a real-time fault location method for a drainage system according to an embodiment of the present invention. Detailed Implementation

[0043] The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0044] This invention provides a real-time fault location method for a drainage system, referring to... Figure 1 ,include:

[0045] Step S1: Based on the geographic information system data of the drainage pipe network and the location information of the deployed flow sensors, construct a directed topology network model of the drainage system.

[0046] Specifically, in step S1, the core of the implementation is to transform the physical drainage network entity into a standardized network structure that can be used for mathematical and logical analysis. This process begins with in-depth processing of the drainage network geographic information system (GIS) data.

[0047] In GIS data, each facility such as a manhole, pumping station, and overflow outlet is abstracted as a node, which is usually denoted as in mathematical terms. (Representing a vertex), all nodes constitute a node set. Each pipe connecting two facilities is abstracted as a directed edge, which is usually denoted as in mathematical notation. (Represents an edge), and all edges constitute the edge set. Therefore, the entire drainage system is formally defined as a directed graph. .

[0048] Assigning edge directions is crucial for constructing a directed topological network model. This is based on the potential flow paths of water, primarily determined through a combination of pipeline design parameters (such as slope and elevation) and historical hydrological monitoring data. Elevation is a key geographical parameter, referred to here in uppercase letters. Representation, for example, a node The elevation is recorded as For pipe sections with complete design data, the direction is determined by the elevation, pointing from the higher elevation node to the lower elevation node; for pipe sections lacking design data or with complex flow directions (such as flat areas), it is necessary to analyze the flow meter data deployed over a long period of time, and take the statistically dominant flow direction as the standard.

[0049] This model not only describes the connection relationships of the pipelines, but more importantly, it depicts the expected direction of water flow under static design, laying a structural foundation for subsequent graph theory-based flow analysis and fault tracing. This step transforms chaotic spatial entity data into a standardized network model that can be recognized and reasoned about by computers, which is the first and crucial step in realizing intelligent diagnosis.

[0050] In one specific embodiment, the implementation process is as follows:

[0051] The system imports a pipeline GIS spatial database containing node coordinates, attributes, and edge connection relationships.

[0052] First, perform a topology check to correct errors such as dangling wires and redundant nodes, ensuring that the network logical connectivity is correct.

[0053] Subsequently, the system reads or constructs a node elevation attribute table for any edge. Connecting nodes and If it is known elevation Greater than elevation Then the initial edge direction is from point to For elevation differences less than a set threshold For pipe sections with missing elevation data, the system queries the historical database: if a flow meter is installed on or immediately upstream or downstream of the pipe section, its historical readings are collected, and the percentage of forward flow time is calculated and recorded as follows. ;like Then follow the direction indicated by the flow meter; if If so, it is marked as "direction undetermined"; if If the direction is not specified, it is determined to be a reverse flow, and the direction is set accordingly. All edges with "direction pending" will trigger an alarm, and the direction will be determined manually based on the regional catchment boundaries and the overall pipeline network plan.

[0054] Finally, the system constructs the adjacency list data structure of the directed graph in memory, denoted as For each node , It stores a list of all its direct downstream neighbors. Simultaneously, the system generates a network visualization graph with arrows, allowing operators to ultimately verify the model's accuracy.

[0055] The following points should be paid special attention to during the implementation process:

[0056] First, the topological integrity of the original GIS data must be guaranteed, as this is a fundamental prerequisite for a correct model. The data cleaning process must eliminate all logical inconsistencies, such as ensuring that pipeline connections are consistent with the actual site conditions and avoiding the creation of "isolated" nodes.

[0057] Secondly, a clear priority rule needs to be established for assigning the initial direction: "Measured data takes precedence over design data, and design data takes precedence over manual inference." Special modeling is required for hydraulic control facilities such as pump stations, gates, and spillways. A pump station should be modeled as a special node with a forced direction, whose downstream direction is determined; a gate should be modeled as an edge with a variable state (open / closed), whose direction is valid when open and temporarily removed from the network when closed.

[0058] Third, the model should have version management capabilities. Any manual corrections or subsequent dynamic updates should have complete operation logs to ensure the model's traceability.

[0059] Fourth, the advantage of this step is that it creates a digital twin skeleton that is faithful to the physical system and easy to compute, transforming complex engineering problems into clear graph theory problems. This allows for the systematic application of mature graph algorithms for fault analysis, which is the core foundation that distinguishes the method of this invention from traditional empirical methods.

[0060] Step S2: Based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system, construct the node-pipe segment association matrix.

[0061] Specifically, in step S2, the objective is to express the flow conservation constraints inherent in the directed topology network constructed in step S1 using precise linear algebra. Its core output is the node-edge incidence matrix. Let the network be a directed graph. Include Each node and Segment, i.e. Define the correlation matrix. For one The matrix.

[0062] Each element in the matrix The definition strictly follows the topological relationship between nodes and pipe segments: if the pipe segment Termination at node (Right now Inflow ),but If the pipe section Starting from node (Right now outflow ),but If the pipe section With nodes If there is no direct connection, then Therefore, the matrix Each row corresponds to the flow balance equation of a node, and each column corresponds to the contribution of a pipe segment to the node connected to it.

[0063] According to Kirchhoff's current law, in a steady flow state, for any ordinary node that is neither a source nor a sink... The algebraic sum of the flow rates through all pipe sections flowing through it should be zero. Let The column vector consisting of the flow rates of all pipe segments, where For pipe section The flow rate (defined as positive when it is in the same direction as the preset direction of the pipe section). Then the node... The flow balance equation can be written as: The flow balance constraints of the entire pipeline network can be simply represented by a matrix equation: , where 0 is Zero-dimensional vector.

[0064] This step uses mathematical abstraction to transform the complex physical connections and water flow distribution patterns of the pipe network into a unified, concise, and computable linear algebraic model, providing a solid mathematical framework for subsequent quantitative diagnosis.

[0065] In one specific embodiment, the implementation process is as follows:

[0066] The system has obtained the directed graph from step S1. and its adjacency list .

[0067] First, for all nodes and pipe section Establish a globally unique and fixed index mapping. Construct an all-zero index. Sparse matrix .

[0068] Then, each pipe segment was traversed. Let its starting node be The termination node is By definition, in a matrix In the middle, at the first Line number Enter -1 in the column position, at the 1st position. Line number Enter +1 in the column position. After traversal, the matrix... The construction is now complete.

[0069] Given that in actual pipeline networks, each node is only connected to a few pipe segments (the degree of a node is typically 2 to 4), the matrix It is highly sparse. To optimize storage and computational efficiency, the system immediately converts this matrix into a sparse matrix storage format, such as Compressed Sparse Row (CSR) format. This format only stores the values ​​of non-zero elements and their row and column indices, which greatly reduces memory usage. After construction, the matrix... With network model The data is bound and stored, serving as the core data object for all subsequent analysis and computation. Simultaneously, the system can output a sparse pattern diagram of the matrix to aid in understanding the network's topological characteristics.

[0070] The following points should be paid special attention to during the implementation process:

[0071] First, once the index mapping between nodes and pipe segments is established, it must remain stable throughout the entire system operation cycle. Any changes must be updated synchronously to update all data structures that depend on this index; otherwise, it will lead to serious calculation errors.

[0072] Secondly, for a real drainage network, there are flow "source" nodes (such as rainwater inlets and sewage connection points) and "sink" nodes (such as sewage treatment plant inlets and outlets). These nodes do not strictly meet the following conditions. ,in It is a matrix The Okay. For this, additional processing is needed. A common approach is to introduce a source-sink term vector on the right-hand side of the matrix equation. The equilibrium equation is extended to For the source node, This indicates injected traffic; for sink nodes, This indicates the outflow of traffic; for ordinary nodes, .vector A portion of the ____ can be obtained through flow meter measurement or estimation.

[0073] Third, matrix The construction depends entirely on directed graphs The direction setting is crucial. Therefore, when the direction of a pipe segment is dynamically updated in subsequent steps, the matrix must be updated accordingly. The corresponding column elements (with the original +1 and -1 swapped).

[0074] Fourth, the advantage of this step is that it transforms qualitative topological connections into quantitative mathematical objects that can be linearly calculated. This allows us to use mature linear algebra theory and numerical calculation tools to conduct systematic flow analysis and status assessment of large and complex pipeline networks containing tens of thousands of nodes and pipe segments. This is the mathematical foundation for achieving automated and refined fault diagnosis.

[0075] Step S3: Based on the real-time monitoring data of all flow sensors and the node-pipe segment correlation matrix, calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system.

[0076] Specifically, in step S3, the core of the implementation is based on real-time monitored traffic data and correlation matrix. The residuals of the flow balance equation at each node are quantitatively calculated to obtain an index characterizing the degree of anomaly—the real-time flow imbalance at each node. Let the flow vector of the pipe segment be... ,Right now It is an m-dimensional column vector of real numbers. Since not all pipe sections are equipped with flow meters, Divided into known parts (correspond (monitored sections) and unknown sections (correspond (Unmonitored pipe sections). Accordingly, the correlation matrix will be... Divided by column .

[0077] For nodes Its complete flow balance equation is: (Consider source and sink items). Due to Because the information is unknown, the total residual cannot be directly calculated. Therefore, the nodes... The real-time traffic imbalance is denoted as , defined as the absolute value of net flow calculated only based on the known flow portion, and normalized to enhance comparability. Its calculation formula is:

[0078]

[0079] in This refers to the absolute value of the node's net flow calculated based on known data. Physically, this value is explicitly defined as the total inflow minus the total outflow, and is denoted as... ,Right now .like A positive value indicates that the known inflow is greater than the known outflow (flow surplus); a negative value indicates that the known outflow is greater than the known inflow (flow deficit).

[0080] denominator This is the reference flow, used for normalization. Typically, it can be taken as the sum of the absolute values ​​of all known inflows to that node, i.e.:

[0081]

[0082] The summation iterates through all satisfying conditions. Pipe segment index That is, all known inflow nodes The pipe section, of which It is a very small positive number to prevent division by zero errors.

[0083] To intelligently distinguish between normal fluctuations and true anomalies, a dynamic threshold needs to be set for each node, denoted as . Its size is adaptively adjusted based on historical data. For example, it can be based on nodes. During the same historical period (considering daily and weekly cycles) Value sequence, calculate its history Statistical mean of values and statistical standard deviation And set the threshold to ,in This is the sensitivity coefficient (usually taken as 2 or 3).

[0084] This step is like installing a "flow balance" at each node of the pipeline network. By weighing the weight difference between the "known inflow" and the "known outflow" in real time and comparing it with the historical normal fluctuation range, any abnormal imbalance signal can be sensitively detected.

[0085] In one specific embodiment, the implementation process is as follows:

[0086] The system periodically (e.g., every 5 minutes) retrieves readings from all online flow meters from the Supervisory Control and Data Acquisition (SCADA) system and populates the vector according to the pipe segment index mapping. The corresponding position. Simultaneously, the known portion of the association matrix is ​​loaded from the model. Source and sink vectors (part (Possibly from the main inlet or the flow meter at the wastewater treatment plant inlet). For each node The program performs the following calculations:

[0087] (1) Extraction (a row vector) and Perform dot product operation, plus The original imbalance value is obtained. ;

[0088] (2) Calculate the reference flow rate The program identified For all elements with a value of +1 (corresponding to known inflow pipe segments), assign these positions to... Summation and (e.g. 10) -6 m 3 ( / s) Compare and take the larger value;

[0089] (3) Calculate the normalized real-time flow imbalance. ;

[0090] (4) Query the time period pattern (e.g., "weekday daytime") to which the node belongs at the current time, and retrieve the corresponding time period pattern from the historical database. and Calculate the dynamic threshold. ;

[0091] (5) Comparison and ,like Then the node Mark it as an "abnormal node" and record its imbalance value. The positive and negative signs;

[0092] After all nodes have been calculated, the system obtains a feature snapshot containing a list of abnormal nodes and their detailed status.

[0093] The following points should be paid special attention to during the implementation process:

[0094] First, data time synchronization is crucial. Data from all flow meters and source / sink devices must be aligned to the same timestamp. The system needs data caching and interpolation capabilities to handle issues such as different sensor acquisition periods or communication delays.

[0095] Secondly, refer to traffic The computational strategy needs to be carefully designed. For nodes with a known total inflow that is extremely small, even tiny fluctuations in absolute value can lead to problems. Drastic changes can produce false alarms. Besides... In comparison, the design flow rate of this node or the flow rate estimated by the total upstream catchment area can also be used as a reference to improve stability.

[0096] Third, dynamic threshold Modeling needs to fully consider the periodicity, seasonality, and rainfall impact of the drainage system. Historical statistical models should be able to distinguish between dry and rainy weather patterns, and even further differentiate between different scenarios such as light rain, moderate rain, and heavy rain. The model needs to be retrained periodically (e.g., monthly) with the latest data to adapt to the slow changes in the pipe network itself.

[0097] Fourth, the advantage of this step is that it compresses continuous, multi-dimensional traffic time-series data into a scalar indicator with clear physical meaning and comparability at each node. It utilizes adaptive thresholds for intelligent initial screening, achieving efficient conversion from "data" to "effective alarm information," providing precise guidance for subsequent steps to focus on the real potential problem areas.

[0098] Step S4: Based on the real-time pressure or flow rate symbol data of the key pipe sections, perform dynamic verification and update on the preset pipe section water flow transmission direction in the directed topology network model of the drainage system.

[0099] Specifically, in step S4, the core implementation is to add dynamic adaptability to the static initial network model, and to verify and correct potential errors in the preset water flow direction of the pipe section through real-time sensor data. Two specific verification methods are provided.

[0100] Method 1: Pressure Difference Sign Method

[0101] For a preset direction as slave node Flow to Node pipe section At its upstream end and downstream end Install a pressure sensor to monitor the pressure value in real time. and In a gravity-driven drainage network, without external power (such as a pump), there should be a head loss along the flow direction, hence the pressure difference. During normal flow, it should be a positive value (or slightly greater than zero). If the system continuously monitors... (in If it is a dead zone threshold that takes into account measurement noise and local disturbances, then it strongly indicates that the actual water flow direction is opposite to the preset direction.

[0102] Method 2: Node Flow Sign Consistency Method

[0103] For a node Consider the set of all inflow pipe sections equipped with flow meters. and outflow pipe section collection According to the flow meter installation agreement, the measured value... The positive or negative sign indicates the actual flow direction (e.g., positive for inflow and negative for outflow). This applies to pipe sections pre-defined as inflow. If its flow reading A consistently negative value indicates actual outflow, contradicting the preset value; conversely, a positive value indicates outflow for a pipe section preset to be outflowing. ,like A consistently positive value also indicates a contradictory direction.

[0104] When any method (or a combination of both) continuously detects directional inconsistencies for more than a set time window If the time limit is 10 minutes, the system determines that the pipe segment's direction setting has failed. At this point, the directed graph model in memory... Inside, the pipe section The direction is flipped (i.e., its starting and ending nodes are swapped), and the association matrix is ​​updated synchronously. The element symbols in the corresponding column (swap +1 and -1).

[0105] This step endows the entire diagnostic system with intelligent characteristics of "adaptive" and "self-correction," ensuring that the core topology model can closely follow the dynamically changing real-world operating conditions, thereby guaranteeing the reliability of all subsequent analysis conclusions.

[0106] In one specific embodiment, the implementation process is as follows:

[0107] The system maintains a list of pipe segments that require dynamic verification, typically including: pipe segments in areas with gentle terrain, pipe segments downstream of combined sewer systems prone to backflow, and pipe segments before and after pumping stations. The system starts a separate monitoring thread to periodically (e.g., every 1 minute) verify these pipe segments.

[0108] For pipe sections using the pressure difference method, the thread reads the pressure values ​​at both ends. and Calculate the instantaneous pressure difference Then, for a sliding time window (e.g., the past 10 sampling points) Perform statistical analysis on the sequence. If more than 90% of the data points within this window meet the following criteria: (For example If a water column reaches 0.5m, the "Suspected Reverse Direction" flag is triggered and a timer begins. If this flag remains in place for more than [time period missing], [the text abruptly ends here]. If so, then a direction update operation will be performed.

[0109] For nodes using the flow sign method, the thread checks all connected pipe segments with flow meters for that node. It calculates the proportion and duration of pipe segments whose preset direction does not match the flow meter sign within a sliding time window. If the proportion and duration of inconsistency for a certain pipe segment exceed a threshold, a direction update process is also triggered.

[0110] When an update occurs, the system first records the event (time, pipe segment, old direction, new direction, triggering basis) in the log, and then calls the model update function to modify the graph. sum matrix After the pipe segment direction is reversed, the element symbols in the corresponding column of the node-pipe segment association matrix need to be updated synchronously (by swapping +1 and -1) to ensure that the matrix is ​​consistent with the current network topology. After the update is completed, the system can set a flag and recommend re-executing steps S3 to S6 in the next analysis cycle to re-evaluate the system status based on the new model.

[0111] The following points should be paid special attention to during the implementation process:

[0112] First, direction verification is a high-risk, high-impact operation that must adhere to the principle of "better to miss a case than to make a wrong one." Triggering conditions (such as proportional thresholds, dead zones) are crucial. Duration The settings should be relatively conservative to avoid accidental reversal caused by short-term sensor noise, water hammer, or temporary hydraulic fluctuations (such as pump start-up and shutdown).

[0113] Secondly, the pressure difference method requires consideration of the fluid static pressure head in practical applications. Strictly speaking, the difference between the pressure head and the position head should be used. Make a judgment (where) For fluid density, The sign convention is determined by gravity (acceleration) rather than simple pressure difference, to eliminate the influence of differences in pipe elevation. The flow sign convention is highly dependent on the flow meter's installation orientation and zero-point calibration; it is essential to ensure that the sign conventions of all flow meter data are completely consistent with the system model.

[0114] Third, the impact of the direction update needs to be carefully assessed. A direction reversal may make previously "abnormal" nodes "normal," or it may cause new nodes to become "abnormal." The system should be able to provide a brief report on the impact of the change after the update.

[0115] Fourth, the advantage of this step is that it effectively solves the "model drift" problem that has long plagued traditional static model methods, making the diagnostic system no longer a fragile "greenhouse flower" but a "wilderness survivor" capable of coping with complex field conditions (such as backflow, scheduling, and accidents), greatly enhancing the practicality, robustness, and long-term reliability of the entire method in actual engineering.

[0116] Step S5: Based on the specific node whose real-time flow imbalance exceeds a preset threshold, in the directed topology network model of the drainage system, upstream reverse tracing is performed along all inflow edge directions of the specific node to locate the set of potentially suspicious pipe segment areas.

[0117] Specifically, in step S5, the objective is to use one or more "abnormal nodes" identified in step S3 as the starting point for tracing back along the flow direction defined by the directed network, and trace all upstream paths that may have carried abnormal traffic to them, thereby defining a set of potentially suspicious pipe segment areas, denoted as... This set contains the spatial range in which the root cause of the failure may exist.

[0118] The physical principle is: a node experiences a traffic imbalance (e.g., a persistent traffic deficit). The root cause must lie in one or more upstream pipe sections (such as blockage or leakage). During implementation, each abnormal node... The node is added to the queue as the initial point for the search. Then, a graph search algorithm (such as breadth-first search (BFS) or depth-first search (DFS) is used for traversal.

[0119] Taking BFS as an example, a node is retrieved from the queue. In the reverse adjacency list (i.e., the data structure that stores all the direct upstream neighbors of each node), look up... All direct upstream nodes For each upstream node If a region has not been visited, it is marked as visited and added to the suspicious region set. In the middle. Then, an assessment will be made as to whether it should be from Continue tracing upwards. Termination conditions for tracing typically include:

[0120] (1) It is itself the source point of the pipeline network (such as the runoff inlet) and has no upstream nodes;

[0121] (2) Real-time traffic imbalance of nodes Below a secondary threshold, denoted as (e.g. 0.5) This indicates that the abnormal signal has significantly attenuated at this node;

[0122] (3) Reach the preset maximum number of trace steps (depth).

[0123] If the termination conditions are not met, then Add a node to the queue to continue exploring its upstream. This process is repeated until the queue is empty. Ultimately, all visited nodes and the internal segments connecting them form a connected subgraph, i.e., For multiple abnormal nodes, they can be traced separately and then merged into a single set.

[0124] This step is similar to retrieving surveillance footage from the accident scene along all possible directions of oncoming traffic in a complex road network, gradually narrowing down the starting area of ​​the suspect vehicle.

[0125] In one specific embodiment, the implementation process is as follows:

[0126] The system obtains the list of abnormal nodes output in step S3. and its imbalance value Initialize an empty collection. and an empty queue Simultaneously, maintain an "visited" flag for each node. For each abnormal node in the list... Perform the following operations:

[0127] (1) If If an item is not visited, mark it as visited and add it to the collection. and queue .

[0128] (2) When the queue When not empty:

[0129] a. Depart one node from the queue. .

[0130] b. Query the reverse adjacency table to obtain... List of all direct upstream nodes .

[0131] c. For each If it has not been visited:

[0132] i. Mark as visited and add to the set ;

[0133] ii. Get the node's and ;

[0134] iii. Determine whether to continue tracing: If Not a known source and and The symbol and the initial abnormal node If the signs are the same (e.g., both are negative), then... Joining the team Otherwise, stop from that Let's continue tracing back.

[0135] (3) Clear the queue Prepare to handle the next abnormal node.

[0136] After all abnormal nodes have been processed, the set This is the suspected area of ​​this fault event. The system then calculates the set. The topological features, such as diameter, number of nodes, and number of edges, are analyzed and the results are visualized.

[0137] The following points should be paid special attention to during the implementation process:

[0138] First, constructing the reverse adjacency list is a prerequisite for efficient tracing. The forward adjacency list can be constructed in step S1. At the same time, a reverse adjacency list is built to trade space for time and ensure tracing speed.

[0139] Secondly, secondary threshold Consistency judgment with the sign is crucial for controlling the accuracy of the traceability range. An overly lenient threshold can lead to… If the value is too large, it loses its significance for localization; if it is too strict, it may miss the real remote root cause. Sign consistency checks (such as negative signs) help filter out irrelevant fluctuations caused by measurement noise, making the tracing more focused on the propagation chain of the same hydraulic problem.

[0140] Third, in mesh topologies (where loops exist), the upstream of a node may trace back to the same ancestor node through multiple paths. The BFS algorithm can naturally handle this situation, ensuring that the region... The integrity of the pipeline is precisely where the advantage of this method in handling complex pipeline networks lies. However, care must be taken to prevent infinite loops within loops, which can be effectively avoided by using "visited" markers.

[0141] The advantage of this step is that it achieves a qualitative leap from "point-based alarms" to "area-based positioning." Maintenance personnel no longer need to blindly investigate scattered alarm points, but instead obtain a "heat map" that clearly marks the most likely hiding area of ​​the fault's root cause, greatly narrowing the scope of manual on-site verification and improving the response speed and accuracy of emergency repairs.

[0142] Step S6: Based on the set of potentially suspicious pipe segments, perform downstream impact analysis along the outflow edges of all end nodes in the set in the directed topology network model of the drainage system to determine the downstream impact range.

[0143] Specifically, in step S6, the objective is to evaluate the suspicious areas located in step S5. If a fault does exist, what extent will it affect the downstream pipeline network? This will determine the downstream impact range, denoted as [missing information]. This is a forward propagation analysis corresponding to the reverse tracing in step S5.

[0144] The logic is that faults (such as blockages) not only affect upstream flow but also obstruct or alter the normal drainage path downstream, potentially leading to secondary risks (such as insufficient flow in downstream pipe sections or overflowing manholes). During implementation, the first step is to identify potential areas. The outflow boundary nodes. These nodes belong to the set. However, there exists at least one outgoing edge pointing to the set. Nodes outside the boundary. Formally, the outflow boundary node set. .

[0145] Then, with Starting from each node in the graph, a graph search algorithm (usually BFS) is used for forward traversal. From the boundary node... Begin by traversing all outgoing edges until you reach a direct downstream node. Add these downstream nodes to the set of nodes affected. The process continues, starting from these newly added nodes and traversing their downstream regions, progressing layer by layer. The traversal can terminate when: the network convergence point (such as a wastewater treatment plant) is reached, the preset maximum propagation depth is reached, or an important hydraulic control facility (such as a pumping station, the operation of which may change the downstream state) is encountered, requiring the analysis to be paused.

[0146] Ultimately, all those visited The nodes outside and the pipe sections connecting them constitute the downstream impact area. This step is similar to delineating the "epicenter" ( After that, the potential "affected areas" were simulated using a disaster propagation model. This provides a basis for disaster prevention plans.

[0147] In one specific embodiment, the implementation process is as follows:

[0148] The system obtains a set of suspicious areas. (List of node IDs).

[0149] First, traverse Each node in Check all outgoing edges using a forward adjacency list. (storage (All direct downstream nodes), determine if any of them do not belong to If so, then set that node. Add to outflow boundary node set .

[0150] Then, initialize an empty set of influence ranges. and a queue .for Each boundary node in (If not accessed), perform the following operations:

[0151] (1) Mark it as visited (for traversal within this step), and its downstream neighbors (belonging to) Add (outside) to queue and set .

[0152] (2) When the queue When not empty:

[0153] a. Depart one node from the queue. .

[0154] b. Obtain List of all direct downstream nodes .

[0155] c. For each If it is not accessed and :

[0156] i. Mark as visited and add to the collection. .

[0157] ii. Determine whether to continue spreading: If If it is not the sink and has not reached the maximum depth, then... Joining the team .

[0158] After the traversal is complete, the set This refers to the downstream impact range. The system can calculate... The geographical scope, the key facilities involved (such as pumping stations and flood-prone areas), and the imbalance of these nodes. (This should usually be a negative value, indicating insufficient flow), generate an impact assessment report.

[0159] The following points should be paid special attention to during the implementation process:

[0160] First, accurately identify the outflow boundary node set. This is crucial for accurate analysis. The algorithm must strictly adhere to the membership relationships of the set to ensure that no starting point of any outflow path is missed.

[0161] Secondly, the propagation termination conditions need to be flexibly configured. For large pipe networks, unrestricted downstream propagation may lead to an excessively large and unnecessary analysis scope. Setting a maximum propagation depth (e.g., 10 hops) or stopping upon encountering a pump station is a reasonable strategy, as the hydraulic conditions of the downstream pool at the pump station are relatively independent.

[0162] Third, the downstream impact range Possibly related to suspicious areas Geographically discontinuous, separated by normal areas, this is determined by the pipeline topology and needs to be clearly expressed in visualization.

[0163] Fourth, this step of the analysis is predictive, based on the current topology and fault assumptions. If the actual fault point is not in... If the core issue is different, or the type of failure is different (such as leakage), the actual impact may differ from the prediction in this step. The analysis results should be used as a risk warning rather than an absolute conclusion.

[0164] The advantage of this step is that it upgrades fault diagnosis from simply "locating the problem" to "assessing the risk," achieving a fusion of diagnosis and early warning. The operations and maintenance department can not only know "where might be broken," but also anticipate "what secondary problems might be caused by the breakage," thus enabling the development of more comprehensive emergency response strategies, such as pre-emptively deploying drainage equipment. The notification of key nodes in the region, or of potentially affected areas, significantly enhances the initiative and foresight of public safety protection.

[0165] Step S7: Based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of all network nodes, perform collaborative analysis and comprehensive judgment to determine the specific type and precise geographical location of the fault.

[0166] Specifically, in step S7, the objective is to act as the center for information fusion and intelligent decision-making, integrating the suspicious areas output from step S5. The downstream impact range of step S6 output And the real-time traffic imbalance of all network nodes output in step S3. and its symbols The spatial distribution pattern, in which Indicates taking The sign (positive, negative, or zero) is used. By analyzing the synergistic and contradictory relationships between multi-dimensional features, the specific type of fault is ultimately determined and its most likely precise location is inferred. Judgment logic for two typical fault modes is defined.

[0167] Mode 1: Congestion fault, characterized by the following collaborative features:

[0168] (1) Regional concentration: Suspicious areas Topologically highly concentrated in a specific pipe segment Upstream. This means that through tracing back in step S5, Most nodes in the pipeline converge to the pipe section via one or more paths. The upstream node.

[0169] (2) Downstream persistent shortage: pipeline section downstream impact range The unbalanced value of the nodes within. It exhibits continuous, significant negative values ​​(i.e.) ,in (A set deficit threshold) indicates a systematic deficiency in downstream flow.

[0170] (3) Comparison of upstream and downstream symbols: Nodes within the region It may be a slightly positive or negative value, but it is consistent with... The strong negative values ​​in the region create a distinct gradient. When these characteristics are met, a blockage fault is identified, and the pipe segment is... The location of the fault is marked as precise.

[0171] Mode 2: Instrument measurement drift fault, characterized by the following features:

[0172] (1) Regional discreteness: Suspicious areas The nodes in the system are topologically and geographically dispersed, lacking clear convergence paths.

[0173] (2) Insignificance of impact: downstream impact range It is very small or empty because the instrument malfunctions and does not accurately reflect the water flow.

[0174] (3) The uniqueness and low amplitude of the abnormal pattern: Abnormal nodes It manifests as a sustained low amplitude of positive or negative values ​​(i.e.) ,in (For a set drift amplitude threshold), and the signs are usually consistent (both positive or both negative) in the same fault event.

[0175] If these characteristics are met, it is determined to be an instrument drift fault, and... The flow meter column corresponding to the middle node is a suspicious instrument.

[0176] This step is similar to that of an experienced doctor who synthesizes the patient's medical history. ), physical signs ( ) and lab report ( , (Distribution), through logical reasoning and pattern matching, to give clear diagnostic conclusions.

[0177] In one specific embodiment, the implementation process is as follows:

[0178] The system operates based on a preset rule base. After a complete S1-S6 analysis process is completed, the engine extracts the feature vector of this event, denoted as... This includes: The number of nodes topological diameter, Middle node Statistic (mean) ,variance (positive and negative ratios) The number of nodes Middle node mean Is there a critical pipeline section? Its upstream covers most of Nodes, etc.

[0179] The rule for determining a blockage fault is: when there is a pipe segment This results in the number of suspicious nodes located upstream accounting for a significant portion of the suspicious area. The proportion of the total number of nodes exceeds 80%, and the downstream impact range is... Not empty, and Mean of internal node imbalance If the value is below -0.3 (indicating a persistent deficit) and the variance of the imbalance values ​​at these nodes is small (consistent in performance), then the fault type is determined to be "blockage," and the fault location is the pipe segment. And assign a high confidence level.

[0180] The rule for determining instrument measurement drift faults is: when the suspected area... The topological diameter is relatively large (nodes are discretely distributed), and the downstream influence range is also large. The number of nodes accounts for The proportion of nodes is less than 20%, and Mean of absolute values ​​of imbalance at mid-nodes If the value is between 0.1 and 0.4 (low amplitude anomaly) and its variance is small, the fault type is determined to be "instrument drift," and the list of suspected devices is as follows: The flow meter associated with the middle node is assigned a medium confidence level.

[0181] After the determination is completed, the system generates a structured diagnostic report, which includes the fault type, location / device, confidence level, triggering rules, and key evidence (such as...). ID, The data (value) and timestamp are then pushed to the monitoring center interface and mobile terminals.

[0182] The following points should be paid special attention to during the implementation process:

[0183] First, the rules need continuous evolution and optimization. The initial rule base is based on physical principles and expert experience, but it may not cover all complex scenarios (such as congestion accompanied by minor leaks). The system should be designed with a feedback learning mechanism, allowing operations and maintenance personnel to confirm or correct the automatic judgment results, and use this feedback data to iteratively optimize rule parameters or introduce new features and rules.

[0184] Secondly, the relationship between confidence level and manual review must be clearly defined. For judgments with high confidence (such as typical blockages), the system can directly trigger a maintenance work order; for judgments with medium or low confidence (such as suspected instrument drift or complex situations), the system should prominently mark "manual review recommended" in the report and provide all relevant data links, which must be ultimately confirmed by maintenance personnel.

[0185] Third, in addition to the two modes mentioned above, efforts should be made to expand the failure mode library. For example, the characteristics of "pipeline rupture and leakage" might be... Regional concentration, downstream impact Not significant, but nodes The readings show a significant positive value (excessive flow), and may be accompanied by an abnormal drop in pressure sensor readings in the area. It is necessary to continuously accumulate case studies in actual operation to enrich diagnostic capabilities.

[0186] The advantage of this step is that it achieves a closed-loop and automated output of the fault diagnosis process. The system no longer merely provides a bunch of intermediate data and charts that require expert interpretation, but can directly output actionable and semantically clear diagnostic conclusions, greatly reducing the technical threshold and improving the intelligence level and emergency response efficiency of drainage system operation and maintenance. This is the ultimate embodiment of the core practical value and creativity brought by this invention patent.

[0187] A real-time fault location system for a drainage system, referring to Figure 2 ,include:

[0188] Network construction module 210 is used to construct a directed topology network model of the drainage system based on geographic information system data of the drainage pipe network and the location information of the deployed flow sensors.

[0189] Matrix construction module 220 is used to construct a node-pipe segment association matrix based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system.

[0190] The imbalance calculation module 230 is used to calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system based on the real-time monitoring data of all flow sensors and the node-pipe segment association matrix.

[0191] The direction verification and update module 240 is used to perform dynamic verification and update of the water flow conduction direction of the pipe segment in the directed topology network model of the drainage system based on the real-time pressure or flow symbol data of the key pipe segment.

[0192] The upstream tracing module 250 is used to perform upstream reverse tracing along all inflow edge directions of the specific node in the directed topology network model of the drainage system based on the specific node whose real-time flow imbalance exceeds a preset threshold, and to locate a set of potentially suspicious pipe segment areas.

[0193] The downstream analysis module 260 is used to perform downstream impact analysis along the outflow edge direction of all end nodes in the directed topology network model of the drainage system based on the set of potentially suspicious pipe segment areas, and to determine the downstream impact range.

[0194] The fault determination module 270 is used to perform collaborative analysis and comprehensive judgment based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of the entire network nodes, to determine the specific type and precise geographical location of the fault.

[0195] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time fault location in a drainage system, characterized in that, Includes the following steps: Step S1: Based on the geographic information system data of the drainage pipe network and the location information of the deployed flow sensors, construct a directed topology network model of the drainage system. Step S2: Based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system, construct the node-pipe segment association matrix; Step S3: Based on the real-time monitoring data of all flow sensors and the node-pipe segment correlation matrix, calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system. Step S4: Based on the real-time pressure or flow rate symbol data of key pipe sections, perform dynamic verification and update on the preset pipe section water flow transmission direction in the directed topology network model of the drainage system. Step S5: Based on the specific node whose real-time flow imbalance exceeds a preset threshold, in the directed topology network model of the drainage system, upstream reverse tracing is performed along all inflow edge directions of the specific node to locate the set of potentially suspicious pipe segment areas. Step S6: Based on the set of potentially suspicious pipe segment areas, perform downstream impact analysis along the outflow edge direction of all end nodes in the set in the directed topology network model of the drainage system to determine the downstream impact range. Step S7: Based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of all network nodes, perform collaborative analysis and comprehensive judgment to determine the specific type and precise geographical location of the fault.

2. The real-time fault location method for a drainage system according to claim 1, characterized in that, In step S1, the initial water flow conduction direction of each edge in the network is determined based on the pipeline design parameters and historical water flow data.

3. The real-time fault location method for a drainage system according to claim 1, characterized in that, In step S2, the node-pipe segment association matrix is ​​expressed in matrix form, where the rows of the matrix correspond to the nodes in the network, the columns of the matrix correspond to the pipe segments in the network, and values ​​are assigned to the corresponding positions in the matrix according to the inflow and outflow relationship between the pipe segments and the nodes.

4. The real-time fault location method for a drainage system according to claim 3, characterized in that, In step S3, when calculating the real-time flow imbalance, for each node, the absolute value of the difference between the total inflow and the total outflow calculated based on the node-pipeline correlation matrix and real-time monitoring data is used as the real-time flow imbalance of that node.

5. The real-time fault location method for a drainage system according to claim 4, characterized in that, The difference is defined as the total inflow minus the total outflow. A positive difference indicates a flow surplus, while a negative difference indicates a flow deficit.

6. The real-time fault location method for a drainage system according to claim 1, characterized in that, In step S3, the preset threshold is a dynamic threshold whose size is adaptively adjusted based on the historical flow data and real-time flow change rate of the nodes in the directed topology network model of the drainage system, so as to eliminate the systematic flow deviation under normal operating conditions.

7. The real-time fault location method for a drainage system according to claim 6, characterized in that, In step S4, when dynamically verifying and updating the network direction, the direction of water flow transmission is determined by comparing the signs of the real-time pressure difference between upstream and downstream nodes of the same pipe segment, or by analyzing the signs and magnitudes of the real-time flow rates between different inflow pipe segments of the same node.

8. The real-time fault location method for a drainage system according to claim 3, characterized in that, In steps S5 and S6, when performing upstream reverse tracing or downstream impact analysis, a graph search algorithm is used for traversal, wherein the graph search algorithm is either a breadth-first search algorithm or a depth-first search algorithm.

9. The real-time fault location method for a drainage system according to claim 1, characterized in that, When the set of potentially suspicious pipe segments is concentrated upstream of a certain pipe segment, and the nodes within its downstream influence range show a continuous negative flow imbalance, it is determined that the pipe segment has a blockage fault. When the potentially suspicious pipe section area is discretely distributed and its downstream influence range is not significant, and the imbalance of the abnormal node is a continuous low-amplitude positive value, it is determined to be an instrument measurement drift fault.

10. A real-time fault location system for a drainage system, characterized in that, include: The network construction module is used to construct a directed topology network model of the drainage system based on the geographic information system data of the drainage pipe network and the location information of the deployed flow sensors. The matrix construction module is used to construct a node-pipe segment association matrix based on all inflow edges and all outflow edges of each node in the directed topology network model of the drainage system. The imbalance calculation module is used to calculate the real-time flow imbalance of each node in the directed topology network model of the drainage system based on the real-time monitoring data of all flow sensors and the node-pipe segment correlation matrix. The direction verification and update module is used to perform dynamic verification and update of the water flow conduction direction of the pipe segments in the directed topology network model of the drainage system based on the real-time pressure or flow symbol data of the key pipe segments. The upstream tracing module is used to perform upstream reverse tracing along all inflow edges of the specific node in the directed topology network model of the drainage system, based on the specific node whose real-time flow imbalance exceeds a preset threshold, to locate a set of potentially suspicious pipe segment areas. The downstream analysis module is used to perform downstream impact analysis along the outflow edges of all end nodes in the directed topology network model of the drainage system, based on the set of potentially suspicious pipe segment areas, to determine the downstream impact range. The fault determination module is used to perform collaborative analysis and comprehensive judgment based on the spatial distribution information of the potentially suspicious pipe segment area set, the spatial distribution information of the downstream impact range, and the spatial distribution pattern of the real-time traffic imbalance of all network nodes, to determine the specific type and precise geographical location of the fault.

Images