Drainage network rain and sewage mixed connection tracing and spatial positioning method of micro monitoring station network

Abstract

The application relates to the technical field of data processing, and discloses a rain and sewage mixed connection tracing and spatial positioning method for a drainage pipe network of a micro monitoring station network, which comprises the following steps: according to the flow connection relationship of the drainage pipe network and the layout nodes of the micro monitoring station, the upstream confluence pipe network range of each micro monitoring station is defined, and a unique mapping relationship database between each pipe section and the corresponding multi-level monitoring topological sequence thereof is established; water quality and water level data of the monitoring station are acquired, and an analysis time period is intercepted; based on the space-time propagation correlation attribute of the data between the stations, a target monitoring topological sequence currently producing a response is identified in the mapping relationship database; a homomorphic mapping candidate pipe section set corresponding to the sequence is locked, and a pollutant source discharge inversion model of each candidate pipe section is constructed according to the fluid transport and diffusion mechanism of the pipe network; the model is used to calculate the fitting deviation degree of each candidate pipe section source hypothesis to the monitoring data, and the pipe section with the minimum deviation degree is selected as the positioning result.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks using a micro-monitoring station network. Background Technology

[0002] The separation of rainwater and sewage in urban drainage systems is a core task in water environment management. However, in actual operation, complex misconnections and mixed connections often exist between rainwater and sewage pipe networks, leading to direct discharge of sewage into rivers during dry weather or abnormal concentrations of influent to sewage treatment plants during rainy weather. With the advancement of smart water management, a large number of low-cost, distributed micro-monitoring stations are widely installed in inspection wells and key nodes of drainage pipe networks to collect continuous monitoring data such as water level and water quality in real time. Utilizing these high-density monitoring networks to quickly identify and locate sources of mixed connections has become a key technological direction for improving the efficiency of pipe network operation and maintenance.

[0003] Existing methods for investigating combined sewer overflows mainly fall into two categories: manual investigation and data analysis. Manual investigation methods typically rely on closed-circuit television (CCTV) detection, sonar detection, or dye tracing. While intuitive, these methods are time-consuming, labor-intensive, and struggle to provide real-time coverage of the entire pipe network. Data analysis-based methods primarily trigger alarms by setting numerical thresholds or using the correlation between upstream and downstream monitoring stations for rough assessment. For example, an alarm is triggered when a monitored value exceeds a preset warning line, or the presence of abnormal flow injection is inferred by comparing upstream and downstream liquid level differences. These methods focus more on detecting the existence of abnormal events than precisely pinpointing the exact location of the anomaly.

[0004] However, existing technologies have significant limitations when dealing with complex pipeline network topologies. Since micro-monitoring stations are typically sparsely deployed and concentrated in the main pipeline network, a single monitoring station often corresponds to a vast upstream confluence network area. When abnormal fluctuations occur in monitoring data, existing methods struggle to effectively distinguish which upstream branch pipe is responsible for the discharge, resulting in a many-to-one source ambiguity problem. Furthermore, methods relying on iterative simulations using a full-network hydraulic model have excessive computational loads, making them unsuitable for online real-time analysis. Simple correlation analysis, on the other hand, ignores the time delays and attenuation caused by fluid transport and diffusion within the pipeline network, leading to insufficient accuracy and spatial resolution in the location results, failing to provide precise pipe-segment-level guidance for frontline maintenance personnel. Summary of the Invention

[0005] This invention provides a method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks using a micro-monitoring station network, thus solving the technical problems mentioned in the background art.

[0006] Methods for tracing and spatially locating the source of combined sewer overflows in drainage pipe networks using micro-monitoring station networks include:

[0007] Based on the flow direction connection of the drainage pipe network and the deployment nodes of the micro monitoring stations, the upstream confluence pipe network range of each micro monitoring station is defined, and a mapping relationship library is established between each pipe segment in the drainage pipe network and the multi-level monitoring topology sequence formed by the downstream micro monitoring stations it flows through.

[0008] Water quality and water level monitoring data from each micro-monitoring station are acquired and the analysis period is extracted. Based on the spatiotemporal propagation correlation attributes of data between upstream and downstream micro-monitoring stations, the target monitoring topology sequence that generates a response in the current period is identified in the mapping relationship database.

[0009] The set of isomorphic mapping candidate pipe segments corresponding to the target monitoring topology sequence is locked, and a pollutant source emission inversion model for each candidate pipe segment in the set is constructed based on the pipeline fluid transport and diffusion mechanism.

[0010] The source emission inversion model is used to calculate the fitting deviation of the source assumptions of each candidate pipe segment to the monitoring data, and the candidate pipe segment with the smallest fitting deviation is selected as the spatial location result of the source of mixed rainwater and sewage.

[0011] The beneficial effects of this invention are as follows:

[0012] 1. Existing technologies largely rely on single-point threshold alarms or simple upstream-downstream correlation analysis. When faced with complex pipeline networks, it is difficult to distinguish specific emission sources within the vast upstream confluence area. This invention transforms the complex network-based source tracing problem into a structured sequence matching problem by constructing a multi-level monitoring topology sequence and mapping relationship library. This allows for precise targeting of the investigation scope from the entire watershed to a specific set of isomorphic mapped pipe segments, even when monitoring stations are sparse, by utilizing topological constraints, effectively solving the many-to-one source tracing problem.

[0013] 2. Addressing the technical challenge of multiple pipe segments potentially mapping to the same set of monitoring stations, this invention does not rely on expensive and time-consuming full-network hydraulic simulation. Instead, it utilizes the fluid transport and diffusion mechanism of the pipe network to construct a deterministic source emission inversion model. By introducing a source strength back-calculation operator and a ridge regression algorithm, this invention can capture the minute transport delay and attenuation differences (i.e., fitting bias) between different candidate pipe segments reaching the monitoring stations. This allows even adjacent pipe segments on the same confluence path to be effectively distinguished through data fit, thus achieving precise pipe segment-level positioning beyond the density of monitoring stations.

[0014] 3. This invention does not rely on threshold-based conditional logic, but instead employs energy-maximizing analysis time segmentation and least-squares-based global optimal solution search. This avoids false negatives or missed alarms caused by improper threshold settings, and eliminates the need to maintain complex historical state caches. This invention has low computational overhead, is easily deployed on edge computing gateways or cloud servers, and can meet the real-time online monitoring needs of large-scale urban drainage networks. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method for tracing and spatially locating the source of rainwater and sewage mixing in the drainage pipe network of the present invention using a micro monitoring station network;

[0016] Figure 2 This is a schematic diagram of an implementation scenario of the present invention. Detailed Implementation

[0017] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0018] like Figure 1 As shown, the method for tracing and spatially locating the source of stormwater and sewage mixing in the drainage pipe network of the micro-monitoring station network includes:

[0019] Based on the flow direction connection of the drainage pipe network and the deployment nodes of the micro monitoring stations, the upstream confluence pipe network range of each micro monitoring station is defined, and a mapping relationship library is established between each pipe segment in the drainage pipe network and the multi-level monitoring topology sequence formed by the downstream micro monitoring stations it flows through.

[0020] Water quality and water level monitoring data from each micro-monitoring station are acquired and the analysis period is extracted. Based on the spatiotemporal propagation correlation attributes of data between upstream and downstream micro-monitoring stations, the target monitoring topology sequence that generates a response in the current period is identified in the mapping relationship database.

[0021] The set of isomorphic mapping candidate pipe segments corresponding to the target monitoring topology sequence is locked, and a pollutant source emission inversion model for each candidate pipe segment in the set is constructed based on the pipeline fluid transport and diffusion mechanism.

[0022] The source emission inversion model is used to calculate the fitting deviation of the source assumptions of each candidate pipe segment to the monitoring data, and the candidate pipe segment with the smallest fitting deviation is selected as the spatial location result of the source of mixed rainwater and sewage.

[0023] In a preferred embodiment, the upstream catchment area of ​​each micro-monitoring station is defined based on the flow direction connection of the drainage network and the deployment nodes of the micro-monitoring stations, including:

[0024] The manholes and pipe segments in the drainage pipe network are constructed as a directed graph model. ,in For a set of nodes, Generate the edge set and generate a directed adjacency matrix. ;

[0025] Set the maximum effective topology hops for network propagation. Calculate the directed reachability matrix of the entire network. :

[0026] ;

[0027] in, Represents the directed adjacency matrix Power of 1 This is a sign function; it takes the value 1 if the element value is greater than 0, and 0 otherwise.

[0028] For each micro-monitoring station node Define its upstream manifold network range for:

[0029] ;

[0030] in, Indicates from the upstream node Flowing to downstream nodes pipe section, Indicates from node To the micro monitoring station node A directed path exists.

[0031] Preferably, in a drainage network in the old urban area of ​​a city after the rainwater and sewage separation renovation, the network includes 120 inspection wells and 150 pipe sections, and miniature monitoring stations have been deployed at 8 key inspection well nodes. First, a directed graph model is constructed of the inspection wells and pipe sections in this drainage network. ,in This is a set of nodes, corresponding to 120 manholes in the pipeline network, with each manhole corresponding to... One of the elements; Let the set of edges correspond to the 150 pipe segments in the pipeline network, with each pipe segment corresponding to... A directed edge in ,in This indicates the upstream inspection well node of this pipe section. This represents the downstream inspection well node of this pipe segment. Based on this directed graph model, a directed adjacency matrix is ​​generated. , It is a dimension (i.e., a 120×120 matrix, the elements in the matrix) The rule for determining the value is: when there is a slave node To the node When the pipe section, ,otherwise For example, if there is a pipeline in the network from inspection well 5 (corresponding node) ) to inspection well 6 (corresponding node) For the pipe section, then ,and ( The value of ) is 0.

[0032] Next, we define the maximum effective topology hop count for network propagation. , The preferred value for is 10. This value is based on the fact that the longest path in the actual topology of the drainage network contains 10 pipe segments, which ensures coverage of all possible directed paths in the network while avoiding unnecessary computational redundancy. The value of is used to calculate the directed reachability matrix of the entire network. The corresponding calculation formula is: ,in Represents a directed adjacency matrix of Power of 1 The element represents the node. To the node After no more than The total number of paths in the pipeline segment; It is a symbolic function, its function is to... Each element in the matrix is ​​processed, taking a value of 1 if the element's value is greater than 0, and 0 otherwise. Therefore, the directed reachability matrix... elements in The meaning is: from node To the micro monitoring station node Does it exist that does not exceed A directed path for a pipe segment, if it exists. ,otherwise .

[0033] Subsequently, for each micro-monitoring station node (i.e., the nodes corresponding to the 8 miniature monitoring stations deployed at the inspection wells), defining the upstream catchment network range. The corresponding calculation formula is: ,in Indicates from the upstream node Flowing to downstream nodes pipe section, Indicates from node To the micro monitoring station node There exists a directed path, therefore The corresponding set is all those that satisfy its upstream nodes Can be reached via a directed path A collection of pipe segments, for example, if a micro monitoring station node The equipment corresponding to the installation in inspection well 20 corresponding nodes Including the nodes corresponding to inspection wells 15, 16, 17, etc., then the pipe section , , All of these belong to The corresponding upstream manifold network range.

[0034] Specifically, by constructing a directed graph model The connection relationships of drainage pipe networks can be quantified into topological structures; directed adjacency matrix This serves as a numerical representation of the topology, providing a foundation for subsequent reachability analysis; it also sets the effective maximum number of topological hops. This involves controlling the computational complexity while ensuring coverage of all valid paths in the pipeline network; calculating the directed reachability matrix. This is to clarify the reachability relationship from any node to the micro-monitoring station node, and based on Defined This allows for the accurate identification of all factors that can affect the micro-monitoring station through water flow. The pipe section range is the topological boundary for determining the source of pollution when conducting subsequent source tracing of combined sewer overflows.

[0035] In a preferred embodiment, a mapping relationship library is established between each pipe segment in the drainage network and a multi-level monitoring topology sequence formed by the downstream micro-monitoring stations through which it flows, including:

[0036] For any pipe segment According to the upstream confluence network range Identify the set of associated monitoring stations for this pipe section. :

[0037] ;

[0038] in, It is a collection of micro monitoring stations across the entire network;

[0039] Compute set Various micro monitoring stations in China Shortest directed path length to the designated outlet This is used as the basis for sorting to generate a multi-level monitoring topology sequence. :

[0040] ;

[0041] The entire network will have the same multi-level monitoring topology sequence The pipe segments are grouped into the same isomorphic mapping set. :

[0042] ;

[0043] Establish a multi-level monitoring topology sequence Pointer to isomorphic mapping set The index dictionary forms the mapping relationship library.

[0044] Preferably, a mapping relationship database is established between each pipe segment in the drainage network and the multi-level monitoring topology sequence formed by the downstream micro-monitoring stations through which it flows. The specific process is as follows: First, for any pipe segment... pipe section For example, based on the upstream catchment network range of each micro-monitoring station Identify the set of associated monitoring stations for this pipe section. The corresponding calculation formula is: ,in It is a collection of all the network's micro monitoring stations (i.e., the set of nodes corresponding to the 8 monitoring stations in this scenario). Indicates pipe section Belongs to monitoring station The upstream confluence network range, if the monitoring station (Corresponding to node 20) Includes pipe section Monitoring Station (Corresponding to node 30) If it also includes this pipe section, then .

[0045] Next, calculate the set. Various micro monitoring stations in China Shortest directed path length to the designated outlet (node ​​120) The value of this length is derived from the monitoring station. The number of pipe segments included in the path from the corresponding node to the specified outlet node, such as the number of monitoring stations. The shortest path from node 20 to node 120 is 20→25→30→35→…→120, which includes 12 pipe segments. Monitoring station The shortest path from node 30 to node 120 contains 10 pipe segments, therefore Then, a multi-level monitoring topology sequence is generated based on this sorting. The corresponding calculation formula is: ,in The operation is to The monitoring stations in the middle are arranged according to The values ​​are arranged in ascending order, therefore for After sorting, we get .

[0046] Subsequently, the entire network with the same multi-level monitoring topology sequence will be... The pipe segments are grouped into the same isomorphic mapping set. The corresponding calculation formula is: For example, pipe section of Also ,and , Therefore ,at this time .

[0047] Finally, a multi-level monitoring topology sequence was established. Pointer to isomorphic mapping set The index dictionary, i.e., the record , The corresponding relationships are used to form the mapping relationship library.

[0048] Specifically, through It can clearly identify all micro monitoring stations that a single pipe segment can affect, ensuring the coverage of subsequent signal correlation; This reflects the downstream location of the monitoring stations within the pipeline network, and is obtained by sorting them accordingly. This can reflect the topological order of the monitoring stations corresponding to the pipe segment, avoiding matching chaos caused by unordered sets; and will include the same The pipe section was included It can classify pipe segments with consistent topological monitoring characteristics in the pipeline network, reducing the number of computational objects in subsequent analysis; while the establishment of the index dictionary can realize the rapid mapping of multi-level monitoring topology sequences to pipe segment sets, providing an efficient retrieval basis for subsequent pipe segment location based on monitoring signals.

[0049] In a preferred embodiment, acquiring water quality and water level monitoring data from each micro-monitoring station and selecting the analysis period includes:

[0050] For each micro monitoring station The first collection Time series of parameters Standardization and weighted fusion are performed to generate a one-dimensional state signal. :

[0051] ;

[0052] in, For the total number of parameters, For the first Preset fusion weights for each parameter, and These are the historical mean and standard deviation of the parameter, respectively.

[0053] Calculate the first-order difference of a one-dimensional state signal The total energy curve of the entire network is obtained by superimposing the data. :

[0054] ;

[0055] ;

[0056] in, The sampling period is It is a collection of micro monitoring stations across the entire network;

[0057] The point at which the total energy curve of the entire network reaches its global maximum value. :

[0058] ;

[0059] Centered on this time point, the length is... Analysis period :

[0060] ;

[0061] Preferably, each micro monitoring station collects 4 parameters (total number of parameters). (corresponding to conductivity, COD, ammonia nitrogen, and water level respectively), sampling period The preferred interval is 1 minute. This value is chosen because the rate of change of water quality parameters in the drainage network is moderate; a 1-minute interval can capture effective changes without generating excessive redundant data. For each micro-monitoring station... , collect its first Time series of parameters Meanwhile, historical monitoring data for this parameter over the past 7 days were selected to calculate the historical average. and standard deviation .

[0062] Next, the time series Standardization and weighted fusion are performed to generate a one-dimensional state signal. The corresponding calculation formula is: ,in For the first The preset fusion weights for each parameter have the preferred values. , , , The consideration for this value is that conductivity is more sensitive to the response of rainwater and sewage mixing, so it is given a higher weight, while water level is relatively weaker and is therefore given a lower weight. Standardization of this parameter can eliminate the differences in the dimensions of different parameters.

[0063] Then, the first-order difference of the one-dimensional state signal is calculated. The corresponding calculation formula is: This formula is used to reflect The rate of change at adjacent time points; then superimposed to obtain the total energy curve of the entire network. The corresponding calculation formula is: This formula amplifies the abrupt changes in the signal by summing the squares of the rates of change over all monitoring stations. For example, monitoring stations exist hour , hour ,but If other monitoring stations If they are 0.5, 0.6, etc., then This is the sum of the squares of these values.

[0064] Subsequently, the time it takes for the total energy curve of the entire network to reach its global maximum value is searched. The corresponding calculation formula is: This formula is used to locate the moment when the signal changes most drastically, that is, the core response period of a rainwater and sewage mixing event, for example by traversing... Time series data, find .

[0065] Finally, taking that time point as the center, a section of length is extracted. Analysis period The corresponding calculation formula is: ,in The preferred value is 10 minutes. This value is chosen because the propagation process of the stormwater and sewage mixing signal can usually cover adjacent monitoring stations within 10 minutes, thus fully encompassing the event's response process; substituting... , minutes, can get .

[0066] Specifically, standardization eliminates the dimensional differences between different parameters, while weighted fusion highlights parameters that are more sensitive to mixing. It can comprehensively reflect the pollution status of the monitoring station; first-order difference Capture the rate of change of state signals, total network energy This aggregates abrupt change information from all monitoring stations, avoiding noise interference from a single monitoring station; and locates the global maximum energy. It can accurately pinpoint the core time period of an event, with a truncated length of [length missing]. of This ensures that the complete response process of the event is included, providing a valid data period for subsequent signal correlation analysis.

[0067] In a preferred embodiment, based on the spatiotemporal propagation correlation attributes of data between upstream and downstream micro-monitoring stations, the target monitoring topology sequence that generates a response in the current time period is identified in the mapping relationship database, including:

[0068] For any multi-level monitoring topology sequence in the mapping relation database Calculate the adjacent first and second digits in the sequence. Mini monitoring station With the Mini monitoring station During the analysis period Spatiotemporal propagation delay within :

[0069] ;

[0070] in, and For the corresponding one-dimensional state signal, For search latency;

[0071] Calculate the amplitude attenuation ratio between adjacent stations :

[0072] ;

[0073] Calculate the sequence matching residuals of this multi-level monitoring topology sequence. :

[0074] ;

[0075] Selecting sequence matching residuals minimal sequence As a target monitoring topology sequence:

[0076] ;

[0077] in, It is the set of all multi-level monitoring topology sequences in the mapping relationship database.

[0078] Preferably, the set of all multi-level monitoring topology sequences in the mapping relationship database is denoted as... , including , Sequences, analysis period For any multi-level monitoring topology sequence in the mapping relationship database ,by For example, among which This indicates the number of micro-monitoring stations contained in the sequence. First, the number of adjacent stations in the sequence is calculated. Mini monitoring station With the Mini monitoring station During the analysis period Spatiotemporal propagation delay within The corresponding calculation formula is: ,in For mini monitoring stations One-dimensional state signal, For mini monitoring stations exist One-dimensional state signal at time t. To minimize search latency, the preferred search range is set as follows: The timeframe of 5 minutes is chosen because the signal propagation time between adjacent micro-monitoring stations in the drainage network typically does not exceed 5 minutes, allowing for coverage of potential time delays while controlling computational load; for ,Pick Corresponding adjacent pairs and traversal exist The values ​​within are calculated for each. The corresponding summation result, when The summation result reaches its maximum at this time, therefore minute.

[0079] Next, the amplitude attenuation ratio between adjacent stations is calculated. The corresponding calculation formula is: ,in For mini monitoring stations exist One-dimensional state signal at time t. For mini monitoring stations exist A one-dimensional state signal at a given time.

[0080] Then, the sequence matching residuals of this multi-level monitoring topology sequence were calculated. The corresponding calculation formula is: ,in For mini monitoring stations One-dimensional state signal and Delay And scale The deviation between the signals after the signal.

[0081] Then the set All multi-level monitoring topology sequences (such as Repeat the above steps to calculate their respective... .

[0082] Specifically, by calculating the spatiotemporal propagation delay The time difference in signal propagation between upstream and downstream micro-monitoring stations can be determined because pollution signals from mixed rainwater and sewage flow along the pipe network, and there is a fixed time delay between upstream and downstream signals, maximizing cross-correlation. That is, to correspond to the time delay; calculate the amplitude attenuation ratio. This is because the amplitude of the pollution signal attenuates during propagation in the pipeline network due to factors such as dilution, and this proportion can quantify the degree of attenuation; sequence matching residual. By considering the matching of time delay and attenuation, the smaller the residual, the more the signal of the micro-monitoring station corresponding to the sequence conforms to the propagation correlation characteristics of rainwater and sewage mixing. By selecting the sequence with the smallest residual as the target, the monitoring topology sequence that best matches the signal response of the current time period can be locked from the mapping relationship library, providing a basis for subsequent pipe segment location.

[0083] In a preferred embodiment, locking the set of isomorphic mapping candidate pipe segments corresponding to the target monitoring topology sequence includes:

[0084] With the target monitoring topology sequence As an index, in the dictionary of the mapping relation library Retrieve the corresponding isomorphic mapping set Lock it into the set of candidate pipe segments for isomorphic mapping:

[0085] ;

[0086] in, For the entire network management segment collection, For pipe section The corresponding multi-level monitoring topology sequence.

[0087] Preferred target monitoring topology sequence for The dictionary of the mapping relation library It is an index dictionary previously established, pointing to corresponding isomorphic mapping sets from multi-level monitoring topology sequences, and a complete set of network management segments. Includes 150 pipe segments in this scenario. For pipe section The corresponding multi-level monitoring topology sequence (e.g., pipe segment) of Pipeline section of ).

[0088] Next, using the target monitoring topology sequence As an index, retrieve the corresponding isomorphic mapping set from the dictionary of the mapping relation library. The corresponding calculation formula is: ,in Dictionary China and Israel The value corresponding to the key. This indicates that the pipe segment belongs to the entire network of pipe segments. Indicates pipe section Corresponding multi-level monitoring topology sequence and target monitoring topology sequence Consistency; Search Dictionary Chinese correspondence The entries, obtained And lock this set as a set of candidate pipe segments for isomorphic mapping.

[0089] Specifically, since pipe segments with the same multi-level monitoring topology sequence have been grouped into corresponding isomorphic mapping sets, and an index relationship between the sequence and the set has been established through a dictionary, the target monitoring topology sequence... It is the sequence that best matches the characteristics associated with signal propagation in the current time period, therefore it is... By indexing and retrieving the corresponding isomorphic mapping set, all pipe segments with the sequence characteristics can be directly identified. These pipe segments are the possible source pipe segments corresponding to the current rainwater and sewage mixing event, thereby narrowing the scope of subsequent positioning and focusing the analysis on the set of pipe segments that match the signal propagation characteristics.

[0090] In a preferred embodiment, based on the pipeline network fluid transport and diffusion mechanism, a pollutant source emission inversion model is constructed for each candidate pipe segment in the set, including:

[0091] Obtain target monitoring topology sequence Pipeline transport distance between adjacent micro monitoring stations Combined with spatiotemporal propagation delay Calculate the equivalent transport velocity of the interval :

[0092] ;

[0093] For the isomorphic mapping candidate pipe segment set Any candidate pipe segment Calculate the number of its arrival sequence. Mini monitoring station transmission time :

[0094] ;

[0095] in, Indicates pipe section The index of the interval it belongs to;

[0096] Constructing this candidate pipe segment for the first Unit impulse response function of a micro-monitoring station :

[0097] ;

[0098] in, Where is the diffusion coefficient. The function is a time variable. These constitute the basic operators of the pollutant source emission inversion model.

[0099] Preferred target monitoring topology sequence First, the pipeline transport distance between adjacent micro-monitoring stations in the sequence is obtained. The adjacent micro monitoring station here is and (correspond ), for The corresponding node to The total length of the pipe segments between the corresponding nodes, obtained from the pipeline network GIS data, is 200 meters; this is combined with the previously calculated spatiotemporal propagation delay of this adjacent pair. Calculate the equivalent transport velocity in this interval within minutes. The corresponding calculation formula is: ,in Monitoring topology sequence for the target The middle adjacent The pipeline transport distance between individual micro-monitoring stations The spatiotemporal propagation delay of this adjacent pair, The equivalent transport velocity in this interval can be obtained by substituting the values. meters per minute.

[0100] Next, for the isomorphic mapping candidate pipe segment set Any candidate pipe segment ,by For example, calculate the number of arrivals in the sequence. Mini monitoring station transmission time The corresponding calculation formula is: ,in Indicates pipe section The interval index, the segment here belong The corresponding upstream interval, and the corresponding interval index. ; Let be the sequence number of the micro-monitoring station in the sequence, if correspond ,but Substitute rice, meters per minute, can be obtained minutes; if correspond ,but The summation term is empty. minute.

[0101] Then, the candidate pipe segment was constructed for the first... Unit impulse response function of a micro-monitoring station The corresponding calculation formula is: ,in The diffusion coefficient is preferably 0.5 square meters per minute, which is determined based on the typical diffusion characteristics of sewage in urban drainage networks. It is a time variable; For pipe section Reaching the The propagation time of a micro-monitoring station. , For example, substitute minute, square meters per minute, which yields This function is the candidate pipe segment pair. The unit impulse response function constitutes the basic operator of the pollutant source emission inversion model.

[0102] Specifically, the equivalent transport velocity is calculated by combining the pipeline transport distance with the spatiotemporal propagation delay. This is based on the correspondence between the distance and time of fluid transport in the pipeline network, which can quantify the flow velocity of pollution signals between adjacent monitoring stations. The propagation time from the pipe section to the micro-monitoring station is calculated because the pollution emitted from the pipe section will flow along the pipeline network, and the time it takes to reach the monitoring station is determined by both the flow velocity and distance in its respective section, thus clarifying the time delay pattern of the pollution signal. The unit impulse response function is constructed because pollutants undergo both transport and diffusion in the pipeline network. This function can describe the change in concentration of a unit intensity of pollution at the monitoring station over time after being emitted from the pipe section. This serves as the basic operator for the inversion model, reflecting the actual process of pollution propagation and providing a physical calculation basis for subsequent inversion of the pollution source pipe section.

[0103] In a preferred embodiment, the pollutant source emission inversion model is used to calculate the fitting deviation of the source assumptions of each candidate pipe segment to the monitoring data, and the candidate pipe segment with the smallest fitting deviation is selected as the spatial location result of the combined sewer overflow source, including:

[0104] In the target monitoring topology sequence Single-dimensional state signal of a micro-monitoring station Stacked into full sequence observation vectors :

[0105] ;

[0106] For each candidate pipe segment in the isomorphic mapping candidate pipe segment set Using the unit impulse response function Constructing a block-shaped Topplitz form of the transmission response matrix ;

[0107] Introducing regularization parameters The ridge regression algorithm is used to obtain the source emission intensity sequence estimate for this candidate pipe segment. :

[0108] ;

[0109] in, It is the identity matrix;

[0110] Calculate the fitting deviation of the candidate pipe segment. , defined as the sum of squares of the reconstructed residuals:

[0111] ;

[0112] Selecting the fit deviation Minimum candidate pipe segment Spatial location results as the source of combined sewer overflow:

[0113] ;

[0114] in, This is the set of candidate pipe segments for isomorphic mapping.

[0115] Preferred target monitoring topology sequence ,in This indicates the number of micro-monitoring stations included in the sequence. First, this... Single-dimensional state signal of a micro-monitoring station Stacked into full sequence observation vectors The corresponding expression is ,in It is the first micro-monitoring station in the sequence. A one-dimensional state signal time-series vector (containing all sample values ​​within the analysis period). It is the second micro-monitoring station in the sequence. The single-dimensional state signal time-series vector, after stacking It integrates the observation data from all monitoring stations in the target sequence to form a unified computational object.

[0116] Next, for the isomorphic mapping candidate pipe segment set Each candidate pipe segment ,by For example, using the unit impulse response function Constructing a block-shaped Topplitz form of the transmission response matrix , The structure is as follows: each block corresponds to the first... The impulse response of each micro-monitoring station, with elements within the block arranged in a Toplitz pattern (i.e., elements are equal along the antidiagonal), thereby diverting pollution from the pipe section. The propagation process to each monitoring station is transformed into a transmission matrix of a linear system, realizing a linear mapping from the source strength sequence to the observed signal.

[0117] Then, regularization parameters were introduced. The preferred value is 0.01. This value balances fitting accuracy and solution stability, avoiding solution fluctuations caused by column correlation in the transmission response matrix. The ridge regression algorithm is used to estimate the source emission intensity sequence for this candidate pipe segment. The corresponding calculation formula is: ,in yes The transpose of the matrix, Is with The same-dimensional identity matrix, this formula is derived by introducing Correction The diagonal elements are used to ensure that the matrix is ​​invertible and to obtain stable source strength estimation results.

[0118] Then, the fit deviation of the candidate pipe segment is calculated. It is defined as the sum of squares of the reconstructed residuals, and the corresponding calculation formula is: ,in It is the square of the 2-norm, representing the sum of the squares of the elements of the vector. This value quantifies the... When the pollution source is identified, the difference between the simulated observation signal and the actual monitoring signal is considered. For candidate pipe sections... Repeat the above steps to calculate the corresponding... ,For example of , of .

[0119] Finally, the fit deviation was selected. Minimum candidate pipe segment As the spatial location result of the source of combined sewer overflow, the corresponding calculation formula is: , here Less than Therefore .

[0120] Specifically, stacking signals from multiple monitoring stations into an observation vector integrates scattered observation data into a unified computational input, facilitating the subsequent solution of the linear system. Constructing a block-shaped Topplitz-form transport response matrix is ​​beneficial because pollutant propagation in the pipeline network exhibits time-invariant characteristics, and the Topplitz matrix accurately describes this time-invariant linear transport relationship. Introducing a ridge regression algorithm with regularization parameters addresses potential ill-conditioned problems in the transport response matrix, ensuring the stability of source strength estimation. The calculation of the fitting bias measures the degree of matching between candidate pipe segments as pollution sources by quantifying the difference between simulated signals and actual observations. The pipe segment with the smallest bias is selected because the propagation process corresponding to this segment best explains the actual monitored signal changes, thereby achieving accurate source location.

[0121] like Figure 2 As shown, for the scenario of tracing the source and spatial positioning of mixed stormwater and sewage connections in urban drainage pipe networks: multiple micro monitoring stations are deployed at key nodes such as stormwater pipes, sewage pipes and their inspection wells. The micro monitoring stations collect continuous monitoring data such as water level and water quality, and upload the monitoring data to the platform via wireless link; the platform analyzes the abnormal propagation process, outputs the positioning results of the pipe section where the suspected source of the mixed connection is located, and can send the positioning results to on-site inspection personnel to guide the verification and handling of the corresponding inspection well / pipe section.

[0122] It is important to note that all input data described in this solution is acquired in real-time through legal and compliant hardware interfaces with the user's full knowledge, explicit consent, and active cooperation. The preset parameters, prior constants, and statistical means are all derived from publicly available scientific literature data, de-identified general research datasets, or calibration data from laboratory environments, and do not contain any unauthorized sensitive third-party information. The system's data processing is limited to local or volatile memory computation transmitted via encrypted channels. There is no illegal collection, theft, or retention of user biometric data or infringement of user privacy without the user's knowledge. All parameter calls and generation comply with the principles of data minimization, legality, legitimacy, and necessity.

[0123] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A method for tracing and spatially locating the source of combined sewer overflows in drainage pipe networks using a micro-monitoring station network, characterized in that, include: Based on the flow direction connection of the drainage pipe network and the deployment nodes of the micro-monitoring stations, the upstream confluence pipe network range of each micro-monitoring station is defined, and a mapping relationship library is established between each pipe segment in the drainage pipe network and the multi-level monitoring topology sequence formed by the downstream micro-monitoring stations it flows through. This includes: traversing each pipe segment in the drainage pipe network, identifying all micro-monitoring stations that can be topologically affected when the pipe segment discharges, based on the upstream confluence pipe network range, forming an associated monitoring station set for the pipe segment; calculating the topological distance from each micro-monitoring station in the associated monitoring station set to the drainage pipe network outlet, and sorting the associated monitoring station set in order of topological distance from farthest to nearest, generating the multi-level monitoring topology sequence corresponding to the pipe segment; performing equivalence class grouping on all pipe segments in the entire network, grouping pipe segments with completely identical multi-level monitoring topology sequences into the same isomorphic mapping set, and establishing an index dictionary from the multi-level monitoring topology sequence to the isomorphic mapping set, forming the mapping relationship library; The process involves acquiring water quality and water level monitoring data from each micro-monitoring station and selecting an analysis period. This includes: performing statistical standardization on the multidimensional water quality and water level monitoring data collected from each micro-monitoring station, and using preset fusion weights to weighted sum the multidimensional data to generate a single-dimensional state signal for that micro-monitoring station; calculating the first-order difference of the single-dimensional state signal with respect to time, and using the square of this first-order difference as the instantaneous change energy; superimposing the instantaneous change energy of all micro-monitoring stations in the entire network at the same time to obtain the total energy curve of the entire network; searching for the time point when the total energy curve of the entire network reaches its global maximum value, and selecting a time window of a preset fixed length centered on this time point as the analysis period. Based on the spatiotemporal propagation correlation attributes of data between upstream and downstream micro-monitoring stations, the target monitoring topology sequence that generates a response in the current time period is identified in the mapping relationship database. This includes: traversing each multi-level monitoring topology sequence in the mapping relationship database, extracting adjacent upstream and downstream micro-monitoring stations in the multi-level monitoring topology sequence, calculating the peak value of the cross-correlation function of the corresponding single-dimensional state signals of the upstream and downstream micro-monitoring stations in the analysis time period, and obtaining the spatiotemporal propagation delay; using the spatiotemporal propagation delay to perform time shifting on the single-dimensional state signal of the upstream micro-monitoring station, calculating the least squares scaling factor between the shifted signal and the single-dimensional state signal of the downstream micro-monitoring station as the amplitude attenuation ratio; constructing a prediction model using the spatiotemporal propagation delay and the amplitude attenuation ratio, calculating the sum of squared signal fitting residuals between all adjacent stations in the multi-level monitoring topology sequence as the sequence matching residual; comparing the sequence matching residuals of all multi-level monitoring topology sequences, and selecting the multi-level monitoring topology sequence with the smallest sequence matching residual value as the target monitoring topology sequence that generates a response in the current time period. The set of isomorphic mapping candidate pipe segments corresponding to the target monitoring topology sequence is locked, and a pollutant source emission inversion model for each candidate pipe segment in the set is constructed based on the pipeline fluid transport and diffusion mechanism. The source emission inversion model is used to calculate the fitting deviation of the source assumptions of each candidate pipe segment to the monitoring data, and the candidate pipe segment with the smallest fitting deviation is selected as the spatial location result of the source of mixed rainwater and sewage.

2. The method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks according to claim 1, characterized in that, Based on the flow direction and connection relationship of the drainage pipe network and the deployment nodes of the micro monitoring stations, the upstream catchment network range of each micro monitoring station is defined, including: The manholes and pipe segments in the drainage network are constructed as a directed graph model, and a directed adjacency matrix describing the direct connection relationship between nodes is generated according to the water flow direction of the pipe segment. Set the effective maximum topological hop count for network propagation, and use the directed adjacency matrix to calculate the directed reachability matrix of the entire network. The directed reachability matrix is ​​used to characterize whether there is a connected path between any two nodes that satisfies the hop count limit. For each micro-monitoring station, based on the directional reachability matrix, all upstream nodes that can reach the micro-monitoring station along the water flow direction are selected, and the set of connecting pipe segments between these upstream nodes is taken as the upstream confluence network range of the micro-monitoring station.

3. The method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks according to claim 1, characterized in that, Locking the set of isomorphic mapping candidate pipe segments corresponding to the target monitoring topology sequence, including: Using the target monitoring topology sequence as the retrieval key, a matching retrieval is performed in the mapping relationship database, and the isomorphic mapping set pointed to by the retrieval key is extracted as the current isomorphic mapping candidate pipe segment set. The isomorphic mapping candidate pipe segment set includes all drainage pipe segments in the entire network that have the same multi-level monitoring topology sequence in structure.

4. The method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks according to claim 3, characterized in that, Based on the fluid transport and diffusion mechanism in pipeline networks, a pollutant source emission inversion model is constructed for each candidate pipe segment in the dataset, including: The pipeline transport distance between adjacent micro-monitoring stations in the target monitoring topology sequence is calculated based on the geometric properties of the drainage pipeline network. The equivalent transport velocity of each interval is obtained by dividing the pipeline transport distance by the spatiotemporal propagation delay. For each candidate pipe segment in the isomorphic mapping candidate pipe segment set, the propagation time of the candidate pipe segment to each micro-monitoring station in the sequence is calculated according to the equivalent transport velocity. Using the Green's function form of the one-dimensional convection-diffusion equation, combined with the preset diffusion coefficient and the propagation time, a unit impulse response function describing the response generated at each micro-monitoring station when the candidate pipe section experiences unit intensity emission is constructed.

5. The method for tracing and spatially locating the source of rainwater and sewage mixing in drainage pipe networks according to claim 4, characterized in that, The pollutant source emission inversion model is used to calculate the fitting deviation of the source assumptions of each candidate pipe segment to the monitoring data. The candidate pipe segment with the smallest fitting deviation is selected as the spatial location result of the combined sewer overflow source, including: The single-dimensional state signals of all micro-monitoring stations in the target monitoring topology sequence are stacked in the order of the stations to form a full sequence observation vector; For each candidate pipe segment in the isomorphic mapping candidate pipe segment set, a transmission response matrix describing the spatiotemporal convolution relationship between the source emission of the candidate pipe segment and the full sequence observation vector is generated based on the unit impulse response function. By introducing a regularization parameter and using the ridge regression algorithm to perform deconvolution operation on the transmission response matrix and the full sequence observation vector, the source emission intensity sequence estimate of the candidate pipe segment is obtained. The theoretical observation vector is reconstructed using the source emission intensity sequence estimate and the transmission response matrix. The squared Euclidean distance between the theoretical observation vector and the actual full sequence observation vector is calculated to obtain the fitting deviation of the candidate pipe segment. Traverse the set of candidate pipe segments in the isomorphic mapping, compare the fitting deviation of all candidate pipe segments, and determine the candidate pipe segment with the smallest fitting deviation value as the spatial location result of the source of rainwater and sewage mixing.

Images