Water pollution source tracking analysis method and system based on internet of things

Abstract

The application discloses a water pollution source tracking analysis method and system based on Internet of Things, and relates to the technical fields of water quality detection and water pollution source tracking analysis. The method comprises the following steps: obtaining Internet of Things water quality monitoring data and determining a pollution response time window; generating a candidate pollution source set and its propagation path to each monitoring node according to a water system topology; calculating a pollution propagation time range; forming a discharge time positive constraint domain for pollution positive response monitoring nodes, and performing intersection update on the feasible discharge time domain of the candidate pollution source; forming a contradictory discharge time sub-domain for unresponsive monitoring nodes that simultaneously meet the propagation reachability, monitoring time coverage, theoretical detectability and effective detection data, and performing difference set update; performing concentration propagation consistency verification on the remaining candidate pollution sources; when new detection data arrives, only the candidate pollution sources with propagation correlation are subjected to local incremental update; and finally, the pollution source position, feasible discharge time range and pollution propagation path are output.

Description

Technical Field

[0001] This invention relates to the field of water quality testing and water pollution source tracing and analysis technology, specifically to a water pollution source tracing and analysis method and system based on the Internet of Things, which is particularly applicable to rivers, river networks, drainage pipe networks and other aquatic systems with clear pollution propagation topological relationships and flow directions. Background Technology

[0002] Following a water pollution incident, it is typically necessary to trace the pollution source upstream by combining water quality monitoring results, river or pipe network topology, flow direction, flow velocity, and pollutant migration patterns. With the development of Internet of Things (IoT) technology, an increasing number of water quality monitoring nodes are being deployed in rivers, river networks, and drainage pipe networks, enabling continuous collection of time-series data on the concentration of target pollutants, providing a rich data foundation for pollution source tracing.

[0003] Existing technologies include methods that use sampling time, distance, and flow velocity to perform stepwise time-correlation backtracking, and infer the location of pollution sources by comparing measured concentrations with simulated concentrations. Other methods continuously update pollution source parameters through drainage behavior profiling, time-varying priors, and probabilistic inversion. Related research also widely employs hydrodynamic-water quality models, Bayesian / Markov chain Monte Carlo methods, genetic algorithms, and data assimilation to invert pollution source location, emission time, and emission intensity.

[0004] However, the above-mentioned existing technical solutions still have the following shortcomings:

[0005] First, the pollution response of a single or small number of monitoring nodes can typically correspond to multiple pollution source locations and multiple emission times. In existing schemes, time information is mainly used for matching, scoring, or probabilistic inversion, lacking a mechanism to uniformly transform the pollution response of multiple nodes into a deterministic constraint on the emission time of candidate pollution sources.

[0006] Secondly, monitoring nodes that fail to detect a pollution response are easily mistakenly identified as "uncontaminated nodes" to rule out pollution sources. However, this approach carries a significant risk of misjudgment. In reality, a lack of detected pollution response can be caused by various reasons: the monitoring period may not cover the theoretical arrival time of the pollutant, the theoretical pollution concentration may be lower than the node's effective detection conditions, or the detection data may be invalid due to equipment malfunction or communication problems. Directly excluding pollution sources based on "no response" can easily lead to the incorrect exclusion of the true pollution source.

[0007] Third, IoT monitoring data arrives continuously. As the amount of data increases, if a global simulation or inversion calculation is re-executed every time new data arrives, the computational load will increase dramatically, which is not conducive to the real-time requirements of continuous tracking.

[0008] In summary, the existing technology lacks a technical solution that directly addresses the water quality detection results of multiple nodes in the Internet of Things, uses the feasible emission time domain of candidate pollution sources as a unified state variable, and simultaneously utilizes the positive response detection results and the effective non-response detection results for intersection and difference constraints, respectively. Summary of the Invention

[0009] 1. Technical problem to be solved:

[0010] To address the problems existing in the prior art, this invention provides a water pollution source tracking and analysis method and system based on the Internet of Things. It can achieve deterministic hard constraint screening of multi-node evidence, avoid invalid monitoring data leading to the incorrect exclusion of real pollution sources, and significantly reduce the global recalculation cost brought by new data, thereby balancing the accuracy, robustness and online real-time performance of pollution source tracking.

[0011] 2. Technical Solution:

[0012] To solve the above problems, the present invention adopts the following technical solution.

[0013] This invention constructs an Internet of Things (IoT)-based water pollution source tracking and analysis system, applicable to rivers, river networks, drainage pipe networks, or other water systems with clearly defined pollution propagation topologies and flow directions. Multiple IoT water quality monitoring nodes continuously collect time-series data on target pollutant concentrations and simultaneously record monitoring time, node location, equipment status, and data quality status. This invention does not limit the specific sensor structure, communication protocol, or cloud platform implementation.

[0014] The system generates a pollution response time window based on multi-node detection results of confirmed water pollution events. It then combines this with water system topology, flow direction, and potential discharge locations to generate a set of candidate pollution sources and their propagation paths to each monitoring node. The system determines the pollution propagation time range based on path length and hydraulic parameters. For each candidate pollution source, the system establishes and updates a currently feasible discharge time domain, which represents the set of times during which the candidate pollution source could still potentially discharge pollution, given the current effective water quality monitoring evidence.

[0015] For monitoring nodes that detect a response to a target pollutant, a positive emission time constraint is formed for the feasible emission time domain of the candidate pollutant source, and the system performs an intersection update accordingly. For monitoring nodes that do not detect a corresponding pollution response, a contradictory emission time subdomain is formed for the candidate pollutant source only if the node simultaneously meets four conditions: propagation reachability, monitoring time coverage, theoretical detectability, and valid detection data. The system performs a difference update accordingly. When the feasible emission time domain of a candidate pollutant source becomes empty, the candidate pollutant source becomes invalid. For the remaining candidate pollutant sources, a concentration propagation allowable envelope verification is further performed. When new IoT water quality monitoring data arrives, the system only updates the time domain for candidate pollutant sources that have a propagation association with the new evidence, and finally outputs the pollutant source location, feasible emission time range, and pollution propagation path.

[0016] Specifically, the method of the present invention includes the following steps:

[0017] S1. Acquire water quality monitoring data from multiple IoT nodes and generate pollution response information.

[0018] For confirmed water pollution events, obtain the time sequence of target pollutant concentrations, sampling time, node location, equipment operating status, and data quality status of each IoT water quality monitoring node in the target water area. Let the first... The water quality monitoring nodes are ,node At any moment The measured concentration of the target pollutant is recorded as , representing a node At any moment The measured pollutant concentration, with the unit determined according to the target pollutant, is usually expressed in mg / L.

[0019] Based on confirmed pollution events at nodes The actual water quality response period is used to determine the pollution response time window. ,in The pollution response start time, This is the end time of the pollution response. If necessary, it can be accessed from... Auxiliary features such as peak concentration and peak time are extracted, but are not used as necessary constraint variables in this invention.

[0020] To differentiate between monitoring periods and data quality, nodes are set. The actual monitoring time set is At the same time, the valid status of the detection data is defined as .in, Represents a node At any moment The detection data meets the preset quality requirements and can be used for pollution source tracing and analysis; This indicates that the data at that moment does not meet the usage requirements due to equipment failure, communication failure, calibration abnormality, obvious out-of-bounds error, maintaining a constant value for a preset duration, or other reasons. Used to describe whether the node is in the monitoring period. These two are used to describe whether the corresponding monitoring data is valid; they are used for subsequent time coverage and data validity judgment, respectively.

[0021] S2. Generate a set of candidate pollution sources and establish pollution propagation paths.

[0022] Based on the location of the monitoring node or pollution section where the pollution event occurred, the water system topology, the direction of water flow, and the location of potential discharge outlets or pollution sources, potential pollution sources with hydraulic connectivity are retrieved from the location of the pollution event in the reverse direction of the pollution propagation direction, forming an initial set of candidate pollution sources:

[0023] ,

[0024] in, This is the initial set of candidate pollution sources; For the first One candidate pollution source, Indexing candidate pollution sources; This represents the initial number of candidate pollution sources.

[0025] For candidate pollution sources and monitoring nodes The directed pollution propagation path between the two can be determined based on the topology of the river network or pipeline network:

[0026] ,

[0027] in, Candidate pollution sources To monitoring nodes The pollution transmission pathway; For the first on this path Each river section or pipe section; For path segment index; For path The number of river sections or pipe segments included. If and If there is no downstream reachable path between nodes, then the node... No candidate pollution source This creates subsequent time constraints.

[0028] S3. Calculate the pollution propagation time range from candidate pollution sources to relevant monitoring nodes.

[0029] For the transmission path Let the first The length of each propagation segment is The unit is length; the effective velocity range of this propagation segment is... ,in and The first The effective minimum and effective maximum flow velocities within the propagation segment length are expressed in units of length / time.

[0030] When using segmented travel time calculations, candidate pollution sources To monitoring nodes The shortest propagation time is:

[0031] ;

[0032] The longest propagation time is:

[0033] ;

[0034] This leads to the candidate pollution sources. To monitoring nodes The time range of pollution transmission:

[0035] ;

[0036] in, Indicates that the pollutants are from spread to The possible propagation time is used. A range of propagation time is used instead of a single value to accommodate changes in flow velocity, hydraulic fluctuations, and parameter errors. If an existing hydrodynamic model, historical tracer data, or measured travel time can provide a propagation time range, the corresponding result can be used directly.

[0037] S4. Establishing feasible emission time domains for candidate pollution sources using positive pollution responses.

[0038] For monitoring nodes that detect the target pollutant response Its pollution response time window is Candidate pollution sources To node The time range of its spread is This allows for the reverse determination of nodes. For candidate pollution sources The resulting emission time positive constraint domain:

[0039] ;

[0040] in, For monitoring nodes For candidate pollution sources The resulting emission time is a positive constraint domain. Indicates candidate pollution sources To explain the node The actual pollution response must fall within a certain range for the emission time.

[0041] Based on the pollution response time window and the maximum possible propagation time, candidate pollution sources were identified. Set the initial emission time search range:

[0042] ;

[0043] in, Candidate pollution sources The initial feasible emission time domain; and These are the earliest and latest emission search times, preset or determined based on pollution events.

[0044] set up In order to be related to candidate pollution sources A set of monitoring node indices that demonstrates a transmission relationship, has valid detection data, and has detected a response to the target pollutant constitutes a candidate pollution source. The current feasible emission time domain is:

[0045] ;

[0046] in, Candidate pollution sources Current feasible emission time domain; This represents the intersection operation of sets; Indicates to Find the intersection of the emission time positive constraint domains formed by all valid positive response monitoring nodes.

[0047] like:

[0048] ;

[0049] in, The empty set indicates that no emission time can simultaneously explain all the current valid positive response detection results, thus identifying candidate pollution sources. The candidate pollution sources are invalidated and removed from the set of candidate pollution sources. The core of this step is to use the existence of a common feasible emission time domain for candidate pollution sources as a deterministic criterion, rather than scoring or ranking based solely on time differences.

[0050] S5. Constructing a contradictory emission time subdomain using valid non-responding results.

[0051] For monitoring nodes where no corresponding pollution response was detected ,in To monitor the node index, instead of directly excluding candidate pollution sources based on their "not responding" status, we analyze the candidate pollution sources... Current feasible emission time domain Each candidate emission time Determine whether the non-response result meets the criteria to serve as valid negative evidence.

[0052] First, define the propagation reachability state. When candidate pollution sources To monitoring nodes When there is a pollution transmission path under the current effective flow direction, ;otherwise For any Pollutant theory reaches node The time range is:

[0053] ;

[0054] in, Candidate pollution sources At the time of emission The following corresponds to the monitoring node The theoretical arrival time range; and They are respectively to The shortest and longest propagation times.

[0055] Based on candidate pollution source hypotheses and the current feasible emission time And pollutant propagation conditions, calculation nodes Concentration propagation within the theoretical arrival time range allows for envelope:

[0056] ;

[0057] in, Candidate pollution sources At the time of emission Next pair of nodes The theoretical concentration propagation allows for an envelope; and These are the lower and upper bounds of the theoretical concentration, respectively, after considering the uncertainties in dilution, attenuation, diffusion, and propagation parameters.

[0058] For candidate emission times Further define four decision states: In the time coverage state, when the theoretical arrival time range is covered by nodes Actual monitoring time set Set the value to 1 if coverage is required according to the preset coverage requirements, otherwise set the value to 0. For the theoretically detectable state, a value of 1 is assigned when the theoretical concentration propagation allowable envelope is within the effective monitoring period and meets the preset effective detection conditions for the target pollutant; otherwise, a value of 0 is assigned. The effective detection conditions are determined based on the monitoring node. The determination is based on a comprehensive consideration of reliable detection capabilities, water quality background fluctuations, and target pollution response requirements. For data to be considered valid, detection data that meets the preset ratio or duration requirements within the theoretical time range has the following characteristics: The value is 1 if the condition is met, otherwise it is 0. This represents the actual pollution response state, when the node In Within the corresponding theoretical arrival time range, the value is 1 when the target pollutant response is formed, and 0 when no corresponding response is formed.

[0059] All candidate emission moments that simultaneously satisfy the criteria of propagation reachability, time coverage, theoretical detectability, valid data, and no actual pollution response are constituted as contradictory emission time subdomains:

[0060] ,

[0061] in, For monitoring nodes For candidate pollution sources The resulting contradictory emission time subdomain; vertical line This indicates that the following condition is met.

[0062] Subsequently, a difference set update is performed on the current feasible emission time domain of the candidate pollution sources:

[0063] ,

[0064] in, This represents the set difference operation, the updated... All nodes were deleted. Candidate emission times that contradict valid non-responding detection results. If updated Then candidate pollution sources Failure. Therefore, monitoring results showing no pollution response can only be used as a basis for excluding a pollution source if the monitoring node should theoretically have detection capabilities and the detection data is valid.

[0065] S6. Conduct a fine screening of the remaining candidate pollution sources based on the consistency of concentration propagation.

[0066] After S4 and S5 complete the hard constraints in the time domain, only those that still satisfy... Candidate pollution sources Perform a concentration propagation consistency screening. This step does not require re-performing global parameter inversion on all initial candidate sources. Instead, within the current feasible emission time domain of the candidate pollution sources, it calculates the candidate pollution sources based on flow dilution, pollutant attenuation, dispersion, the possible emission levels of the candidate sources, and the allowable range of relevant propagation parameters. At monitoring nodes The corresponding candidate emission time Concentration propagation allows for envelope propagation:

[0067] ,

[0068] in, Candidate pollution sources At the time of emission Below the monitoring node The resulting concentration propagation allows for an envelope; and These are the lower and upper bounds of the theoretical concentration, respectively. .

[0069] Consider nodes Water quality testing error The measured concentration is expressed as the allowable detection range:

[0070] ,

[0071] in, For nodes At any moment The allowable range of measured concentrations; This is the measured concentration; This represents the detection error or permissible measurement uncertainty for the target pollutant at this node, with units consistent with concentration.

[0072] Define concentration consistency status When candidate emission time The corresponding theoretical concentration propagation allows the envelope Compared with the allowable range of measured concentration When the preset consistency requirements are met within the preset critical response time range, Otherwise, take 0. This forms a monitoring node. For candidate pollution sources Concentration consistency time subdomain:

[0073] ,

[0074] in, For the node The set of candidate emission times that can still be interpreted based on the concentration detection results; The concentration is consistent.

[0075] set up To form the effective monitoring node index set for participation in the concentration consistency screening, the current feasible emission time domain of candidate pollution sources can be further updated as follows:

[0076] ,

[0077] in, This is the set of monitoring node indices used for concentration consistency screening; the meanings of other symbols are the same as before. If sufficient parameter information is lacking to construct a reliable allowable envelope for concentration propagation, this step can only serve as an auxiliary consistency check, and the candidate pollution source should not be directly determined to be invalid based on the concentration consistency check results.

[0078] S7. Utilize newly added IoT water quality monitoring data for localized incremental tracking

[0079] When IoT water quality monitoring nodes generate new and valid detection data, they first identify candidate pollution sources that are associated with the newly added monitoring evidence based on the water system topology, pollution propagation path, and theoretical arrival time. They then update the current feasible emission time domain only for these candidate pollution sources and do not re-perform global source tracing calculations for all candidate pollution sources.

[0080] set up For incremental update rounds, For the first The set of candidate pollution sources before the next update. Candidate pollution sources In the The feasible emission time domain before the update. If the new detection result is a positive pollution response, then the new emission time positive constraint domain is generated according to method S4. and execute:

[0081] ,

[0082] in, For the first The next updated feasible emission time domain; To provide additional positive response detection evidence for candidate pollution sources The resulting emission time is a positive constraint domain.

[0083] If the newly added detection result is a valid non-response result, and the result meets the conditions described in step S5, then a new contradictory emission time subdomain is generated according to the method in S5. and execute:

[0084] ,

[0085] in, To add valid non-response detection evidence for candidate pollution sources The resulting contradictory emission time subdomain.

[0086] After completing the time-domain update of the relevant candidate pollution sources, the candidate pollution source set is updated as follows:

[0087] ,

[0088] in, For the first The updated set of candidate pollution sources; This is the set of candidate pollution sources before the update; Candidate pollution sources Updated feasible emission time domain. Candidate pollution sources that do not have a transmission association with newly added monitoring evidence. It remains unchanged.

[0089] When there is only one candidate pollution source, or only one candidate pollution source simultaneously meets the time domain constraints and concentration propagation consistency requirements, the pollution source location, feasible emission time range, and pollution propagation path are output. If multiple candidate pollution sources still exist, incremental updates are performed using subsequent IoT water quality monitoring results; when new monitoring data causes changes in the allowable range of measured concentrations at affected monitoring nodes, S6 is re-executed only for the remaining affected candidate pollution sources.

[0090] This invention also provides an Internet of Things-based water pollution source tracking and analysis system, comprising:

[0091] The data acquisition module is used to acquire the target pollutant concentration time series, sampling time, node location, equipment operation status and data quality status of each IoT water quality monitoring node in the water area involved in the confirmed water pollution incident, and to determine the pollution response time window of each monitoring node based on the actual water quality response time period on each monitoring node.

[0092] The candidate pollution source generation module is used to search for potential pollution sources with hydraulic connectivity in the reverse direction of the pollution propagation direction based on the location of the monitoring node or pollution section where the pollution event is located, the water system topology, the water flow direction, and the location of potential sewage outlets or pollution sources, generate an initial set of candidate pollution sources, and establish a directed pollution propagation path from each candidate pollution source to each monitoring node.

[0093] The propagation time calculation module is used to calculate the pollution propagation time range from each candidate pollution source to each monitoring node based on the path length and hydraulic parameters of each propagation path.

[0094] The positive constraint update module is used to determine the emission time positive constraint domain formed by the monitoring node for the candidate pollution source based on the pollution response time window of the monitoring node and the pollution propagation time range from the candidate pollution source to the monitoring node, and update the intersection of the initial feasible emission time domain of the candidate pollution source and the emission time positive constraint domain formed by all effective positive response monitoring nodes.

[0095] The negative constraint update module is used to determine whether each candidate emission moment in the current feasible emission time domain of the candidate pollution source simultaneously meets the conditions of propagation reachability, monitoring time coverage, theoretical detectability, and valid detection data for monitoring nodes that have not detected corresponding pollution responses. The candidate emission moments that simultaneously meet all the above conditions constitute the contradictory emission time subdomain, and the difference set update is performed on the current feasible emission time domain of the candidate pollution source.

[0096] The concentration consistency verification module is used to calculate the allowable envelope of concentration propagation within the current feasible emission time domain for the remaining candidate pollution sources after the positive constraint update and negative constraint update are completed, and to perform consistency verification with the measured allowable concentration interval of the monitoring node. The current feasible emission time domain of the candidate pollution sources is updated according to the verification results.

[0097] The incremental update module is used to identify candidate pollution sources that have a propagation association with the newly added IoT water quality detection data when new detection data arrives, and only update the current feasible emission time domain of the candidate pollution sources that have a propagation association.

[0098] The pollution source output module is used to output the location, feasible emission time range, and pollution propagation path of a candidate pollution source when only one or only one candidate pollution source meets the time domain constraints and concentration propagation consistency requirements.

[0099] 3. Beneficial effects:

[0100] Compared with the prior art, the technical solution provided by this invention has the following advantages:

[0101] (1) This invention uses the "feasible emission time domain" of candidate pollution sources as the core state variable, combines the pollution response time window of each IoT monitoring node with the pollution propagation time range from the candidate pollution source to the monitoring node, and forms an emission time positive constraint domain through set intersection operation, thus unifying multiple valid positive response detection evidence into a deterministic hard constraint on the emission time of candidate pollution sources. This method differs from the processing mode in the prior art that only scores or ranks based on time difference, and can directly determine whether there is an emission time of candidate pollution sources that can simultaneously explain all valid positive response detection results, thereby achieving effective screening and rapid convergence of candidate pollution sources.

[0102] (2) For monitoring nodes where no pollution response was detected, this invention does not directly use them as "uncontaminated nodes" to exclude pollution sources. Instead, for each candidate emission time of a candidate pollution source, it comprehensively judges four conditions: propagation reachability, monitoring time coverage, theoretical detectability, and detection data validity. Only when all the above conditions are met and no pollution response has actually occurred, is the candidate emission time included in the contradictory emission time subdomain, and a difference set update is performed. This mechanism fully considers the actual situation of no response due to reasons such as incomplete monitoring period coverage, insufficient detection capability, or invalid data, effectively avoiding the risk of incorrectly excluding real pollution sources, and significantly improving the reliability and robustness of pollution source tracing results.

[0103] (3) When new IoT water quality monitoring data arrives, this invention only updates the feasible emission time domain of candidate pollution sources that have a propagation association with the new monitoring data, while keeping the feasible emission time domain unchanged for candidate pollution sources that do not have a propagation association with the new evidence. This local incremental update mechanism avoids re-performing global propagation calculations or parameter inversions for all candidate pollution sources every time new data arrives, effectively controlling the growth of computational load. This enables the invention to adapt to the continuous and high-frequency arrival characteristics of IoT water quality monitoring data, meeting the real-time and online processing requirements of continuous pollution source tracking.

[0104] (4) After completing the positive and negative constraints update in the time domain, this invention further calculates the theoretical concentration propagation allowable envelope for the remaining candidate pollution sources within their current feasible emission time domain, and performs consistency verification with the measured concentration allowable interval of the monitoring nodes. This process combines temporal feasibility with physical feasibility in terms of concentration, and uses measured concentration data to perform secondary verification of the emission time of candidate pollution sources, further narrowing down the range of candidate pollution sources and improving the accuracy and reliability of the final output pollution source location and emission time range.

[0105] (5) When calculating the pollution propagation time from candidate pollution sources to monitoring nodes, this invention uses a range of propagation time (from the shortest to the longest propagation time) instead of a fixed single value, which can effectively accommodate the uncertainty of propagation time caused by changes in flow velocity, hydraulic fluctuations, and parameter errors in actual water systems. Based on this, the positive constraint domain and contradictory emission time subdomain constructed are both based on interval operations, which allows this invention to keep the real pollution sources and their actual emission times within a feasible range even when the actual hydraulic conditions deviate from the calibration conditions, thus exhibiting good engineering applicability and robustness.

[0106] It should be noted that the structures not described in this invention are not related to the design points and improvement directions of this invention, and are the same as or can be implemented using existing technologies, so they will not be elaborated here. Attached Figure Description

[0107] Figure 1 This is a flowchart of the overall scheme for tracing and analyzing water pollution sources in this invention;

[0108] Figure 2 This is a flowchart of the water pollution source tracing and analysis method of the present invention;

[0109] Figure 3 This is a flowchart of the update and failure determination process for the positive and negative constraints of the feasible emission time domain of candidate pollution sources in this invention.

[0110] Figure 4 This is a schematic diagram of the local topology of the pipeline network and the layout of candidate pollution sources and monitoring nodes in an embodiment of the present invention. Detailed Implementation

[0111] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0112] This invention constructs an Internet of Things (IoT)-based water pollution source tracking and analysis system, applicable to rivers, river networks, drainage pipe networks, or other water systems with clearly defined pollution propagation topologies and flow directions. Multiple IoT water quality monitoring nodes continuously collect time-series data on target pollutant concentrations and simultaneously record monitoring time, node location, equipment status, and data quality status. This invention does not limit the specific sensor structure, communication protocol, or cloud platform implementation.

[0113] The system generates a pollution response time window based on multi-node detection results of confirmed water pollution events. It then combines this with water system topology, flow direction, and potential discharge locations to generate a set of candidate pollution sources and their propagation paths to each monitoring node. The system determines the pollution propagation time range based on path length and hydraulic parameters. For each candidate pollution source, the system establishes and updates a currently feasible discharge time domain, which represents the set of times during which the candidate pollution source could still potentially discharge pollution, given the current effective water quality monitoring evidence.

[0114] For monitoring nodes that detect a response to a target pollutant, a positive emission time constraint is formed for the feasible emission time domain of the candidate pollutant source, and the system performs an intersection update accordingly. For monitoring nodes that do not detect a corresponding pollution response, a contradictory emission time subdomain is formed for the candidate pollutant source only if the node simultaneously meets four conditions: propagation reachability, monitoring time coverage, theoretical detectability, and valid detection data. The system performs a difference update accordingly. When the feasible emission time domain of a candidate pollutant source becomes empty, the candidate pollutant source becomes invalid. For the remaining candidate pollutant sources, a concentration propagation allowable envelope verification is further performed. When new IoT water quality monitoring data arrives, the system only updates the time domain for candidate pollutant sources that have a propagation association with the new evidence, and finally outputs the pollutant source location, feasible emission time range, and pollution propagation path.

[0115] The propagation correlation is as follows: there is a directed pollution propagation path from the candidate pollution source to the monitoring node where the newly added detection data is located, and the theoretical arrival time range calculated based on the current propagation time range intersects with the detection time of the newly added detection data.

[0116] Specifically, the method of the present invention includes the following steps:

[0117] S1. Acquire water quality monitoring data from multiple IoT nodes and generate pollution response information.

[0118] For confirmed water pollution events, obtain the time sequence of target pollutant concentrations, sampling time, node location, equipment operating status, and data quality status of each IoT water quality monitoring node in the target water area. Let the first... The water quality monitoring nodes are ,node At any moment The measured concentration of the target pollutant is recorded as , representing a node At any moment The measured pollutant concentration, with the unit determined according to the target pollutant, is usually expressed in mg / L.

[0119] Based on confirmed pollution events at nodes The actual water quality response period is used to determine the pollution response time window. ,in The pollution response start time, This is the end time of the pollution response. If necessary, it can be accessed from... Auxiliary features such as peak concentration and peak time are extracted, but are not used as necessary constraint variables in this invention.

[0120] To differentiate between monitoring periods and data quality, nodes are set. The actual monitoring time set is At the same time, the valid status of the detection data is defined as .in, Represents a node At any moment The detection data meets the preset quality requirements and can be used for pollution source tracing and analysis; This indicates that the data at that moment does not meet the usage requirements due to equipment failure, communication failure, calibration abnormality, obvious out-of-bounds error, maintaining a constant value for a preset duration, or other reasons. Used to describe whether the node is in the monitoring period. These two are used to describe whether the corresponding monitoring data is valid; they are used for subsequent time coverage and data validity judgment, respectively.

[0121] In step S1, the valid status of the detection data includes: when the monitoring node The test data is marked as valid when it meets the preset quality requirements, and as invalid when the test data fails to meet the usage requirements due to equipment failure, communication failure, calibration abnormality, data exceeding the limit, or continuously maintaining a constant value within a preset time period.

[0122] The above valid state can be denoted as An invalid state can be denoted as The preset quality requirements can be determined based on the monitoring nodes. The detection capability, communication status, calibration status, measurement range, and data fluctuation characteristics are preset. Equipment malfunctions, communication failures, calibration anomalies, data exceeding limits, and maintaining a constant value for a preset duration are all considered invalid states that do not meet the usage requirements. Data in invalid states will not be used in subsequent emission time positive constraints, contradictory emission time subdomain construction, and concentration consistency verification.

[0123] S2. Generate a set of candidate pollution sources and establish pollution propagation pathways.

[0124] Based on the location of the monitoring node or pollution section where the pollution event occurred, the water system topology, the direction of water flow, and the location of potential discharge outlets or pollution sources, potential pollution sources with hydraulic connectivity are retrieved from the location of the pollution event in the reverse direction of the pollution propagation direction, forming an initial set of candidate pollution sources:

[0125] ,

[0126] in, This is the initial set of candidate pollution sources; For the first One candidate pollution source, Indexing candidate pollution sources; This represents the initial number of candidate pollution sources.

[0127] For candidate pollution sources and monitoring nodes The directed pollution propagation path between the two can be determined based on the topology of the river network or pipeline network:

[0128] ,

[0129] in, Candidate pollution sources To monitoring nodes The pollution transmission pathway; For the first on this path Each river section or pipe section; For path segment index; For path The number of river sections or pipe segments included. If and If there is no downstream reachable path between nodes, then the node... No candidate pollution source This creates subsequent time constraints.

[0130] S3. Calculate the pollution propagation time range from candidate pollution sources to relevant monitoring nodes.

[0131] For the transmission path Let the first The length of each propagation segment is The unit is length; the effective velocity range of this propagation segment is... ,in and The first The effective minimum and effective maximum flow velocities within the propagation segment length are expressed in units of length / time.

[0132] When using segmented travel time calculations, candidate pollution sources To monitoring nodes The shortest propagation time is:

[0133] ;

[0134] The longest propagation time is:

[0135] ;

[0136] This leads to the candidate pollution sources. To monitoring nodes The time range of pollution transmission:

[0137] ;

[0138] in, Indicates that the pollutants are from spread to The possible propagation time is used. A range of propagation time is used instead of a single value to accommodate changes in flow velocity, hydraulic fluctuations, and parameter errors. If an existing hydrodynamic model, historical tracer data, or measured travel time can provide a propagation time range, the corresponding result can be used directly.

[0139] S4. Establishing feasible emission time domains for candidate pollution sources using positive pollution responses.

[0140] For monitoring nodes that detect the target pollutant response Its pollution response time window is Candidate pollution sources To node The time range of its spread is This allows for the reverse determination of nodes. For candidate pollution sources The resulting emission time positive constraint domain:

[0141] ;

[0142] in, For monitoring nodes For candidate pollution sources The resulting emission time is a positive constraint domain. Indicates candidate pollution sources To explain the node The actual pollution response must fall within a certain range for the emission time.

[0143] Based on the pollution response time window and the maximum possible propagation time, candidate pollution sources were identified. Set the initial emission time search range:

[0144] ;

[0145] in, Candidate pollution sources The initial feasible emission time domain; and These are the earliest and latest emission search times, preset or determined based on pollution events.

[0146] set up In order to be related to candidate pollution sources A set of monitoring node indices that demonstrates a transmission relationship, has valid detection data, and has detected a response to the target pollutant constitutes a candidate pollution source. The current feasible emission time domain is:

[0147] ;

[0148] in, Candidate pollution sources Current feasible emission time domain; This represents the intersection operation of sets; Indicates to Find the intersection of the emission time positive constraint domains formed by all valid positive response monitoring nodes.

[0149] like:

[0150] ;

[0151] in, The empty set indicates that no emission time can simultaneously explain all the current valid positive response detection results, thus identifying candidate pollution sources. The candidate pollution sources are invalidated and removed from the set of candidate pollution sources. The core of this step is to use the existence of a common feasible emission time domain for candidate pollution sources as a deterministic criterion, rather than scoring or ranking based solely on time differences.

[0152] S5. Constructing a contradictory emission time subdomain using valid non-responding results.

[0153] For monitoring nodes where no corresponding pollution response was detected ,in To monitor the node index, instead of directly excluding candidate pollution sources based on their "not responding" status, we analyze the candidate pollution sources... Current feasible emission time domain Each candidate emission time Determine whether the non-response result meets the criteria to serve as valid negative evidence.

[0154] In step S5, the method for determining the reachability of propagation is: candidate pollution source To monitoring nodes When a pollution propagation path exists under the current effective flow direction, the propagation is considered reachable; the method for determining the monitoring time coverage is: the theoretical arrival time of the pollutant at the monitoring node. The time range is defined by the monitoring node Actual monitoring time set When coverage is achieved according to the preset coverage requirements, the monitoring time coverage is satisfied; the method for determining theoretically detectable sources is: candidate pollution sources. At candidate emission time Next monitoring node When the theoretical concentration propagation envelope formed meets the preset effective detection conditions of the target pollutant, it is theoretically detectable; the method for judging the validity of the detection data is: when the detection data that meets the preset proportion or duration requirements within the theoretical arrival time range is in a valid state, the detection data is valid.

[0155] Specifically, the reachability of propagation can be determined based on the water system topology and the current effective flow direction. to If there is no downstream reachable path, then Monitoring time coverage can be based on the theoretical arrival time range. With the actual monitoring time set The intersection or coverage ratio is used to determine the theoretical detectability; the theoretical detectability can be determined based on the relationship between the theoretical concentration propagation allowable envelope and the target pollutant detection limit, background fluctuations, and response threshold; the validity of the detection data can be determined based on the theoretical arrival time range. The data ratio or continuous effective duration is used to determine the coverage requirements, effective detection conditions, and ratio or duration requirements mentioned above. These can all be preset based on the detection capabilities of the monitoring nodes, the characteristics of the target pollutants, and fluctuations in the water quality background.

[0156] First, define the propagation reachability state. When candidate pollution sources To monitoring nodes When there is a pollution transmission path under the current effective flow direction, ;otherwise For any Pollutant theory reaches node The time range is:

[0157] ;

[0158] in, Candidate pollution sources At the time of emission The following corresponds to the monitoring node The theoretical arrival time range; and They are respectively to The shortest and longest propagation times.

[0159] Based on candidate pollution source hypotheses and the current feasible emission time And pollutant propagation conditions, calculation nodes Concentration propagation within the theoretical arrival time range allows for envelope:

[0160] ;

[0161] in, Candidate pollution sources At the time of emission Next pair of nodes The theoretical concentration propagation allows for an envelope; and These are the lower and upper bounds of the theoretical concentration, respectively, after considering the uncertainties in dilution, attenuation, diffusion, and propagation parameters.

[0162] For candidate emission times Further define four decision states: In the time coverage state, when the theoretical arrival time range is covered by nodes Actual monitoring time set Set the value to 1 if coverage is required according to the preset coverage requirements, otherwise set the value to 0. For the theoretically detectable state, a value of 1 is assigned when the theoretical concentration propagation allowable envelope is within the effective monitoring period and meets the preset effective detection conditions for the target pollutant; otherwise, a value of 0 is assigned. The effective detection conditions are determined based on the monitoring node. The determination is based on a comprehensive consideration of reliable detection capabilities, water quality background fluctuations, and target pollution response requirements. For data to be considered valid, detection data that meets the preset ratio or duration requirements within the theoretical time range has the following characteristics: The value is 1 if the condition is met, otherwise it is 0. This represents the actual pollution response state, when the node In Within the corresponding theoretical arrival time range, the value is 1 when the target pollutant response is formed, and 0 when no corresponding response is formed.

[0163] All candidate emission moments that simultaneously satisfy the criteria of propagation reachability, time coverage, theoretical detectability, valid data, and no actual pollution response are constituted as contradictory emission time subdomains:

[0164] ,

[0165] in, For monitoring nodes For candidate pollution sources The resulting contradictory emission time subdomain; vertical line This indicates that the following condition is met.

[0166] Subsequently, a difference set update is performed on the current feasible emission time domain of the candidate pollution sources:

[0167] ,

[0168] in, This represents the set difference operation, the updated... All nodes were deleted. Candidate emission moments that contradict valid non-response detection results. This indicates an assignment operation, where the result of the operation on the right is assigned to the variable on the left; if after the update... Then candidate pollution sources Failure. Therefore, monitoring results showing no pollution response can only be used as a basis for excluding a pollution source if the monitoring node should theoretically have detection capabilities and the detection data is valid.

[0169] S6. Conduct a fine screening of the remaining candidate pollution sources based on the consistency of concentration propagation.

[0170] After S4 and S5 complete the hard constraints in the time domain, only those that still satisfy... Candidate pollution sources Perform a concentration propagation consistency screening. This step does not require re-performing global parameter inversion on all initial candidate sources. Instead, within the current feasible emission time domain of the candidate pollution sources, it calculates the candidate pollution sources based on flow dilution, pollutant attenuation, dispersion, the possible emission levels of the candidate sources, and the allowable range of relevant propagation parameters. At monitoring nodes The corresponding candidate emission time Concentration propagation allows for envelope propagation:

[0171] ,

[0172] in, Candidate pollution sources At the time of emission Below the monitoring node The resulting concentration propagation allows for an envelope; and These are the lower and upper bounds of the theoretical concentration, respectively. .

[0173] Consider nodes Water quality testing error The measured concentration is expressed as the allowable detection range:

[0174] ,

[0175] in, For nodes At any moment The allowable range of measured concentrations; This is the measured concentration; This represents the detection error or permissible measurement uncertainty for the target pollutant at this node, with units consistent with concentration.

[0176] Define concentration consistency status When candidate emission time The corresponding theoretical concentration propagation allows the envelope Compared with the allowable range of measured concentration When the preset consistency requirements are met within the preset critical response time range, Otherwise, take 0. This forms a monitoring node. For candidate pollution sources Concentration consistency time subdomain:

[0177] ,

[0178] in, For the node The set of candidate emission times that can still be interpreted based on the concentration detection results; The concentration is consistent.

[0179] set up To form the effective monitoring node index set for participation in the concentration consistency screening, the current feasible emission time domain of candidate pollution sources can be further updated as follows:

[0180] ,

[0181] in, This is the set of monitoring node indices used for concentration consistency screening; the meanings of other symbols are the same as before. If sufficient parameter information is lacking to construct a reliable allowable envelope for concentration propagation, this step can only serve as an auxiliary consistency check, and the candidate pollution source should not be directly determined to be invalid based on the concentration consistency check results.

[0182] S7. Utilize newly added IoT water quality monitoring data for localized incremental tracking

[0183] When IoT water quality monitoring nodes generate new and valid detection data, they first identify candidate pollution sources that are associated with the newly added monitoring evidence based on the water system topology, pollution propagation path, and theoretical arrival time. They then update the current feasible emission time domain only for these candidate pollution sources and do not re-perform global source tracing calculations for all candidate pollution sources.

[0184] In step S7, the method for determining the candidate pollution source that has a propagation association with the newly added detection data is as follows: based on the water system topology and pollution propagation path, determine whether the theoretical arrival time range from the candidate pollution source to the monitoring node where the newly added detection data is located intersects with the detection time of the newly added detection data. If there is an intersection, it is determined that there is a propagation association.

[0185] Specifically, for candidate pollution sources The theoretical arrival time range to the monitoring node where the newly added detection data is located can be calculated based on the propagation time range obtained in step S3; if the theoretical arrival time range intersects with the detection time of the newly added detection data, then it is determined that... There is a transmission link with the newly tested data, only for Perform a local incremental update; if no intersection exists, then... Current feasible emission time domain It remains unchanged and no global recalculation is performed.

[0186] set up For incremental update rounds, For the first The set of candidate pollution sources before the next update. Candidate pollution sources In the The feasible emission time domain before the update. If the new detection result is a positive pollution response, then the new emission time positive constraint domain is generated according to method S4. and execute:

[0187] ,

[0188] in, For the first The next updated feasible emission time domain; To provide additional positive response detection evidence for candidate pollution sources The resulting emission time is a positive constraint domain.

[0189] If the newly added detection result is a valid non-response result, and the result meets the conditions described in step S5, then a new contradictory emission time subdomain is generated according to the method in S5. and execute:

[0190] ,

[0191] in, To add valid non-response detection evidence for candidate pollution sources The resulting contradictory emission time subdomain.

[0192] After completing the time-domain update of the relevant candidate pollution sources, the candidate pollution source set is updated as follows:

[0193] ,

[0194] in, For the first The updated set of candidate pollution sources; This is the set of candidate pollution sources before the update; Candidate pollution sources Updated feasible emission time domain. Candidate pollution sources that do not have a transmission association with newly added monitoring evidence. It remains unchanged.

[0195] When there is only one candidate pollution source, or only one candidate pollution source simultaneously meets the time domain constraints and concentration propagation consistency requirements, the pollution source location, feasible emission time range, and pollution propagation path are output. If multiple candidate pollution sources still exist, incremental updates are performed using subsequent IoT water quality monitoring results; when new monitoring data causes changes in the allowable range of measured concentrations at affected monitoring nodes, S6 is re-executed only for the remaining affected candidate pollution sources.

[0196] This invention also provides an Internet of Things-based water pollution source tracking and analysis system, comprising:

[0197] The data acquisition module is used to acquire the target pollutant concentration time series, sampling time, node location, equipment operation status and data quality status of each IoT water quality monitoring node in the water area involved in the confirmed water pollution incident, and to determine the pollution response time window of each monitoring node based on the actual water quality response time period on each monitoring node.

[0198] The candidate pollution source generation module is used to search for potential pollution sources with hydraulic connectivity in the reverse direction of the pollution propagation direction based on the location of the monitoring node or pollution section where the pollution event is located, the water system topology, the water flow direction, and the location of potential sewage outlets or pollution sources, generate an initial set of candidate pollution sources, and establish a directed pollution propagation path from each candidate pollution source to each monitoring node.

[0199] The propagation time calculation module is used to calculate the pollution propagation time range from each candidate pollution source to each monitoring node based on the path length and hydraulic parameters of each propagation path.

[0200] The positive constraint update module is used to determine the emission time positive constraint domain formed by the monitoring node for the candidate pollution source based on the pollution response time window of the monitoring node and the pollution propagation time range from the candidate pollution source to the monitoring node, and update the intersection of the initial feasible emission time domain of the candidate pollution source and the emission time positive constraint domain formed by all effective positive response monitoring nodes.

[0201] The negative constraint update module is used to determine whether each candidate emission moment in the current feasible emission time domain of the candidate pollution source simultaneously meets the conditions of propagation reachability, monitoring time coverage, theoretical detectability, and valid detection data for monitoring nodes that have not detected corresponding pollution responses. The candidate emission moments that simultaneously meet all the above conditions constitute the contradictory emission time subdomain, and the difference set update is performed on the current feasible emission time domain of the candidate pollution source.

[0202] The concentration consistency verification module is used to calculate the allowable envelope of concentration propagation within the current feasible emission time domain for the remaining candidate pollution sources after the positive constraint update and negative constraint update are completed, and to perform consistency verification with the measured allowable concentration interval of the monitoring node. The current feasible emission time domain of the candidate pollution sources is updated according to the verification results.

[0203] The incremental update module is used to identify candidate pollution sources that have a propagation association with the newly added IoT water quality detection data when new detection data arrives, and only update the current feasible emission time domain of the candidate pollution sources that have a propagation association.

[0204] The pollution source output module is used to output the location, feasible emission time range, and pollution propagation path of a candidate pollution source when only one or only one candidate pollution source meets the time domain constraints and concentration propagation consistency requirements.

[0205] Example 1: Pollution source tracing under normal monitoring conditions

[0206] (1) Simulation object and event settings

[0207] To verify the technical effectiveness of the above method, Example 8 of the Storm Water Management Model (SWMM) application manual published by the U.S. Environmental Protection Agency (EPA) was selected as the simulation object, and the emission and propagation process of ammonia nitrogen (NH3-N) pollution was simulated based on this network. Four candidate pollution sources were set according to the network topology, among which... For J1, For Aux3, For J13, J12 is designated as the water quality monitoring node; JI4, JI5, JI7, J7 and JI1 are designated as water quality monitoring nodes.

[0208] In the simulation, J12 was used as the actual pollution source, and a short-duration pollution pulse was applied from approximately 10:42 to 10:47. However, the location of this actual source was not provided to the tracking algorithm. To obtain the pollution propagation characteristics of different candidate sources, the water quality propagation model was run with J1, Aux3, J13, and J12 as sources, respectively. Each candidate source was used with approximately... The nominal source-end ammonia nitrogen concentration is reported in 1-minute increments.

[0209] In actual pollution incidents, both JI4 and JI5 showed significant ammonia nitrogen responses. As the effective pollution response determination criterion in this embodiment, the effective response time window for JI4 is 10:43 to 11:01, and the peak value is... The peak time was 10:48; the effective response time window for JI5 was 10:45 to 11:02, with a peak value of The peak time was 10:49. JI7, J7, and JI1 did not generate an ammonia nitrogen response under this actual pollution event.

[0210] (2) Time-domain update of candidate pollution sources

[0211] Based on the pipeline topology, all four candidate pollution sources can propagate to JI4 and JI5; therefore, the initial set of candidate pollution sources is: .

[0212] Based on the propagation response time obtained from the SWMM benchmark simulation of each candidate source, the emission time positive constraint is formed according to the actual positive response of S4 to JI4 and JI5 and the intersection is obtained, resulting in the following information.

[0213]

[0214] At this time, the four candidate pollution sources Neither of them is empty, and the source of pollution cannot be uniquely determined based solely on the two positive downstream responses.

[0215] Further elimination was performed using valid non-responding results according to S5. In the J1 candidate source model, the peak ammonia nitrogen level of JI7 was... In the J13 candidate source model, the peak ammonia nitrogen concentration of J7 is... The concentrations mentioned above are all significantly higher than the effective detection conditions of this embodiment. However, in the actual event, neither JI7 nor J7 generated a pollution response, and the corresponding monitoring periods and detection data were valid. Therefore, their non-response results are respectively... and This creates an effective negative constraint. After the difference set is updated... , Therefore and Invalid.

[0216] In the Aux3 candidate source model, the peak ammonia nitrogen concentration of JI1 is Theoretically, it possesses significant discriminative capabilities. However, in the initial tracking phase, assuming that historical water quality data for JI1 has not yet been uploaded to the analysis platform due to communication interruptions, the platform lacks the conditions to determine the actual response status and data validity for that period. Therefore, JI1 is not used to construct contradictory emission time subdomains. At this time, the candidate pollution source set is... .

[0217] (3) Concentration consistency screening and incremental update

[0218] right and Concentration propagation consistency screening was performed according to S6. Combining the actual peak times and peak concentrations of JI4 and JI5 with the SWMM baseline propagation responses of each candidate source, and under the 1-minute peak time tolerance and allowable measurement uncertainty conditions set in this embodiment, The feasible emission time domain narrows to approximately 10:18–10:19. The feasible emission time domain narrows to approximately 10:41–10:43, that is:

[0219]

[0220]

[0221] JI1 communication was subsequently restored and historical water quality data for the event period was retransmitted. After data validity checks, JI1... The actual ammonia nitrogen concentration during the corresponding theoretical pollution arrival period is Since the propagation from Aux3 to JI1 is reachable, the theoretical arrival time is covered by effective monitoring data, and the candidate source model indicates that the node should theoretically form a clear and detectable response, the newly added non-response result satisfies the negative evidence condition described in S5.

[0222] According to S7, only those with a propagation association with JI1 are considered. Perform local incremental updates:

[0223]

[0224] There is no corresponding pollution transmission relationship with JI1, and its current feasible emission time domain remains unchanged. The final result is... Therefore, the pollution source location was determined to be J12, and the feasible emission start time was approximately 10:41–10:43. This result covers the actual pollution emission start time of 10:42 set in the simulation.

[0225] This embodiment demonstrates that when multiple candidate pollution sources can explain the positive pollution response of downstream monitoring nodes, by using positive response intersection constraints, non-response difference constraints that meet the detection validity conditions, concentration propagation consistency screening, and local incremental updates of new monitoring evidence, the feasible emission time domain of candidate pollution sources that do not conform to actual water quality detection evidence can be gradually transformed into an empty set, thereby determining the location of the pollution source and its feasible emission time range.

[0226] Example 2: Pollution Source Tracing under Invalid Monitoring Data

[0227] To verify the pollution source tracing effect of the present invention when some IoT water quality monitoring data is invalid due to communication failure, this embodiment still uses the aforementioned EPA SWMM Example 8 combined sewer network.

[0228] In this embodiment, J1, Aux3, J13, and J12 are designated as four candidate pollution sources. The five water quality monitoring nodes are denoted as follows: Corresponding to JI4, Corresponding to JI5, Corresponding to JI7, Corresponding to J7, Corresponding to JI1. The initial set of candidate pollution sources is:

[0229]

[0230] In this embodiment, J1 is used as the actual pollution source, and a nominal source-end concentration of approximately [missing information] is applied from 10:42 to 10:47. The simulation generates short-duration NH3-N pollution pulses, but the actual pollution source location and actual emission time are not provided to the tracking algorithm. The simulation report step size is... .

[0231] This embodiment will As a specific implementation parameter for forming an effective response to the target pollutant and for making theoretically detectable judgments, the allowable measurement uncertainty of concentration is taken as... Peak times and Time tolerance serves only as an auxiliary feature in subsequent concentration propagation consistency screening and does not replace step S4 based on the pollution response time window. and the time range of transmission The resulting emission time is a positive constraint.

[0232] (1) Setting up scenarios with invalid monitoring data and responding to actual pollution

[0233] To simulate a communication interruption at an IoT monitoring node, a monitoring node is configured. This means that JI7 experienced a communication failure between 10:40 and 11:05.

[0234] Based on the valid status of the data detected in step S1 The definition, within this time range, satisfies:

[0235]

[0236] It should be noted that, This only indicates that the data for this time period cannot be used as valid evidence for IoT water quality monitoring; it does not change the actual water quality propagation process calculated by the SWMM model.

[0237] After performing SWMM simulation with J1 as the real pollution source, the following results were obtained:

[0238]

[0239] As can be seen from the table, in the water body In fact, a distinct NH3-N response was formed, with its peak value reaching [value missing]. However, the entire main pollution response area was within the communication interruption time range.

[0240] Therefore, from the perspective of the actual water quality process in SWMM, There is a pollution response; however, based on the detection evidence available from the IoT tracking system, due to the corresponding time period... This response is not considered valid positive response evidence, nor can the platform's failure to obtain valid response data be interpreted as actual non-response.

[0241] (2) Update the feasible emission time domain using the positive pollution response.

[0242] In this embodiment, and The detection data are valid and all show obvious pollution responses. Therefore, according to step S4, based on their pollution response time window... , The propagation time range from each candidate pollution source to the corresponding monitoring node forms the corresponding positive emission time constraint domain. .

[0243] Using the same initial feasible emission time range of 09:30 to 10:55 as in Example 1, according to:

[0244]

[0245] After performing an intersection update, we get:

[0246]

[0247] At this time, the four candidate pollution sources correspond to None of them are empty, indicating that only the following are used: and Two effective positive response monitoring nodes cannot uniquely identify the source of pollution.

[0248] (3) Constructing a contradictory emission time subdomain using valid non-response results

[0249] Further process the monitoring nodes that have not formed corresponding pollution responses according to step S5.

[0250] For candidate pollution source J13, its monitoring nodes This means that there is an effective pollution transmission route for J7, therefore, for J13 and This candidate combination of pollution sources and monitoring nodes includes:

[0251]

[0252] In the SWMM candidate source model, J13 is... The peak value of the formed NH3-N is approximately Consider what can explain and After considering the allowable range of candidate emission levels for actual pollution response, The theoretical concentration propagation allows the lower bound of the peak value of the envelope to remain approximately [missing value]. This is significantly higher than the preset effective detection conditions in this embodiment.

[0253] At the same time, the corresponding theoretical arrival time was The actual monitoring time set coverage is valid, and the detection data is effective, while in actual events... No pollution response was generated. Therefore, for the relevant candidate emission times corresponding to J13... The conditions of reachability, time coverage, theoretical detectability, valid data, and actual non-response must be met simultaneously.

[0254] thus, The candidate pollution source J13 forms a contradictory emission time subdomain .according to:

[0255]

[0256] After performing the difference update, the current feasible emission time domain corresponding to J13 becomes an empty set:

[0257]

[0258] Therefore, J13 is invalid.

[0259] For candidate pollution source Aux3, down to the monitoring node This means that JI1 has an effective pollution transmission route.

[0260] In the SWMM candidate source model, Aux3 is The peak value of the formed NH3-N is approximately In the ability to explain and Within the allowable range of candidate emission levels for actual pollution response, The theoretical concentration propagation allows the lower bound of the peak value of the envelope to remain approximately [missing value]. The levels are significantly higher than those required for effective detection.

[0261] In real events No pollution response was generated, and the corresponding theoretical arrival time period met the conditions for monitoring time coverage and valid detection data. Therefore, Aux3 forms contradictory emission time subdomains .

[0262] After performing the difference update, the current feasible emission time domain corresponding to Aux3 is empty:

[0263]

[0264] Therefore, Aux3 is disabled.

[0265] For candidate pollution source J1, its This means that JI7 has an effective transmission path, and candidate source models indicate that the pollutant should theoretically be able to spread through... A clear and detectable response is formed.

[0266] However, within the corresponding theoretical arrival period The original detection data is valid. Therefore, for the relevant candidate emission time corresponding to J1... The data validity determination status in step S5 satisfies:

[0267]

[0268] Due to the construction of contradictory emission time subdomains The conditions for propagation reachability, time coverage, theoretical detectability, valid data, and actual non-response must all be met simultaneously; therefore, the aforementioned candidate emission times cannot be included. .

[0269] Therefore, it cannot be used The communication missing state for the current feasible emission time domain corresponding to J1 Perform negative constraint deletion.

[0270] For candidate pollution source J12, its , and There is no corresponding pollution transmission path, therefore the relevant unresponsive nodes do not form a contradictory emission time subdomain for J12.

[0271] Therefore, the candidate pollution sources retained after step S5 are:

[0272]

[0273] (4) Concentration propagation consistency screening

[0274] For J1 and J12 that remain after step S5, perform concentration propagation consistency screening according to step S6.

[0275] According to step S6, the candidate pollution sources are evaluated at the candidate emission time. Below, at the monitoring node The theoretical concentration propagation allowed the envelope to be uniformly denoted as:

[0276]

[0277] The allowable range for measured concentrations at monitoring nodes is as follows:

[0278]

[0279] When peak moments are used as auxiliary features for pre-validation, J1 can simultaneously interpret... and The candidate emission start times for peak emissions are mainly concentrated between 10:41 and 10:43; while the candidate emission start times for J12 are mainly concentrated between 10:50 and 10:52.

[0280] The aforementioned peak times are only used to help narrow down the candidate emission times that require concentration envelope verification; the final times will still be determined according to... and Whether the preset consistency requirements are met, determine the concentration consistency time subdomain. .

[0281] Taking the earliest candidate emission time of J12, 10:50, as an example, within the allowable emission level and propagation parameters, it is in The theoretical concentration propagation formed at 10:50 allows for an upper bound of the envelope of approximately .

[0282] In real events The measured concentration at 10:50 was ,consider back:

[0283]

[0284] The theoretical concentration propagation allowable envelope and the measured concentration allowable range for J12 do not overlap:

[0285]

[0286] Even using an earlier candidate emission time that is more favorable to J12 cannot explain... The early pollution response has already occurred. This contradiction persists for later candidate emission times.

[0287] according to and The concentration detection results respectively form the concentration consistency time subdomain corresponding to J12. Then, according to:

[0288]

[0289] After the update, the current feasible emission time domain corresponding to J12 eventually becomes an empty set:

[0290]

[0291] Therefore, J12 is invalid.

[0292] For J1, within the candidate emission timeframe of approximately 10:41–10:43, its theoretical concentration propagation allows the envelope to simultaneously explain… and The response start, peak value, and main concentration change process.

[0293] After updating the intersection of concentration-consistent time subdomains, the current feasible emission time domain corresponding to J1 is:

[0294]

[0295] Ultimately, only J1 satisfies the time domain constraint and the concentration propagation consistency requirement.

[0296] Therefore, the system outputs the pollution source location as J1, and the feasible emission start time range is 10:41 to 10:43, which includes the actual pollution emission start time of 10:42 set in the SWMM simulation.

[0297] (5) Results of Example 2

[0298] In this embodiment, the downstream monitoring node of the actual pollution source J1 The actual water quality process actually forms a peak. There was a clear pollution response, but due to the period of the response... The corresponding test data cannot be used as valid evidence of water quality testing.

[0299] Step S5 does not simply interpret "the analysis platform did not receive an effective pollution response" as "the monitoring node did not actually respond," but further utilizes... Determine whether the non-response result meets the criteria for valid negative evidence. Since the data validity criteria are not met, The emission time subdomain does not contradict the actual pollution source J1.

[0300] Subsequently, through other valid positive responses, valid non-responses, and evidence of consistency in concentration propagation, Aux3, J13, and J12 were gradually ruled out, and J1 was ultimately identified as the actual source of pollution.

[0301] This verifies that even when some IoT water quality monitoring data is invalid due to communication failure, the present invention can avoid incorrectly excluding the real pollution source due to invalid or unresponsive results, and can still complete the pollution source tracing.

[0302] Example 3: Tracing Pollution Sources Under Fluctuating Hydraulic Conditions

[0303] To verify the pollution source tracing effect of the present invention when actual hydraulic conditions change and the pollution propagation time is uncertain, this embodiment still uses the aforementioned EPA SWMM Example 8 combined sewer system.

[0304] The candidate pollution sources remain J1, Aux3, J13, and J12. The monitoring nodes remain the same. Corresponding to JI4, Corresponding to JI5, Corresponding to JI7, Corresponding to J7, Corresponds to JI1.

[0305] This embodiment simulates an application scenario where IoT water quality monitoring nodes are relatively sparse, wherein , , and As an online water quality monitoring node that actually participates in pollution source tracing, J7 remains in the SWMM hydraulic-water quality model and corresponds to... However, this embodiment does not utilize The water quality results form positive constraints on discharge time, contradictory discharge time subdomains, or concentration consistency constraints.

[0306] In this embodiment, J13 is used as the actual pollution source, and a nominal source-end concentration of approximately [missing information] is applied from 10:42 to 10:47. Short-duration NH3-N pollution pulses.

[0307] The dry weather inflow of a real pollution event is set as the baseline operating condition. The value is multiplied to simulate situations where the actual hydraulic conditions deviate from the pre-calibrated operating conditions.

[0308] The pollution source tracing algorithm does not use this directly. The actual hydraulic conditions are doubled, while the baseline dry weather inflow is established in advance. times, Double Three permissible hydraulic conditions were used, and the SWMM pollution propagation results under these different hydraulic conditions were used to determine the propagation time range and the permissible concentration propagation envelope.

[0309] This embodiment continues to use... As a specific preset effective detection condition, the allowable measurement uncertainty of concentration is: Peak times and Time tolerance is only used as an auxiliary feature for the concentration consistency analysis in step S6.

[0310] (1) Pollution response under real hydraulic conditions

[0311] exist Under conditions of double-dry weather inflow, after operating SWMM with J13 as the actual pollution source, the monitoring nodes involved in pollution source tracing were obtained as follows:

[0312]

[0313] in:

[0314]

[0315]

[0316] In other words, J7 still calculates the water quality propagation results in the SWMM model, but in this embodiment, it is not used as evidence for online water quality detection, and therefore it is not involved in subsequent positive constraints, negative constraints, and concentration consistency screening.

[0317] (2) Time range of pollution propagation under different hydraulic conditions

[0318] In respectively times, Double Under the permissible hydraulic conditions, SWMM water quality propagation simulations were performed on J1, Aux3, J13, and J12.

[0319] Based on the pollution propagation response obtained under different hydraulic conditions, candidate pollution sources were identified. and The permissible range for the spread of pollution is as follows:

[0320]

[0321] The above results are used as the basis for step S3. and The specific implementation value.

[0322] Using a range of propagation time instead of a single fixed propagation time allows subsequent step S4 to accommodate deviations in pollution propagation time caused by changes in actual flow rate and velocity.

[0323] (3) Feasible emission time domain for the formation of positive pollution response

[0324] according to and The pollution response time window and the propagation time range corresponding to each of the above candidate pollution sources are determined according to:

[0325]

[0326] Construct positive constraint regions for emission time respectively.

[0327] Based on:

[0328]

[0329] After performing an intersection update, we get:

[0330]

[0331] The four candidate pollution sources correspond to None of them are empty, therefore, based solely on and The positive response to pollution still cannot uniquely identify the actual source of pollution.

[0332] (4) Effective non-response negative constraint

[0333] Further utilize step S5 and The valid non-response results are used to exclude candidate pollution sources. For candidate pollution source J1, its at least There is an effective transmission path, therefore the corresponding candidate pollution source-monitoring node combination satisfies:

[0334]

[0335] exist times, Double Under the permissible hydraulic conditions, J1 in The theoretical pollution responses formed were all significantly higher than the preset effective detection conditions of this embodiment.

[0336] Further consideration can explain and After the allowable emission levels of the actual pollution response, The theoretical concentration propagation allows the lower bound of the peak value of the envelope to remain approximately [missing value]. .

[0337] In real events No pollution response was generated, the corresponding theoretical arrival time was covered by the actual monitoring time set, and the detection data was valid. Therefore, the relevant candidate emission times meet the conditions of propagation reachability, time coverage, theoretical detectability, valid data, and no actual response, forming a contradictory emission time subdomain. .

[0338] Updated via difference set:

[0339]

[0340] Therefore, J1 is invalid.

[0341] For candidate pollution source Aux3, it is even more There are effective transmission pathways.

[0342] Under three permissible hydraulic conditions, Aux3 performs as follows: The resulting pollution responses all exhibit clear theoretical detectability. This allows for explanation... and Within the allowable emission levels of the actual response, The theoretical concentration propagation allows the lower bound of the peak value of the envelope to remain approximately [missing value]. .

[0343] In real events No pollution response was generated, and the corresponding monitoring time coverage and detection data were valid, therefore Aux3 forms contradictory emission time subdomains .

[0344] After the difference set update, the current feasible emission time domain corresponding to Aux3 becomes an empty set:

[0345]

[0346] Therefore, Aux3 is disabled.

[0347] For J13 and J12, there is no... and The pollution transmission path corresponds to the branch road, therefore the relevant accessibility conditions for transmission do not hold. and The unresponsive results do not form a contradictory emission time subdomain for J13 or J12.

[0348] After step S5, the candidate pollution sources are retained:

[0349]

[0350] (5) The impact of hydraulic fluctuations on the determination of discharge time

[0351] To illustrate the role of using the propagation time range, let's first take J13 as an example, and consider a single... A comparison of the propagation time ranges formed by the double reference hydraulic condition and multiple hydraulic conditions.

[0352] exist Under the reference hydraulic conditions, J13 propagates to and The peak propagation hysteresis is approximately and .

[0353] In real events, and The peak times were 10:56 and 10:57, respectively. The peak times were used as auxiliary water quality characteristics and... When time tolerance is present, The corresponding candidate emission time is approximately 10:41 to 10:43. The corresponding candidate emission times are approximately 10:39 to 10:41, with the common part concentrated only at approximately 10:41.

[0354] Therefore, if only using Using the single propagation characteristic corresponding to the baseline hydraulic state as an auxiliary judgment, the actual discharge start time of 10:42 set by SWMM cannot simultaneously satisfy the peak time relationship of the two nodes.

[0355] comprehensive times, Double After the allowable hydraulic conditions are doubled, J13 to and The pollution propagation time has a corresponding allowable range. In peak-assisted verification, the candidate emission times that can simultaneously explain the peak times of two nodes are expanded to approximately 10:39 to 10:43, thus preserving the true emission start time of 10:42.

[0356] It should be noted that the above peak time comparison is used to illustrate the changes in propagation time caused by hydraulic fluctuations and serves as an auxiliary feature for step S6; whether the candidate pollution source is ultimately retained is still determined according to the concentration propagation allowable envelope and concentration consistency time subdomain specified in the main text.

[0357] (6) Concentration propagation consistency screening

[0358] According to step S6, calculate the concentration propagation allowable envelope for J13 and J12 that are still retained after step S5.

[0359] For J13, according to times, Double The SWMM water quality time series under permissible hydraulic conditions is constructed by considering both candidate discharge levels and permissible ranges of propagation parameters. and Theoretical concentration propagation allows for envelope:

[0360]

[0361] reality The measured concentration of the double-hydraulic event is based on:

[0362]

[0363] This establishes the permissible range for measured concentrations.

[0364] This embodiment will The critical response time range is defined as 10:48, from the start of an effective pollution response to 11:14, when the response ends. Report step length, to and A time-by-time concentration consistency verification was conducted, resulting in a total of 54 "monitoring node-time" verification points.

[0365] This embodiment specifies a critical response time range of no less than [amount missing]. The verification point satisfies the requirement that the theoretical concentration propagation allowable envelope and the measured concentration allowable interval intersect, serving as a specific requirement for determining the consistency of concentration at the corresponding candidate emission time. These are specific implementation parameters for this embodiment and do not constitute a limitation on the scope of protection of this invention.

[0366] For J13, within the candidate emission time range of approximately 10:39 to 10:43, all 54 verification points met the corresponding concentration consistency conditions, meaning that 54 / 54 verification points met the requirements.

[0367] Therefore, by and Formed concentration-consistent time subdomain They share a common non-empty time range. According to:

[0368]

[0369] After the update, the current feasible emission time domain for J13 is:

[0370]

[0371] For J12, the same method is used to verify the theoretical concentration propagation allowable envelope. Near its most favorable candidate emission time, there are at most 35 verification points that satisfy the condition that the theoretical concentration propagation allowable envelope intersects with the measured concentration allowable interval, i.e.:

[0372]

[0373] Lower than the preset value in this embodiment Consistency requirements.

[0374] Therefore, after performing a concentration consistency time subdomain intersection update on J12:

[0375]

[0376] J12 is faulty.

[0377] Ultimately, only J13 simultaneously meets the positive emission time constraint, the negative effective non-response constraint, and the concentration propagation consistency requirement. Therefore, the output pollution source location is J13, and the corresponding feasible emission start time range is:

[0378]

[0379] This range includes the pollution emission start time of 10:42 preset for real-world events in SWMM.

[0380] (7) Results of Example 3

[0381] In this embodiment, the actual pollution event occurred during the baseline dry weather inflow. The hydraulic state is twice that of the pollution source tracing algorithm, while the pollution source tracing algorithm only utilizes... times, Double The three permissible hydraulic conditions establish the pollution propagation time range and concentration propagation permissible envelope.

[0382] The results show that when only the propagation characteristics corresponding to a single reference hydraulic state are used, the auxiliary discharge time judgment formed by different monitoring nodes based on the peak time cannot completely cover the actual discharge start time of 10:42.

[0383] By adopting the pollution propagation time range described in this invention, the propagation time deviation caused by changes in actual flow rate and velocity can be accommodated, so that the actual pollution source J13 and its actual emission time can continue to be kept within the current feasible range.

[0384] Subsequently, through and The valid non-responding results form contradictory emission time subdomains, excluding J1 and Aux3; then utilize... and The concentration propagation allowed for fine screening of the envelope and concentration consistency time subdomain, ultimately eliminating J12 and identifying J13 as the actual pollution source.

[0385] This verifies that even when actual hydraulic conditions deviate from the baseline operating conditions, and there are certain uncertainties in the pollution propagation time and pollution concentration process, the present invention can still correctly determine the location of the pollution source and the feasible emission time range by using the pollution propagation time range, positive emission time constraints, contradictory emission time subdomains formed by effective non-response, and concentration propagation consistency screening.

[0386] The above-described embodiments are merely one implementation of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the appended claims.

Claims

1. A method for tracing and analyzing water pollution sources based on the Internet of Things, characterized in that, Includes the following steps: S1. Obtain the target pollutant concentration time series, sampling time, node location, equipment operation status and data quality status of each IoT water quality monitoring node in the water area involved in the confirmed water pollution incident, and determine the pollution response time window of each monitoring node based on the actual water quality response time period on each monitoring node; define the actual monitoring time set of each monitoring node and the validity status of the detection data; S2. Based on the location of the monitoring node or pollution section where the pollution event occurs, the water system topology, the direction of water flow, and the location of potential sewage outlets or pollution sources, search for potential pollution sources with hydraulic connectivity in the reverse direction of the pollution propagation direction, generate an initial set of candidate pollution sources, and establish a directed pollution propagation path from each candidate pollution source to each monitoring node. S3. Based on the path length and hydraulic parameters of each propagation path, calculate the pollution propagation time range from each candidate pollution source to each monitoring node; S4. For a monitoring node that detects a response to the target pollutant, based on the pollution response time window of the monitoring node and the pollution propagation time range from the candidate pollution source to the monitoring node, the positive emission time constraint domain formed by the monitoring node for the candidate pollution source is determined in reverse. The current feasible emission time domain of the candidate pollution source is updated according to the following formula: ， Among them, for each candidate pollution source Establish the initial feasible emission time domain. ,in and These are the earliest and latest emission search times, preset or determined based on pollution events. In order to be related to candidate pollution sources A set of monitoring node indices that have a propagation relationship, valid detection data, and have detected a response to the target pollutant; if the updated feasible emission time domain is an empty set, the candidate pollution source is deemed invalid. S5. For monitoring nodes where no corresponding pollution response was detected, for each candidate emission time in the current feasible emission time domain of the candidate pollution source, determine whether it simultaneously meets the conditions of reachability, monitoring time coverage, theoretical detectability, and valid detection data. Candidate emission times that simultaneously meet all the above conditions constitute a contradictory emission time subdomain. The current feasible emission time domain of the candidate pollution source is updated according to the following formula: ， If the feasible emission time domain after the difference set update is an empty set, then the candidate pollution source is determined to be invalid. S6. For the candidate pollution sources that are still retained after steps S4 and S5, within their current feasible emission time domain, calculate the allowable concentration propagation envelope of the candidate pollution source at the corresponding candidate emission time on the monitoring node based on the information of flow dilution, pollutant attenuation, dispersion and propagation parameters. Perform consistency verification between the allowable concentration propagation envelope and the measured allowable concentration range of the monitoring node. Update the current feasible emission time domain of the candidate pollution source based on the consistency verification result. S7. When new IoT water quality monitoring data arrives, identify candidate pollution sources that are propagated in relation to the new monitoring data, and update only the current feasible emission time domain of the candidate pollution sources with propagation association: If the new monitoring data is a positive pollution response, the feasible emission time domain of the candidate pollution sources is updated according to the following formula: ， If the newly added detection data is a valid non-response result, the feasible emission time domain of the candidate pollution source shall be updated according to the following formula: ， in, For incremental update rounds, For the first Candidate pollution sources before the next update Feasible emission time domain, To provide additional positive response detection evidence for candidate pollution sources The resulting emission time is a positive constraint domain. To add valid non-response detection evidence for candidate pollution sources The contradictory emission time subdomains are formed; the feasible emission time domains of candidate pollution sources that do not form a propagation association with the newly added detection data remain unchanged; when only one candidate pollution source meets the time domain constraints and concentration propagation consistency requirements, the location, feasible emission time range, and pollution propagation path of the candidate pollution source are output.

2. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S1, the valid state of the detection data includes: when the detection data of the monitoring node meets the preset quality requirements, it is marked as valid; when the detection data fails to meet the usage requirements due to equipment failure, communication loss, calibration abnormality, data exceeding the limit, or continuously maintaining a constant value within a preset time period, it is marked as invalid.

3. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S3, the pollution propagation time range is calculated using segmented travel times: Shortest transmission time , Longest transmission time , in, For the first The length of each propagation segment and The first The effective minimum and maximum flow velocities of each propagation segment; candidate pollution sources. To monitoring nodes The time range of pollution transmission is .

4. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S4, the monitoring node The pollution response time window is Candidate pollution sources To monitoring nodes The time range of its spread is The positive constraint domain for emission time is: 。 5. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S5, The method for determining the reachability of propagation is: when there is a pollution propagation path with a current effective flow direction from the candidate pollution source to the monitoring node, the propagation is reachable; The method for determining the monitoring time coverage is as follows: when the time range in which pollutants theoretically arrive at the monitoring node is covered by the actual monitoring time set of the monitoring node according to the preset coverage requirements, the monitoring time coverage is satisfied. The method for determining theoretical detectability is as follows: when the theoretical concentration propagation allowable envelope formed by the candidate pollution source at the candidate emission time for the monitoring node meets the preset effective detection conditions of the target pollutant, it is theoretically detectable. The method for determining the validity of the detection data is as follows: when the detection data that meets the preset proportion or duration requirement within the theoretical arrival time range is in a valid state, the detection data is considered valid.

6. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S5, the contradictory emission time subdomain is represented as: ， in Candidate pollution sources Current feasible emission time domain For candidate emission times, To propagate reachability status, In time-covered state, For theoretically detectable states, The data is in a valid state. This represents the actual pollution response status.

7. The water pollution source tracking and analysis method based on the Internet of Things according to claim 1, characterized in that, In step S6, candidate pollution sources At candidate emission time Below the monitoring node The resulting concentration propagation allows for an envelope of: ， in and These are the lower and upper bounds of the theoretical concentration after considering the uncertainties of flow dilution, pollutant attenuation, dispersion, and propagation parameters, respectively. Monitoring nodes At any moment The permissible range for the measured concentration is: ， in For actual measured concentration, To detect errors or allow for measurement uncertainty; Define concentration consistency status ,when and The value is 1 if the preset consistency requirement is met within the preset critical response time range, and 0 otherwise. Monitoring nodes For candidate pollution sources The concentration consistency time subdomain is: ， Candidate pollution sources The feasible emission time domain is further updated using the following formula: ， in This is a set of monitoring node indexes that participate in the concentration consistency screening.

8. The method for tracing and analyzing water pollution sources based on the Internet of Things according to claim 1, characterized in that, In step S7, the method for determining the candidate pollution source that has a propagation association with the newly added detection data is as follows: based on the water system topology and pollution propagation path, determine whether the theoretical arrival time range from the candidate pollution source to the monitoring node where the newly added detection data is located intersects with the detection time of the newly added detection data. If there is an intersection, it is determined that there is a propagation association.

9. A water pollution source tracking and analysis system based on the Internet of Things, characterized in that, include: The data acquisition module is used to acquire the target pollutant concentration time series, sampling time, node location, equipment operation status and data quality status of each IoT water quality monitoring node in the water area involved in the confirmed water pollution incident, and to determine the pollution response time window of each monitoring node based on the actual water quality response time period on each monitoring node. The candidate pollution source generation module is used to search for potential pollution sources with hydraulic connectivity in the reverse direction of the pollution propagation direction based on the location of the monitoring node or pollution section where the pollution event is located, the water system topology, the water flow direction, and the location of potential sewage outlets or pollution sources, generate an initial set of candidate pollution sources, and establish a directed pollution propagation path from each candidate pollution source to each monitoring node. The propagation time calculation module is used to calculate the pollution propagation time range from each candidate pollution source to each monitoring node based on the path length and hydraulic parameters of each propagation path. The positive constraint update module is used to, for a monitoring node that detects a response to a target pollutant, reversely determine the emission time positive constraint domain formed by the monitoring node for the candidate pollutant source based on the pollution response time window of the monitoring node and the pollution propagation time range from the candidate pollutant source to the monitoring node. The current feasible emission time domain of the candidate pollutant source is updated according to the following formula: ， in, Candidate pollution sources The initial feasible emission time domain, In order to be related to candidate pollution sources A set of monitoring node indices that have a transmission relationship, valid detection data, and have detected a response to the target pollutant; The negative constraint update module is used to determine, for each candidate emission moment in the current feasible emission time domain of a candidate pollution source, whether it simultaneously satisfies the conditions of propagation reachability, monitoring time coverage, theoretical detectability, and valid detection data for monitoring nodes where no corresponding pollution response has been detected. Candidate emission moments that simultaneously satisfy all the above conditions constitute a contradictory emission time subdomain. The current feasible emission time domain of the candidate pollution source is updated according to the following formula: ， The concentration consistency verification module is used to calculate the allowable envelope of concentration propagation within the current feasible emission time domain for the remaining candidate pollution sources after the positive constraint update and negative constraint update are completed, and to perform consistency verification with the measured allowable concentration interval of the monitoring node. The current feasible emission time domain of the candidate pollution sources is updated according to the verification results. The incremental update module is used to identify candidate pollution sources that have a propagational correlation with newly arrived IoT water quality monitoring data when such data arrives. It only updates the current feasible emission time domain of these candidate pollution sources. If the newly arrived monitoring data indicates a positive pollution response, the feasible emission time domain of the candidate pollution source is updated using the following formula: ， If the newly added detection data is a valid non-response result, the feasible emission time domain of the candidate pollution source shall be updated according to the following formula: ， in, For incremental update rounds, For the first Candidate pollution sources before the next update Feasible emission time domain, To provide additional positive response detection evidence for candidate pollution sources The resulting emission time is a positive constraint domain. To add valid non-response detection evidence for candidate pollution sources The resulting contradictory emission time subdomain; The pollution source output module is used to output the location, feasible emission time range, and pollution propagation path of a candidate pollution source when only one or only one candidate pollution source meets the time domain constraints and concentration propagation consistency requirements.

10. The water pollution source tracking and analysis system based on the Internet of Things according to claim 9, characterized in that, In the positive constraint update module, the monitoring node The pollution response time window is Candidate pollution sources To monitoring nodes The time range of its spread is The positive constraint domain for emission time is: ， In the negative constraint update module, the contradictory emission time subdomain is represented as: 。

Images