A method and system for tracing pollution of a distributed water body

Abstract

The present application belongs to the technical field of pollution tracing, and discloses a pollution tracing method and system for distributed water bodies. A plurality of directed node pairs are constructed according to the topological structure of each partition node in the monitored water area. The time lag interval of the pollutant concentration peak value between each directed node pair is determined according to the pollutant concentration data in the monitored water area. Effective node pairs that satisfy the space-time constraint are determined from the plurality of directed node pairs, and a plurality of backtracking paths are constructed according to the effective node pairs. A plurality of simulation paths are obtained by performing particle tracking simulation on the pollutant concentration data, and the particle intersection area of the pollutant is determined according to the plurality of simulation paths. The pollution source area is determined according to the spatial intersection of the plurality of backtracking paths and the particle intersection area, and the source location is further determined. The present application constructs backtracking paths through dynamic time lag analysis, combines particle tracking with grid intersection positioning, realizes high-precision identification of pollution sources, and solves the problems of insufficient space-time dynamics and large positioning deviation of traditional methods.

Description

Technical Field

[0001] This invention discloses a distributed method and system for tracing pollution sources in water bodies, belonging to the field of pollution source tracing technology. Background Technology

[0002] The accelerated pace of global industrialization and urbanization has led to increasingly severe water pollution problems, posing a significant threat to the ecological environment and human health.

[0003] Traditional pollution source tracing mainly relies on manual sampling and laboratory analysis, which is not only inefficient but also makes it difficult to achieve large-scale, real-time dynamic pollution monitoring and source tracing. Especially in distributed water environments, the migration and diffusion of pollutants are affected by multiple factors such as water flow, topography, and weather, making the accurate location of pollution sources a huge challenge.

[0004] While existing sensor networks can achieve multi-point monitoring, water quality monitoring based on sensor networks lacks the ability for multi-node collaborative analysis and dynamic source tracing. In addition, the diffusion of pollutants in water bodies has spatiotemporal dynamics, and existing methods are unable to effectively capture the time-delay propagation characteristics of pollutants, resulting in insufficient accuracy and efficiency in pollution source tracing and location. Summary of the Invention

[0005] The purpose of this invention is to provide solutions to the technical problems in the prior art. To achieve the above objective, this invention proposes a distributed method and system for tracing pollution sources in water bodies, the specific solution of which is as follows:

[0006] A method for tracing pollution sources in distributed water bodies, comprising:

[0007] S1. Construct multiple directed node pairs based on the topological structure of each zone node within the monitored water area;

[0008] S2. Determine the time lag interval of pollutant concentration peak values ​​between each directed node pair based on the pollutant concentration data in the monitored water area.

[0009] S3. Determine the effective node pairs whose time delay intervals satisfy the spatiotemporal constraints from the multiple directed node pairs, and construct multiple backtracking paths based on the effective node pairs;

[0010] S4. Perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle convergence area of ​​the pollutants based on the multiple simulated paths;

[0011] S5. Determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence area, and determine the source location based on the pollution source area.

[0012] Preferably, S2 includes:

[0013] S21. Determine the concentration time series of each partition node based on the pollutant concentration data;

[0014] S22. Extract the peak features of the concentration time series, and determine the time lag interval of the pollutant concentration peaks between each directed node pair based on the peak features.

[0015] Preferably, determining the effective node pairs whose time delay intervals satisfy the spatiotemporal constraints from the plurality of directed node pairs includes:

[0016] Determine the difference between the theoretical time delay and the time delay interval between each pair of directed nodes;

[0017] The directed node pairs whose difference is less than the first preset threshold are denoted as the first node pairs, and the first node pairs that meet the preset conditions are denoted as the valid node pairs.

[0018] Preferably, the preset conditions include:

[0019] The first node meets the preset indicators for medium concentration mutation characteristics, continuous exceedance characteristics, and stable characteristics.

[0020] Preferably, determining the particle convergence zone of pollutants based on the multiple simulated paths includes:

[0021] S41. Determine the distance-concentration data from each downstream path node to the source tracing endpoint for each simulated path. The distance-concentration data from all downstream path nodes to the source tracing endpoint constitute a distance-concentration sequence.

[0022] S42. Perform concentration decay fitting on the pollutant concentration at the source endpoint and terminal node of each simulated path to obtain the concentration fitting sequence for each simulated path.

[0023] S43. Determine the desired path from the multiple simulated paths based on the distance-concentration sequence and the concentration fitting sequence, and determine the particle convergence zone of the pollutant in the monitored water area based on the desired path.

[0024] Preferably, S43 includes:

[0025] The node residuals for each simulated path are determined based on the distance-concentration sequence and the concentration fitting sequence.

[0026] The concentration consistency of the corresponding simulation path is determined based on the node residuals, and the simulation path with a concentration consistency greater than the second preset threshold is recorded as the desired path.

[0027] By intersecting all the expected paths, the particle convergence zone of pollutants in the monitored water area can be obtained.

[0028] Preferably, all desired paths are intersected to obtain the particle convergence zone of pollutants in the monitored water area, including:

[0029] The monitored water area is divided into multiple grid cells, and all desired paths are intersected within these grid cells.

[0030] The trajectory density value of the desired path within each grid cell is statistically analyzed, and the particle convergence zone of the pollutants is determined based on the trajectory density value.

[0031] Preferably, S5 includes:

[0032] The backtracking path and the particle convergence area are spatially superimposed. The area where the backtracking path and the particle convergence area intersect is recorded as the primary pollution source area, and the independent areas where they do not intersect are recorded as the secondary pollution source areas.

[0033] The source location is determined based on the primary pollution source area and the secondary pollution source area.

[0034] Preferably, after S5, it also includes:

[0035] Multiple target sampling areas are determined based on the confidence levels of the primary and secondary pollution source areas;

[0036] Collect pollutant collection data from the multiple target sampling areas;

[0037] The consistency score of the target sampling area is calculated based on the pollutant collection data, and the source location is updated based on the consistency score.

[0038] A distributed water pollution tracing system includes:

[0039] Node pair construction module: used to construct multiple directed node pairs based on the topology of nodes in each zone within the monitored water area;

[0040] The backtracking path construction module is used to determine the time lag interval of the pollutant concentration peak between each directed node pair based on the pollutant concentration data in the monitored water area; determine the effective node pairs whose time lag intervals satisfy the spatiotemporal constraints from the multiple directed node pairs; and construct multiple backtracking paths based on the effective node pairs.

[0041] Particle confluence zone generation module: used to perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle confluence zone of the pollutants based on the multiple simulated paths;

[0042] Source tracing and location module: used to determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence area, and to determine the source tracing and location based on the pollution source area.

[0043] Beneficial effects: This invention accurately identifies effective node pairs that meet spatiotemporal constraints by constraining the difference between the theoretical time delay and the time delay interval of directed node pairs, quantifies the migration time delay of pollutant concentration peaks, reveals the dynamic law of pollution diffusion, solves the problem of insufficient analysis of time-varying characteristics by traditional methods, and provides a time-series basis for the construction of backtracking paths;

[0044] This invention generates multiple simulated paths through particle inverse tracking simulation and uses distance-concentration sequence and concentration decay fitting techniques to screen out desired paths with high concentration consistency. Furthermore, through gridded trajectory intersection analysis, spatial localization of pollutant particle convergence zones is achieved. This overcomes the limitations of traditional source tracing methods that rely on diffusion models, significantly improving the geometric accuracy of pollution source localization.

[0045] This invention divides the pollution source area into primary and secondary pollution source areas by spatially superimposing the backtracking path and the particle convergence zone, constructing a two-level localization framework. By combining confidence assessment of the target sampling area with data consistency verification, the source tracing results are dynamically corrected, solving the localization deviation problem caused by data noise in traditional methods and achieving a closed-loop accuracy for pollution source localization. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the distributed water body source tracing method of the present invention;

[0047] Figure 2 This is a schematic diagram of a distributed water pollution tracing system in an embodiment of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention.

[0049] This application provides a distributed method for tracing the source of water pollution, such as... Figure 1 As shown, the method includes:

[0050] S1. Construct multiple directed node pairs based on the topological structure of each zone node within the monitored water area;

[0051] S2. Determine the time lag interval of pollutant concentration peak values ​​between each directed node pair based on the pollutant concentration data in the monitored water area.

[0052] S3. Determine the effective node pairs whose time delay intervals satisfy the spatiotemporal constraints from the multiple directed node pairs, and construct multiple backtracking paths based on the effective node pairs;

[0053] S4. Perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle convergence area of ​​the pollutants based on the multiple simulated paths;

[0054] S5. Determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence area, and determine the source location based on the pollution source area.

[0055] In specific implementation, the method includes: step S100, interactively acquiring the water features and associated pollution distribution features of the water area to be monitored, performing water area zoning based on the water features and associated pollution distribution features, and establishing water area zoning.

[0056] Specifically, water body characteristics and associated pollution distribution characteristics are obtained through manual input, geographic information systems, remote sensing data, or historical monitoring data. Water body characteristics refer to aquatic environmental parameters that influence pollutant migration and diffusion, including but not limited to hydrological characteristics such as flow velocity, direction, depth, and flow rate; topographic features such as riverbed morphology, channel curvature, and tributary distribution; and water quality characteristics such as pH, dissolved oxygen, turbidity, and temperature. Pollution distribution characteristics refer to the spatiotemporal distribution of pollutants in the water body, including spatial variations in pollutant concentration, such as high-concentration accumulation areas and diffusion trends; temporal variations in pollutant concentration, such as the timing of pollution peaks and periodic fluctuations; and the physicochemical properties of pollutants, such as solubility, sedimentation, and degradation rates. Associated pollution distribution characteristics refer to analyzing how pollutants are affected by the environment in conjunction with water body characteristics. For example, in areas with fast flow, pollutants may diffuse rapidly, forming long-distance migrations; in bends or still water areas, pollutants may deposit due to slowed water flow, forming localized high-concentration pollution areas. Then, based on the characteristics of the water area and the distribution characteristics of associated pollution, the water area is divided into multiple sub-regions. For example, based on the difference in flow velocity, it can be divided into upstream high-speed zone, midstream transition zone, and downstream sedimentation zone; based on the concentration of pollutants, it can be divided into high-pollution zone, low-pollution zone, and clean zone; based on the use of the water area, it can be divided into drinking water source zone, industrial drainage zone, agricultural irrigation zone, etc. The purpose of establishing water area zones is to make pollution monitoring more targeted. Different zones can adopt different sensor deployment strategies, thereby improving the accuracy of pollution source tracing and location.

[0057] Step S200: The water area is divided into regional nodes, a linkage sensor group is deployed, and a wide area network communication is established in the linkage sensor group to establish a water quality field perception network.

[0058] Specifically, after the water area is divided into zones, each zone is designated as a zone node, meaning each sub-region is defined as an independent monitoring node. Each node is responsible for monitoring pollutant concentrations, flow rates, and water quality parameters within its zone, and has a unique identifier for spatial correlation analysis during pollution source tracing. Deploying a coordinated sensor network involves deploying a collaborative sensing unit composed of multiple types of sensors within each node. This may include pollutant sensors to detect the concentrations of target pollutants such as ammonia nitrogen and heavy metals; flow meters, water level gauges, and flow meters to capture hydrodynamic characteristics; and pH meters, dissolved oxygen sensors, and turbidity meters to monitor water quality parameters. All sensors collect data with a unified timestamp, and pollutant data is correlated with flow rates and water quality parameters, such as whether high concentrations of pollution are accompanied by sudden changes in flow rate. When a sensor detects an anomaly, its neighboring nodes automatically increase their monitoring frequency. Then, a wide-area network communication system, such as 5G, LoRa, or NB-IoT, is established within the coordinated sensor network to achieve low-latency and highly reliable communication between nodes, ensuring real-time early warning of pollution events. Ultimately, this forms a water quality field sensing network, i.e., a distributed intelligent monitoring unit. This invention eliminates blind spots in water monitoring by deploying multiple nodes, captures the pollutant diffusion process in real time, and allows nodes in different zones to share data and jointly determine the source of pollution. For example, after the pollution concentration at the upstream node increases, it can predict when the downstream node will reach its peak.

[0059] Step S300: After unifying the timestamp, obtain pollutant concentration data with timestamp and node identifiers based on the water quality field sensing network, and simultaneously establish node flow velocity data.

[0060] Specifically, a unified timestamp refers to assigning a strictly synchronized time stamp to the data collected by all nodes. This ensures that data collected by upstream nodes can be compared with data collected by downstream nodes at the same time, avoiding misjudgments caused by asynchronous device clocks, such as mistaking delayed pollution for new pollution sources. Then, the concentration data of target pollutants such as chemical oxygen demand, ammonia nitrogen, and heavy metals are acquired in real time by the water quality field sensing network, with timestamp and node identifiers. The timestamp records the precise time of data collection, and the node identifier is used to uniquely identify the partition from which the data originates. In addition, it is also necessary to simultaneously collect the hydrodynamic parameters synchronized with the pollutant data of each node, including instantaneous flow velocity, flow direction, water level, and water depth changes, to form node flow velocity data. This data is used to determine the pollutant diffusion rate and pollution propagation path, thereby contributing to the accurate tracing of water pollution sources.

[0061] In a further embodiment, S21, the concentration time series sequence of each partition node is determined based on the pollutant concentration data;

[0062] S22. Extract the peak features of the concentration time series and determine the time lag interval of the pollutant concentration peaks between each directed node pair based on the peak features.

[0063] Furthermore, determining the effective node pairs whose time delay intervals satisfy the spatiotemporal constraints from the plurality of directed node pairs includes:

[0064] Determine the difference between the theoretical time delay and the time delay interval between each pair of directed nodes;

[0065] The directed node pairs whose difference is less than the first preset threshold are denoted as the first node pairs, and the first node pairs that meet the preset conditions are denoted as the valid node pairs.

[0066] Furthermore, the preset conditions include:

[0067] The first node meets the preset indicators for medium concentration mutation characteristics, continuous exceedance characteristics, and stable characteristics.

[0068] In practical implementation, since the first node meets the preset criteria for concentration mutation characteristics, continuous exceedance characteristics, and stable characteristics, multiple features are extracted from the concentration time series in actual applications. The specific implementation is as follows:

[0069] Step S400: Perform concentration characteristic analysis of pollutant concentration data for each partition node to generate key time-series features of the node.

[0070] Specifically, the pollutant concentration data of each node is sorted by time to establish a concentration time series; after time alignment of the concentration time series, outlier removal is performed to establish an updated concentration time series; feature extraction is performed on the peak features, concentration mutations, continuous exceedance features, and stable features of the updated concentration time series, and key time series features of each node are established based on the above features.

[0071] Specifically, for each partition node, the concentration characteristics of pollutant concentration data are analyzed. The pollutant concentration data collected by each node are sorted chronologically to form a continuous concentration sequence. Since the sensor may be affected by environmental interference such as water turbidity and temporary equipment failure, the concentration sequence is time-aligned to unify the data streams of different nodes to the same time base. Then, outlier removal is performed, that is, abnormal data that deviates from the normal fluctuation range is identified and removed by using the sliding window statistical method. Then, the missing time point data is interpolated and compensated by using neighbor interpolation or mean filling to obtain a clean and continuous updated concentration time series to ensure the reliability of feature extraction. Then, feature extraction is performed on the peak features, concentration abrupt changes, persistent exceedance features, and stable features of the updated concentration time series. Peak feature extraction identifies sudden spikes in concentration, recording their amplitude, occurrence time, and duration to capture sudden pollution events. Concentration abrupt change features locate inflection points of steep concentration gradients by calculating the rate of change of concentration between adjacent time points, reflecting the start of pollution input. Persistent exceedance features statistically analyze the length and magnitude of consecutive periods where pollutant concentrations exceed safety thresholds, quantifying the persistent impact of pollution. Stable features analyze steady-state indicators such as the mean and fluctuation range of background concentration levels, characterizing the basic pollution load of the water body. Finally, the multi-dimensional features are fused to form structured key time series features at each node, ensuring efficient and accurate time-lag analysis and reverse tracing.

[0072] Step S500: Use the key temporal characteristics of the partition nodes to perform time-delay flow perception, determine the effective node pairs, and construct multiple backtracking paths based on the effective node pairs.

[0073] Specifically, the process involves pairing any two partition nodes, where the pairing is flow-oriented; calculating the arrival time difference of peak features based on the pairing results, and establishing a time lag interval based on the arrival time difference; matching and authenticating node path distance and node flow velocity data based on the time lag interval and the pairing results to obtain a first node pair; and determining valid node pairs from the first node pair.

[0074] Specifically, time-delay flow perception is achieved by utilizing the key temporal characteristics of nodes. This involves analyzing the time difference characteristics of pollutants appearing at different monitoring nodes and inversely deducing the pollution propagation path and pollution source location by combining the characteristics of water flow. In particular, node pairing is performed between any two regional nodes, that is, a directed connection relationship is established in the water flow network. The distributed monitoring nodes are transformed into a computable pollution propagation path model. Node pairing is flow-directed pairing, allowing only upstream nodes to pair with downstream nodes, following the direction of water flow, and excluding reverse pairing, such as downstream nodes being unlikely to pollute upstream nodes. This allows for the establishment of multiple pairs of directed nodes. Then, based on the node pairing results, the arrival time difference of the peak features is calculated. That is, the peak time of the same pollution event is extracted for directed node pairs, the arrival time difference is calculated and used as the time lag interval. Next, the actual path distance between directed node pairs is obtained, and the average flow velocity during this period is measured using a flow velocity sensor. Matching and authentication are performed based on the time lag interval, the path distance between directed node pairs, and the node flow velocity data. That is, the theoretical time lag is determined based on the ratio of the actual path distance to the average flow velocity, and then the difference between the theoretical time lag and the time lag interval is compared. When the difference between the two time lags is less than a first preset threshold, the matching and authentication are passed, and the first node pair is obtained, which includes the set of effective propagation paths and the time lag credibility score of each path. Finally, a pollution propagation directed graph is constructed using the authenticated paths. The starting point of the path without upstream nodes is listed as a high-risk area, and the overlapping area pointed to by multiple paths is upgraded to a key suspected area, which significantly improves the accuracy of pollution source tracing in complex water network environments.

[0075] In addition, the first node pair is assisted in authentication by invoking concentration mutation, continuous exceedance features, and stability features. The assisted authentication includes node mutation similarity analysis, continuous exceedance similarity analysis, and node data stability matching. Valid node pairs are obtained based on the assisted authentication, and multiple backtracking paths are constructed based on the valid node pairs.

[0076] Specifically, multi-dimensional feature cross-validation of pollutant concentration data is introduced, including the use of concentration abrupt changes, persistent exceedance features, and stable features, to assist in the authentication of the first node pair. This avoids misjudgment due to temporal coincidences of different pollution sources and identifies hidden intermittent emission sources. Specifically, the auxiliary authentication includes inter-node abrupt change similarity analysis, persistent exceedance similarity analysis, and node data stability matching. Inter-node abrupt change similarity analysis involves extracting the rising slope and peak shape of concentration abrupt changes in paired nodes, and then calculating the dynamic time-normalized distance to assess waveform similarity. Persistent exceedance similarity analysis involves determining the proportion of exceedance durations at upstream and downstream nodes and the decay pattern of downstream exceedance concentrations. Node data stability matching... This refers to comparing the standard deviation of concentration at nodes before and after a pollution event and the time taken to recover to the original stable level after a sudden increase, and checking the historical stability of upstream nodes of suspected pollution sources. Valid node pairs are determined through weighted fusion using a three-level weighted scoring system. The pollution propagation network is then reconstructed based on these valid node pairs. This includes excluding node pairs that only meet the time delay criteria but do not match the characteristics, identifying slow-release pollution pathways through the persistent characteristics of exceeding standards, and probabilistically classifying suspected pollution sources based on the consistency of multiple features. Finally, valid node pairs are generated, and multiple backtracking paths are constructed based on these valid node pairs. This process ensures the accuracy of pollution source location and pollution type identification, significantly improving the scientific validity and reliability of source tracing in complex water network environments.

[0077] Furthermore, the particle convergence region of pollutants is determined based on the multiple simulated paths, including:

[0078] S41. Determine the distance-concentration data from each downstream path node to the source endpoint for each simulated path. The distance-concentration data from all downstream path nodes to the source endpoint form a distance-concentration sequence. S42. Fit the pollutant concentration at the source endpoint and the terminal node of each simulated path to obtain the concentration fitting sequence for each simulated path.

[0079] S43. Determine the desired path from the multiple simulated paths based on the distance-concentration sequence and the concentration fitting sequence, and determine the particle convergence zone of the pollutant in the monitored water area based on the desired path.

[0080] Furthermore, S43 includes:

[0081] The node residuals for each simulated path are determined based on the distance-concentration sequence and the concentration fitting sequence.

[0082] The concentration consistency of the corresponding simulation path is determined based on the node residuals, and the simulation path with a concentration consistency greater than the second preset threshold is recorded as the desired path.

[0083] By intersecting all the expected paths, the particle convergence zone of pollutants in the monitored water area can be obtained.

[0084] Furthermore, by intersecting all the expected paths, the particle convergence zone of the pollutants in the monitored water area is obtained, including:

[0085] The monitored water area is divided into multiple grid cells, and all desired paths are intersected within these grid cells.

[0086] The trajectory density value of the desired path within each grid cell is statistically analyzed, and the particle convergence zone of the pollutants is determined based on the trajectory density value.

[0087] In specific implementation, particle tracking simulations are performed on the pollutant concentration data to obtain multiple simulated paths, and the particle convergence zone of the pollutants is determined based on the multiple simulated paths, specifically including:

[0088] Using the pollutant concentration data, particle reverse tracking simulations based on node flow velocity data are performed at each node to establish a set of simulated paths (including multiple simulated paths). The particle intersection zone is established by the dense intersection of the simulated path set trajectories.

[0089] Specifically, particle reverse tracking simulation based on node velocity data refers to the process of simulating the reverse motion of a large number of virtual pollutant particles in a water body, based on the principle of fluid dynamics inversion. This transforms discrete monitoring data into a continuous network of pollution propagation paths. In other words, by combining real-time velocity data from node monitoring, the possible trajectories of pollutants from detection points to potential sources are reconstructed. When the backtracking paths of a large number of simulated particles converge densely in a specific area, that area is identified as a high-probability pollution source location. Specifically, virtual particles are released at each detected pollution node location, where each particle carries pollution characteristic information. The information shows that the number of particles is positively correlated with the concentration of pollutants. Then, the reverse time parameters are configured, including setting the simulated clock to the reverse flow of the actual monitoring time and the direction of particle movement opposite to the direction of the measured flow velocity. At the same time, random disturbances are added to consider the turbulent diffusion effect. Then, the dynamic motion simulation of pollution propagation is carried out. That is, the simulated clock is run in reverse using the measured flow velocity data of the nodes, and each step of the particle movement corresponds to the backtracking of the actual time. When encountering the confluence of tributaries, the particles randomly select simulated paths according to the proportion of the diversion. When the particles reach the water boundary (such as the river source or shoreline), the particles stop moving, thus obtaining a set of simulated paths containing multiple possible pollution propagation paths.

[0090] Next, the water area is divided into multiple grid units, and the frequency of particle propagation paths passing through each grid unit is counted to determine the trajectory density of the grid unit. Finally, a particle convergence zone is established, and high trajectory density, medium trajectory density, and low trajectory density areas are defined to ensure the accuracy of pollution source location.

[0091] Specifically, before trajectory convergence, the process includes: extracting distance-concentration data from the end of the path to the source endpoint from the simulated path set, establishing a distance-concentration sequence for each simulated path; obtaining the node concentrations of the source endpoint and the end nodes of the path, and inputting the corresponding node concentrations as endpoint concentrations into the attenuation fitting channel to generate a concentration fitting sequence for the path; establishing path node residuals for each simulated path based on the distance-concentration sequence and the concentration fitting sequence; generating a concentration consistency score using the node residuals of the simulated path; and completing the dense trajectory convergence analysis after filtering the simulated path set based on the concentration consistency score.

[0092] By establishing a quantitative relationship between pollutant concentration and propagation distance, the massive number of paths generated by reverse tracing simulations are scientifically screened to improve the accuracy of pollution source location. Specifically, the distance from the end of each simulated path to the source monitoring node is extracted and associated with the corresponding pollutant concentration data to form a structured distance-concentration sequence. That is, each distance and the concentration values ​​of each downstream node during the pollution event are recorded to construct a scattered distance-concentration data. Based on the hydrodynamic diffusion theory, an attenuation fitting channel for pollutant concentration decay with distance is established. Then, the node concentrations of the source endpoint and the end nodes of the path are obtained, and the corresponding node concentrations are input as endpoint concentrations into the attenuation fitting channel to generate a theoretical attenuation curve. The fit is then checked with the measured distance-concentration sequence to generate a concentration fitting sequence for the path.

[0093] Specifically, the node residuals of the distance-concentration sequence and the concentration fitting sequence for each simulated path node are calculated. Then, abnormal paths with persistently high residuals are excluded. For example, if a path experiences a sudden increase in concentration at 800 meters, it may indicate secondary pollution input along the path. Next, a concentration consistency score threshold, i.e., a second preset threshold, is set. A concentration consistency score is generated based on the path node residuals and compared with the concentration consistency score threshold. Simulated paths with higher concentration consistency scores are selected as the desired paths. Finally, the selected desired paths are used for dense trajectory intersection analysis. The purpose of the above steps is to eliminate interference areas with dense trajectories but inconsistent concentration patterns, and to distinguish between primary and secondary pollution sources through score differences, ensuring the accuracy of pollution source tracing and location in complex tributary networks or intermittent emission scenarios.

[0094] Furthermore, S5 includes:

[0095] The backtracking path and the particle convergence area are spatially superimposed. The intersection area of ​​the backtracking path and the particle convergence area is recorded as the primary pollution source area, and the independent areas that do not intersect are recorded as the secondary pollution source areas. The source location is determined based on the primary pollution source areas and the secondary pollution source areas.

[0096] In practical applications, this specifically includes:

[0097] Step S700: Source tracing and localization are performed using multiple backtracking paths and particle intersection areas.

[0098] Specifically, by integrating multiple backtracking paths and particle convergence zones, the propagation characteristics in the time dimension and the trajectory characteristics in the spatial dimension are cross-validated to achieve precise pollution source tracing and localization. Specifically, the earliest pollution time node determined by time-lag analysis is matched with the time inversion results of the particle convergence zone, eliminating localization points with temporal logical conflicts, such as pollution release times at particle convergence points that are later than the initial downstream detection time. Then, the linear propagation paths provided by multiple backtracking paths and the point-like probability distributions provided by the particle convergence zone are spatially superimposed, and high-density convergence zones are selected along the extension lines of the backtracking paths. The overlapping area between the endpoint of the backtracking path and the high-density particle zone is then designated as the primary pollution source area, and areas meeting either the backtracking path or the particle convergence zone condition are designated as secondary pollution source areas. Areas with significant contradictions between the two types of results are excluded, ultimately completing the pollution source tracing and localization.

[0099] Furthermore, after S5, it also includes:

[0100] Multiple target sampling areas are determined based on the confidence levels of the primary and secondary pollution source areas;

[0101] Collect pollutant collection data from the multiple target sampling areas;

[0102] The consistency score of the target sampling area is calculated based on the pollutant collection data, and the source location is updated based on the consistency score.

[0103] In a specific embodiment, it includes:

[0104] The target sampling area is determined by the intersection of the backtracking path and the particle convergence zone; pollutant concentration data is collected in the target sampling area by a mobile sampling device to obtain mobile pollutant collection data; residual verification is performed on the backtracking path and particle convergence zone based on the pollutant collection data to generate a mobile sampling consistency score; the source tracing and positioning results are updated based on the mobile sampling consistency score.

[0105] The process involves on-site verification of backtracking results through mobile sampling, leading to dynamic optimization of pollution source location. Specifically, the spatial intersection of the time-delay endpoint of the backtracking path and the particle convergence zone is taken as the target sampling area for mobile sampling. If multiple independent areas exist, they are sorted by confidence level. Then, mobile sampling devices such as UAV water quality samplers and automatic cruise boats are used to collect pollutant concentration data in the target sampling area. For example, sampling points are radially deployed along the water flow direction with the suspected source as the center to obtain pollutant collection data. Then, residual verification is performed on the backtracking path and particle convergence zone based on the pollutant collection data. This includes calculating the relative error between the measured concentration and the predicted value at each sampling point, the Euclidean distance between the sampling point and the predicted core area, and the error between the predicted peak time and the sampling time. Weighting coefficients are configured according to the pollution type, and a weighted mobile sampling consistency score is calculated. Finally, the source location results are updated based on the mobile sampling consistency score. For example, if the mobile sampling consistency score is high, the current source location results are confirmed and the final pollution source coordinates are generated; if the mobile sampling consistency score is medium, the particle tracking parameters are locally adjusted and recalculated; if the mobile sampling consistency score is low, a full-process reanalysis is triggered.

[0106] In a further embodiment, the method further includes configuring a verification monitoring cycle based on the source tracing and positioning results, performing periodic monitoring using a mobile data acquisition device within the verification monitoring cycle, establishing a periodic verification dataset, and generating a source tracing and positioning verification early warning based on the periodic verification dataset.

[0107] Specifically, periodic monitoring enables dynamic tracking and early warning of source tracing conclusions. Specifically, a verification monitoring cycle is configured based on the source tracing results. This involves automatically initializing the monitoring cycle based on the pollutant half-life of the pollution source, and then using a mobile data acquisition device to perform periodic monitoring within the verification monitoring cycle to obtain a periodic verification dataset, including pollutant concentration distribution and time-series fluctuation characteristics. Finally, a source tracing verification early warning is generated based on the periodic verification dataset. For example, if the similarity between the measured pollution range and the predicted range, and the deviation between the actual detected value and the theoretical value, meet preset thresholds, the source tracing verification early warning is green, indicating that the existing treatment plan should be maintained. When the deviation does not meet the preset thresholds, the source tracing verification early warning is yellow or red, indicating that treatment parameters need to be optimized or source control measures need to be initiated, significantly improving the precision and intelligence of distributed water body supervision.

[0108] The distributed water pollution source tracing method according to embodiments of the present invention has been described in detail above. Next, the distributed water pollution source tracing system according to embodiments of the present invention will be described.

[0109] A distributed water pollution tracing system includes a node pair construction module for constructing multiple directed node pairs based on the topology of nodes in each zone of the monitored water area; and a backtracking path construction module for determining the time lag interval of pollutant concentration peaks between each directed node pair based on pollutant concentration data in the monitored water area; determining effective node pairs whose time lag intervals satisfy spatiotemporal constraints from the multiple directed node pairs; and constructing multiple backtracking paths based on the effective node pairs.

[0110] Particle convergence zone generation module: used to perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle convergence zone of the pollutants based on the multiple simulated paths; Source tracing and positioning module: used to determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence zone, and determine the source tracing and positioning based on the pollution source area.

[0111] The distributed water pollution source tracing system of this invention addresses the technical problems of low efficiency, insufficient spatiotemporal dynamic analysis, and limited accuracy of pollution source location in existing technologies, achieving the technical effects of multi-node collaborative sensing, dynamic time-delay analysis, and high-precision pollution source location.

[0112] like Figure 2 As shown, in practical applications, the distributed water pollution source tracing system includes: a water area division module 10, a water quality field perception network establishment module 20, a pollutant concentration data acquisition module 30, a concentration characteristic analysis module 40, a time-delay flow perception module 50, a simulated path set establishment module 60, and a source tracing and positioning module 70.

[0113] The system includes: a water area partitioning module 10, used to interactively acquire water area characteristics and associated pollution distribution characteristics, and perform water area partitioning based on these characteristics to establish water area partitions; a water quality field sensing network establishment module 20, used to deploy linked sensor groups as regional nodes in the water area partitions, and establish wide area network communication within the linked sensor groups to establish a water quality field sensing network; a pollutant concentration data acquisition module 30, used to obtain pollutant concentration data with timestamp and node identifiers based on the water quality field sensing network after a unified timestamp, and simultaneously establish node flow velocity data; and a concentration feature analysis module 40, used to analyze the concentration characteristics of each water area partition. Each node performs concentration feature analysis on pollutant concentration data to generate key temporal features of the nodes; a time-delay flow sensing module 50 is used to perform time-delay flow sensing using the key temporal features of the nodes and construct multiple backtracking paths; wherein the time-delay flow sensing module 50 executes the functions corresponding to the node pair construction module and the backtracking path construction module of the present invention; a simulated path set establishment module 60 is used to perform particle reverse tracking simulation based on node flow velocity data at each node using the pollutant concentration data, establish a simulated path set, and determine the particle intersection area using the dense intersection of the simulated path set trajectories; the simulated path set establishment module 60 executes the functions corresponding to the particle intersection area generation module.

[0114] The source tracing and positioning module 70 is used for source tracing and positioning using multiple backtracking paths and particle intersection areas.

[0115] The specific configuration of the simulated path set establishment module 60 will be described in detail below. The simulated path set establishment module 60 further includes: extracting distance-concentration data from the end of the path to the source node from the simulated path set, establishing a distance-concentration sequence; obtaining the concentrations of the source endpoint and the end node of the path, and inputting the corresponding node concentrations as endpoint concentrations into the attenuation fitting channel to generate path concentration fitting results; establishing the node residuals of the path based on the distance-concentration sequence and the path concentration fitting results; generating a concentration consistency score using the path node residuals; and completing the trajectory dense intersection analysis after screening the simulated path set based on the concentration consistency score.

[0116] The specific configuration of the source tracing and positioning module 70 will be described in detail below. The source tracing and positioning module 70 further includes: determining a mobile target sampling area based on the intersection of the multiple backtracking paths and the particle convergence zone; collecting pollutant concentration data in the mobile target sampling area using a mobile acquisition device to obtain pollutant collection data; performing residual verification on the multiple backtracking paths and the particle convergence zone based on the pollutant collection data to generate a mobile sampling consistency score; and updating the source tracing and positioning results based on the mobile sampling consistency score.

[0117] The specific configuration of the concentration feature analysis module 40 will be described in detail below. The concentration feature analysis module 40 further includes: sorting the pollutant concentration data of each node by time to establish a concentration sequence; performing time alignment on the concentration sequence, removing outliers, and establishing an updated concentration sequence; and extracting the peak features, concentration mutations, continuous exceedance features, and stable features of the updated concentration sequence to establish key temporal features of the nodes.

[0118] The specific configuration of the time-delay flow sensing module 50 will be described in detail below. The time-delay flow sensing module 50 further includes: performing node pairing of any two regional nodes, wherein the node pairing is a flow-direction pairing; calculating the arrival time difference of peak characteristics based on the node pairing result, and establishing a time-delay interval using the arrival time difference; matching and authenticating node path distance and node flow velocity data based on the time-delay interval and the node pairing result, and selecting a first node pair; and constructing multiple backtracking paths based on the first node pair.

[0119] The specific configuration of the time-delay flow sensing module 50 will be described in detail below. The time-delay flow sensing module 50 further includes: calling concentration mutation, continuous exceedance features, and stable features to perform auxiliary authentication on the first node pair, wherein the auxiliary authentication includes node mutation similarity analysis, continuous exceedance similarity analysis, and node data stability matching; filtering valid node pairs according to the auxiliary authentication, and establishing multiple backtracking paths according to the valid node pairs.

[0120] The specific configuration of the source tracing and positioning module 70 will be described in detail below. The source tracing and positioning module 70 further includes: configuring a verification monitoring period based on the source tracing and positioning results; performing periodic monitoring using a mobile data acquisition device within the verification monitoring period to establish a periodic verification dataset; and generating a source tracing and positioning verification early warning based on the periodic verification dataset.

[0121] This invention achieves the collaborative deployment of a distributed sensor network by constructing a topological structure of nodes in each zone within the monitored water area and generating directed node pairs. Through multi-node data interaction, the pollution monitoring range is significantly expanded, enabling the source tracing system to simultaneously capture the pollutant diffusion characteristics of a large area of ​​water, thus improving source tracing efficiency.

[0122] This invention accurately identifies effective node pairs that satisfy spatiotemporal dynamics by constraining the difference between the theoretical time delay and the time delay interval of directed node pairs, quantifies the migration time delay of pollutant concentration peaks, reveals the dynamic laws of pollution diffusion, and solves the problem of insufficient analysis of time-varying characteristics by traditional methods, providing a temporal basis for the construction of backtracking paths. This invention generates multiple simulated paths through particle reverse tracking simulation and uses distance-concentration sequence and concentration decay fitting techniques to screen out desired paths with high concentration consistency. Furthermore, through gridded trajectory intersection analysis, spatial positioning of pollutant particle convergence zones is achieved. This technology overcomes the limitations of traditional source tracing methods that rely on diffusion models, significantly improving the geometric accuracy of pollution source positioning.

[0123] This invention divides the pollution source area into primary and secondary pollution source areas by spatially superimposing the backtracking path and the particle convergence zone, constructing a two-level localization framework. By combining confidence assessment of the target sampling area with data consistency verification, the source tracing results are dynamically corrected, solving the localization deviation problem caused by data noise in traditional methods and achieving a closed-loop accuracy for pollution source localization.

[0124] The source tracing system of this invention achieves decoupled design of algorithm and hardware through the collaborative work of the node pair construction module, the backtracking path construction module, the particle convergence zone generation module, and the source tracing and positioning module. When the monitored water area expands, only sensor nodes need to be added and the topology adjusted, without the need to reconstruct the entire system, which significantly reduces the cost and complexity of large-scale water area deployment.

[0125] The distributed water pollution source tracing system provided in the embodiments of the present invention can execute the distributed water pollution source tracing method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0126] The above description is merely a few embodiments of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any modifications or alterations made by those skilled in the art without departing from the scope of the technical solution of the present invention using the disclosed technical content are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A method for tracing pollution sources in distributed water bodies, characterized in that, include: S1. Construct multiple directed node pairs based on the topological structure of each zone node within the monitored water area; S2. Determine the time lag interval of pollutant concentration peak values ​​between each directed node pair based on the pollutant concentration data in the monitored water area. S21. Determine the concentration time series of each partition node based on the pollutant concentration data; S22. Extract the peak features of the concentration time series, and determine the time lag interval of the pollutant concentration peaks between each directed node pair based on the peak features; S3. Determine the effective node pairs whose time delay intervals satisfy the spatiotemporal constraints from the multiple directed node pairs, and construct multiple backtracking paths based on the effective node pairs; S4. Perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle convergence area of ​​the pollutants based on the multiple simulated paths; The particle convergence zone of pollutants is determined based on the multiple simulated paths, including: S41. Determine the distance-concentration data from each downstream path node to the source tracing endpoint for each simulated path. The distance-concentration data from all downstream path nodes to the source tracing endpoint constitute a distance-concentration sequence. S42. Perform concentration decay fitting on the pollutant concentration at the source endpoint and terminal node of each simulated path to obtain the concentration fitting sequence for each simulated path. S43. Determine the desired path from the multiple simulated paths based on the distance-concentration sequence and the concentration fitting sequence, and determine the particle convergence zone of the pollutant in the monitored water area based on the desired path; S43 includes: The node residuals for each simulated path are determined based on the distance-concentration sequence and the concentration fitting sequence. The concentration consistency of the corresponding simulation path is determined based on the node residuals, and the simulation path with a concentration consistency greater than the second preset threshold is recorded as the desired path. By intersecting all the expected paths, the particle convergence zone of pollutants in the monitored water area can be obtained; S5. Determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence area, and determine the source location based on the pollution source area.

2. The pollution source tracing method according to claim 1, characterized in that, From the plurality of directed node pairs, determine the valid node pairs whose time delay intervals satisfy the spatiotemporal constraints, including: Determine the difference between the theoretical time delay and the time delay interval between each pair of directed nodes; The directed node pairs whose difference is less than the first preset threshold are denoted as the first node pairs, and the first node pairs that meet the preset conditions are denoted as the valid node pairs.

3. The pollution source tracing method according to claim 2, characterized in that, The preset conditions include: The first node meets the preset indicators for medium concentration mutation characteristics, continuous exceedance characteristics, and stable characteristics.

4. The pollution source tracing method according to claim 1, characterized in that, By intersecting all the expected paths, the particle convergence zone of the pollutants in the monitored water area is obtained, including: The monitored water area is divided into multiple grid cells, and all desired paths are intersected within these grid cells. The trajectory density value of the desired path within each grid cell is statistically analyzed, and the particle convergence zone of the pollutants is determined based on the trajectory density value.

5. The pollution source tracing method according to claim 1, characterized in that, S5 include: The backtracking path and the particle convergence area are spatially superimposed. The area where the backtracking path and the particle convergence area intersect is recorded as the primary pollution source area, and the independent areas where they do not intersect are recorded as the secondary pollution source areas. The source location is determined based on the primary pollution source area and the secondary pollution source area.

6. The pollution source tracing method according to claim 5, characterized in that, Following S5 are: Multiple target sampling areas are determined based on the confidence levels of the primary and secondary pollution source areas; Collect pollutant collection data from the multiple target sampling areas; The consistency score of the target sampling area is calculated based on the pollutant collection data, and the source location is updated based on the consistency score.

7. A distributed water pollution source tracing system, characterized in that, For performing the pollution source tracing method as described in any one of claims 1-6, comprising: Node pair construction module: used to construct multiple directed node pairs based on the topology of nodes in each zone within the monitored water area; The backtracking path construction module is used to determine the time lag interval of the pollutant concentration peak between each directed node pair based on the pollutant concentration data in the monitored water area; determine the effective node pairs whose time lag intervals satisfy the spatiotemporal constraints from the multiple directed node pairs; and construct multiple backtracking paths based on the effective node pairs. Particle confluence zone generation module: used to perform particle tracking simulation on the pollutant concentration data to obtain multiple simulated paths, and determine the particle confluence zone of the pollutants based on the multiple simulated paths; Source tracing and location module: used to determine the pollution source area based on the spatial intersection of the multiple backtracking paths and the particle convergence area, and to determine the source tracing and location based on the pollution source area.

Images