Pollution diffusion traceability system fusing hydrological parameters and particle tracking

Abstract

The invention belongs to the technical field of environmental monitoring and hydrological simulation, and particularly relates to a hydrological parameter and particle tracking fused pollution diffusion traceability system, which is characterized in that real-time hydrological observation data and high-resolution hydrodynamic model output are subjected to space-time alignment by constructing a dynamic coupling mechanism; a three-dimensional Lagrange particle tracking model is driven; a quasi-Newton method and finite element method collaborative optimization objective function inversion framework is adopted, and the space coordinates, the release time and the discharge flux of a pollution source are synchronously determined in a multi-source and strong disturbance scene; and non-tampering evidence storage is carried out on the key traceability evidence chain based on the alliance chain architecture. The structural defects of hydrological parameter staticization, inversion algorithm simplification, data credibility deficiency, response mechanism lagging, cross-domain cooperation difficulty and the like in a traditional pollution diffusion traceability system are overcome, and full-chain automation and precision support from pollution identification, path tracking, source inversion to responsibility determination are achieved.

Description

A pollution diffusion tracing system integrating hydrological parameters and particle tracking Technical Field

[0001] This invention belongs to the field of environmental monitoring and hydrological simulation technology, specifically a pollution diffusion tracing system that integrates hydrological parameters and particle tracking. Background Technology

[0002] In the water environment governance and ecological risk prevention and control system, pollution diffusion source tracing technology serves as a core means to support emergency response to sudden water pollution incidents, pollution liability determination, and ecological damage compensation. Its accuracy and timeliness directly affect the effectiveness of watershed management and the level of cross-regional collaborative governance. Especially in current watershed integrated management projects driven by Eco-Oriented Development (EOD) models, rapid and accurate inversion of pollution source location, emission time, and emission volume has become a key technical prerequisite for realizing the principle of "whoever pollutes, bears the responsibility." Traditional pollution diffusion source tracing methods mostly rely on particle tracking models, simulating the migration path of pollutants in water bodies to achieve reverse source tracing. Their theoretical basis stems from the stochastic walk mechanism within the Lagrange framework and is often supplemented by objective function optimization strategies for parameter inversion. Specifically, existing technologies typically set water flow velocity and diffusion coefficient as fixed constants, meaning they are not dynamically adjusted according to actual hydrological conditions during model operation. While such static parameterization strategies have certain applicability in idealized or steady-state hydrodynamic scenarios, they are difficult to reflect the real river system under precipitation conditions. The complex dynamic characteristics of flow velocity abrupt changes, water depth fluctuations, and boundary morphology evolution caused by factors such as rainfall, tides, and dam scheduling pose challenges. Although some studies have attempted to introduce high-resolution hydrodynamic models such as Suntanns to provide background flow fields, their output data often fails to effectively integrate with real-time hydrological observations obtained from on-site automatic monitoring stations. This leads to significant deviations between the physical field upon which particle tracking relies and the actual water body state. Diffusion simulations built on this basis are prone to trajectory deviations and concentration distribution distortions under complex conditions such as strong convection, non-uniform flow, or river network confluence, severely impacting the reliability of source tracing conclusions. Existing source tracing algorithms often employ single numerical optimization methods, such as the quasi-Newton method to directly minimize the sum of squared residuals between observed and simulated concentrations, or the finite element method to perform forward analysis of the spatial concentration field followed by inverse calculation. However, in situations with multiple pollution sources coexisting, staggered emission times, or severe disturbances in the background flow field, the objective function often exhibits highly non-convex characteristics and multiple minima. Single algorithms are prone to getting trapped in local optima, making it difficult to simultaneously and accurately deduce the spatial coordinates of pollution sources. Release time and emission flux The reason for the lack of multi-dimensional parameters is that, although the quasi-Newton method has a relatively fast local convergence speed, it is sensitive to the initial guess and lacks global search capability; while the finite element method has high computational cost and is difficult to embed into the high-frequency iterative inversion process. The two are used separately and have failed to form a composite solution mechanism that complements each other, which restricts the improvement of source tracing accuracy in complex scenarios.

[0003] Therefore, how to construct a pollution diffusion tracing system that can dynamically couple real-time hydrological parameters with particle tracking models, integrate the advantages of multiple inversion algorithms to improve the source tracing accuracy in multi-source pollution scenarios, rely on blockchain technology to ensure data integrity and traceability, and achieve millisecond-level automatic response to excessive events through smart contracts, while supporting cross-regional multi-entity collaborative supervision, has become a key challenge and an urgent technical problem for those skilled in the art.

[0004] Therefore, a pollution diffusion tracing system that integrates hydrological parameters and particle tracking is proposed to address the above problems. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by establishing a pollution diffusion tracing system that integrates hydrological parameters and particle tracking, thereby solving the technical problems mentioned in the background art.

[0006] To address the above technical issues, the following technical solution is adopted: a pollution diffusion tracing system integrating hydrological parameters and particle tracking. This system constructs a dynamic coupling mechanism to spatiotemporally align real-time hydrological observation data with the output of a high-resolution hydrodynamic model, thereby driving a three-dimensional Lagrange particle tracking model. It employs a quasi-Newtonian method and a finite element method collaborative optimization objective function inversion framework to simultaneously determine the spatial coordinates, release time, and emission flux of pollution sources in multi-source, strongly disturbed scenarios. It relies on a consortium blockchain architecture to implement tamper-proof storage of key tracing evidence chains and initiates the inversion process in milliseconds through a smart contract triggered by a preset threshold. Based on the confluence path topology and cross-chain protocol, it breaks down data barriers across administrative regions, supporting pollution liability determination and ecological compensation calculation in EOD projects.

[0007] Preferably, the system includes a hydrological parameter acquisition and fusion module, a three-dimensional particle tracking model module, a composite inversion algorithm module, a blockchain evidence storage and smart contract module, a cross-regional collaborative supervision module, and a visualization and decision support module. The modules are interconnected through standardized interfaces to form a closed-loop pollution incident response system.

[0008] The hydrological parameter acquisition and fusion module is deployed in the network of automatic monitoring stations at key sections of the watershed to acquire water flow velocity components in real time. , diffusion coefficient , and water depth Basic hydrological elements are sampled at a frequency of no less than once every fifteen minutes. Preferably, the module synchronously accesses the background flow field forecast data output by the SUNTANS hydrodynamic model. The two are synchronously calibrated on an hourly basis in the time dimension. In the spatial dimension, the Kriging interpolation algorithm is used to map the discrete monitoring point data to SUNTANS unstructured grid nodes to generate a three-dimensional dynamic hydrological field covering the entire watershed. This hydrological field serves as the input boundary condition for the particle tracking model, ensuring that the flow velocity and dispersion parameters in the simulation process always reflect the real water body state.

[0009] The three-dimensional particle tracking model module constructs a pollutant migration path simulation mechanism based on the Lagrange framework. Its core motion equation is the three-dimensional convection-dispersion equation: ;in: This represents the rate of change of pollutant concentration at a certain location over time, expressed in mg / m³. 3 ·s; (i=x, y, z) represents the migration (directional transport) of pollutants at a background velocity ui, mg / m³ 3 · s; This represents concentration diffusion (non-directional) caused by molecular diffusion or turbulent mixing, in mg / m³. 3 · s; D represents the diffusion coefficient tensor (including , , (Isodirectional components); S represents source and sink terms (positive values ​​for pollution sources, negative values ​​for sinks); the pollutant migration path is simulated using the Lagrange particle tracking method, and the particle displacement within each time step Δt is calculated using a random walk model to handle the dispersion process.

[0010] The displacement equation for Lagrange particle tracking: This represents the particle's displacement in the x-direction; Represents the velocity component in the x-direction; Represents the time step; The dispersion coefficient represents the dispersion coefficient in the x-direction; Represents a standard normally distributed random number (used to simulate the randomness of molecular diffusion). ~N(0, 1).

[0011] The displacement equations for particles in the y and z directions are similar. The model supports the handling of particles under boundary conditions, such as using reflective boundary conditions when encountering riverbanks, and setting parameterized adsorption rates (user-definable) for the riverbed to approximate the physical adsorption process; ; ; All are independent random variables that follow a standard normal distribution N(0,1). When a particle encounters the riverbank boundary, it performs a reflection operation with a reflection coefficient set to 0.8. When it comes into contact with the riverbed interface, its mass decays at an adsorption rate of 15% per hour to approximate the physical adsorption process. All particle trajectory calculations are performed by a GPU parallel acceleration engine, and a single complete simulation takes no more than five minutes.

[0012] The composite inversion algorithm module constructs the objective function: ,in For the i-th monitoring point, the conductivity-concentration conversion formula is used. The actual observed concentration obtained through derivation This represents the simulated concentration output by the particle tracking model at the same location. This represents the historical observation error standard deviation for this monitoring point. The module first uses the finite element method to spatially discretize the three-dimensional water flow control equations, forming a large sparse coefficient matrix K. Then, a quasi-Newton method is used to iteratively update the pollution source parameter vector. ;in n represents spatial coordinates, t represents the release time, and M represents the emission flux; in each iteration, the accompanying model is adjusted according to the current... Calculate the gradient of the objective function And based on this, the search direction is adjusted, when the objective function value drops to The iteration terminates at the following point, and the final inversion result is output.

[0013] The blockchain evidence storage and smart contract module is deployed based on the Hyperledger Fabric consortium blockchain architecture. Participating nodes include ecological and environmental authorities, hydrological monitoring agencies, and third-party testing units. All original hydrological parameters, concentration observation records, and particle trajectory logs are generated into unique fingerprints using the SHA-256 hash algorithm and written to the distributed ledger at a rate of one block per hour to ensure data integrity and traceability. The smart contract pre-codes the pollution event triggering logic. When the conductivity k value of any monitoring point exceeds the threshold corresponding to the Class III limit in the national surface water environmental quality standards, the particle tracking and inversion process is immediately activated, and the entire response cycle is controlled within ten minutes.

[0014] The cross-regional collaborative supervision module establishes a pollution propagation network map based on the confluence path topology extracted from the digital elevation model, identifying administrative boundary nodes crossed by pollutants. This module achieves data interoperability between supervision systems in different administrative regions through the Cosmos cross-chain communication protocol, eliminating information fragmentation caused by inconsistent standards. The model layer calls upon finite element analysis results to quantify the area affected by pollution in each administrative region. Duration and concentration exceeding the standard Substitute into the compensation amount formula: m: Total number of areas affected by pollution; j: Area subscript (j=1,2,3,...,m), representing the j-th affected area; : The pollutant concentration exceeding the standard in the j-th region (the difference between the measured concentration and the emission standard); : The area of ​​the j-th affected region; : Duration of pollution in the j-th area; P: Compensation price per unit area (determined according to ecological damage assessment standards); The compensation amounts for all affected areas are summed up.

[0015] The model layer uses the finite element method to analyze the spatiotemporal distribution of pollution plumes in different administrative regions, providing a quantitative basis for the division of responsibilities, where P is the unit area compensation price determined based on the regional ecological function assessment.

[0016] The visualization and decision support module uses WebGL graphics rendering technology to present particle trajectories, concentration cloud maps, and the evolution of pollution plumes in a three-dimensional dynamic form. It allows users to trace back along the timeline or view the internal structure by cutting out specific sections. This module integrates a random forest machine learning model to predict the expansion trend of the pollution range in the next 24 hours based on historical diffusion patterns and current hydrological conditions. It outputs an assessment report that includes risk level classification. The application layer provides an open RESTful API interface, which can be seamlessly connected with existing emergency command platforms to achieve full-process linkage of command issuance, resource scheduling, and response feedback.

[0017] Preferably, in the hydrological parameter acquisition and fusion module, the automatic monitoring stations are spaced one kilometer apart, the flow velocity data is calibrated using an acoustic Doppler current profiler, and the diffusion coefficient value is set with reference to the regional hydrological manual. cubic meters per second Square meters per second.

[0018] Preferably, the three-dimensional particle tracking model module initializes one thousand virtual particles to represent pollutant points, with a time step of... Fixed at sixty seconds, the particle displacement calculation strictly follows the aforementioned random row preferred method. In the composite inversion algorithm module, the constant term of the conductivity-concentration conversion formula... Set the slope to 0.02. Set the value to 1.5, and the initial iteration step size to 500 meters.

[0019] Preferably, in the blockchain notarization and smart contract module, the data is uploaded to the blockchain once per hour, and the entire response time of the smart contract from triggering to generating a traceability report does not exceed ten minutes.

[0020] Preferably, in the cross-regional collaborative supervision module, the determination of pollution liability adopts a compensation amount formula, and a visual liability list is generated by combining the three-dimensional dynamic rendering results.

[0021] Preferably, the relative error between the diffusion prediction results for the next twelve hours output by the visualization and decision support module and the measured data does not exceed 10%.

[0022] The beneficial effects of this invention are as follows: This invention solves the structural defects of traditional pollution diffusion tracing systems, such as static hydrological parameters, simplistic inversion algorithms, lack of data credibility, lagging response mechanisms, and difficulties in cross-domain collaboration. It achieves full-chain automation and precision support from pollution identification, path tracing, source inversion to responsibility determination. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] In the accompanying drawings: Figure 1 is a schematic diagram of the overall structure of a pollution diffusion tracing system that integrates hydrological parameters and particle tracking according to the present invention.

[0025] Figure 2 is a schematic diagram of the working principle and particle displacement calculation process of the three-dimensional particle tracking model module in this invention.

[0026] Figure 3 is a schematic diagram of the objective function construction and iterative optimization process of the composite inversion algorithm module of the present invention.

[0027] Figure 4 is a schematic diagram of the data interaction and responsibility determination mechanism of the cross-regional collaborative supervision module of the present invention based on the confluence path topology and cross-chain protocol. Detailed Implementation

[0028] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0029] Specific implementation examples are given below.

[0030] Please refer to Figures 1-4. This invention provides a pollution diffusion source tracing system that integrates hydrological parameters and particle tracking. Its overall architecture consists of six functional modules, including a hydrological parameter acquisition and fusion module, a three-dimensional particle tracking model module, a composite inversion algorithm module, a blockchain evidence storage and smart contract module, a cross-regional collaborative supervision module, and a visualization and decision support module. Each module is interconnected through standardized interfaces to form a closed-loop pollution event response system, realizing full-process automated support from pollution identification, path tracing, source inversion to responsibility determination.

[0031] In some implementations, the hydrological parameter acquisition and fusion module is deployed in a network of automatic monitoring stations at key sections of the watershed to acquire water flow velocity components in real time. , diffusion coefficient , and water depth Basic hydrological elements, automatic monitoring stations are deployed at 1 km intervals, flow velocity data are calibrated using acoustic Doppler current profilers, and diffusion coefficient values ​​are set according to the regional hydrological manual. square meters per second With a sampling rate of no less than once every fifteen minutes, the module synchronously receives background flow field forecast data output from the SUNTANS hydrodynamic model. The two are synchronously calibrated on an hourly basis in the time dimension. In the spatial dimension, the Kriging interpolation algorithm is used to map discrete monitoring point data to SUNTANS unstructured grid nodes to generate a three-dimensional dynamic hydrological field covering the entire watershed. This hydrological field serves as the input boundary condition for the particle tracking model, ensuring that the flow velocity and dispersion parameters in the simulation process always reflect the real water body state.

[0032] In some implementations, the three-dimensional particle tracking model module constructs a pollutant migration path simulation mechanism based on the Lagrange framework, and its core motion equation is the three-dimensional convection-dispersion equation: ;in: This represents the rate of change of pollutant concentration at a certain location over time, expressed in mg / m³. 3 ·s; (i=x, y, z) represents the migration (directional transport) of pollutants at a background velocity ui, mg / m³ 3 · s; This represents concentration diffusion (non-directional) caused by molecular diffusion or turbulent mixing, in mg / m³. 3 • s; D represents the dispersion coefficient tensor (containing directional components such as Dx, Dy, and Dz); S represents the source and sink terms (positive values ​​represent pollution sources, and negative values ​​represent sinks); the pollutant migration path is simulated using the Lagrange particle tracking method, and the particle displacement within each time step Δt is calculated using a random walk model to handle the dispersion process.

[0033] The displacement equation for Lagrange particle tracking: ; This represents the particle's displacement in the x-direction; Represents the velocity component in the x-direction; Represents the time step; The dispersion coefficient represents the dispersion coefficient in the x-direction; Represents a standard normally distributed random number (used to simulate the randomness of molecular diffusion). ~N(0, 1).

[0034] The displacement equations for particles in the y and z directions are similar. The model supports the handling of particles under boundary conditions, such as using reflective boundary conditions when encountering riverbanks, and setting parameterized adsorption rates (user-definable) for the riverbed to approximate the physical adsorption process; ; ;in , , All are independent random variables that follow a standard normal distribution N(0,1). When a particle encounters the riverbank boundary, it performs a reflection operation with a reflection coefficient set to 0.8. When it comes into contact with the riverbed interface, its mass decays at an adsorption rate of 15% per hour to approximate the physical adsorption process. All particle trajectory calculations are performed by a GPU parallel acceleration engine, and a single complete simulation takes no more than five minutes.

[0035] The composite inversion algorithm module constructs the objective function: ;in, For the first Each monitoring point was converted using the conductivity-concentration formula. {The actual observed concentration obtained through derivation, This represents the simulated concentration output by the particle tracking model at the same location. This represents the standard deviation of historical observation errors for this monitoring point. The constant term in the conductivity-concentration conversion formula. Set the slope to 0.02. Set to 1.5, this module first uses the finite element method to spatially discretize the three-dimensional water flow control equations, forming a large sparse coefficient matrix. Subsequently, the pollution source parameter vector is iteratively updated using a quasi-Newton method: ;in Let n be the spatial coordinates, t be the release time, and M be the emission flux. The initial iteration step size is set to 500 meters. In each iteration, the accompanying model is adjusted according to the current... Calculate the gradient of the objective function And based on this, the search direction is adjusted, when the objective function value drops to The iteration terminates at the following point, and the final inversion result is output.

[0036] The blockchain evidence storage and smart contract module is deployed based on the Hyperledger Fabric consortium blockchain architecture. Participating nodes include ecological and environmental authorities, hydrological monitoring agencies, and third-party testing units. All original hydrological parameters, concentration observation records, and particle trajectory logs are generated into unique fingerprints using the SHA-256 hash algorithm and written to the distributed ledger at a rate of one block per hour to ensure data integrity and traceability. The data is uploaded to the blockchain once per hour. The smart contract pre-codes the pollution event triggering logic. When the conductivity k value of any monitoring point exceeds the threshold corresponding to the Class III limit in the National Surface Water Environmental Quality Standard GB3838-2002, the particle tracking and inversion process is immediately activated. The entire response cycle is controlled within ten minutes, that is, the entire response time from triggering to generating the source tracing report does not exceed ten minutes.

[0037] The cross-regional collaborative supervision module establishes a pollution propagation network map based on the confluence path topology extracted from the digital elevation model, identifying administrative boundary nodes crossed by pollutants. This module achieves data interoperability between supervision systems in different administrative regions through the Cosmos cross-chain communication protocol, eliminating information fragmentation caused by inconsistent standards. The model layer calls upon finite element analysis results to quantify the area affected by pollution in each administrative region. Duration and concentration exceeding the standard Substitute into the compensation amount formula: m: Total number of areas affected by pollution; j: Area subscript (j=1,2,3,...,m), representing the j-th affected area; : The pollutant concentration exceeding the standard in the j-th region (the difference between the measured concentration and the emission standard); : The area of ​​the j-th affected region; : Duration of pollution in the j-th area; P: Compensation price per unit area (determined according to ecological damage assessment standards); The compensation amounts for all affected areas are summed up.

[0038] The model layer uses the finite element method to analyze the spatiotemporal distribution of pollution plumes in different administrative regions, providing a quantitative basis for liability delineation. Here, P is the unit area compensation unit determined based on the regional ecological function assessment. Ecological damage accounting is completed, where P is the unit area compensation unit price determined based on the regional ecological function assessment. The pollution liability determination adopts the above compensation amount formula, and a visual liability list is generated by combining the three-dimensional dynamic rendering results.

[0039] The visualization and decision support module uses WebGL graphics rendering technology to present particle trajectories, concentration cloud maps, and the evolution of pollution plumes in a three-dimensional dynamic form. It allows users to trace back along the timeline or view the internal structure by cutting out specific sections. This module integrates a random forest machine learning model to predict the expansion trend of the pollution range in the next 24 hours based on historical diffusion patterns and current hydrological conditions, and outputs an assessment report including risk level classification. The relative error between the diffusion prediction results for the next 12 hours output by the visualization and decision support module and the measured data does not exceed 10%. The application layer has an open RESTful API interface, which can be seamlessly connected with the existing emergency command platform to realize the full-process linkage of command issuance, resource scheduling, and response feedback.

[0040] For example, a sudden ammonia nitrogen pollution event occurs in a river basin. At 09:15 on the same day, an automatic monitoring station downstream detected a sharp increase in the conductivity k-value to 0.45 mS / cm, exceeding the Class III water quality standard limit in GB3838-2002 (corresponding to a k-threshold of 0.40 mS / cm). At 09:16, the smart contract triggers the inversion process. The system immediately calls the three-dimensional dynamic hydrological field synchronized and calibrated for the most recent hour from the hydrological parameter acquisition and fusion module, starts the three-dimensional particle tracking model module, and initializes the model with one thousand virtual particles. The reverse tracing was performed with a time step of seconds, while the composite inversion algorithm module began iterative optimization with the initial guess that the source point was located five kilometers upstream of the monitoring station. After seven rounds of quasi-Newton method iterations, the objective function f converged to 8.7 × 10⁻⁶. The spatial coordinates of the pollution source were determined to be 118.325°E, 32.107°N, with a release time of 07:30 and an emission flux of 12.3 kg / h. This result, along with the original observation data, particle trajectory log, and hydrological field snapshot, was used to generate a SHA-256 hash fingerprint, which was written into the consortium blockchain at 10:05. The cross-regional collaborative monitoring module identified the source as being located within the administrative area of ​​City A, while the pollution plume had crossed into the waters of City B. Based on the confluence topology map, the affected area, duration, and excess concentration in City A and City B were calculated respectively. The visualization module simultaneously generated a three-dimensional dynamic source tracing report, showing that the pollution plume spread downstream along the main river channel from 07:30, reached the boundary of City B at 09:00, and covered the monitoring section at 10:30. The relative error between the predicted diffusion range in the next twelve hours and the measured data was 7.2%, which meets the system accuracy requirements.

[0041] To verify the technical effectiveness of this invention, a comparative example was set up: a two-dimensional Eulerian model driven by traditional static hydrological parameters was used for source tracing. The hydrological field was fixed as the average flow velocity and dispersion coefficient of the day before the event. Real-time observation data fusion was not introduced, and the inversion algorithm only used the least squares method. There was no blockchain evidence storage or cross-domain collaboration mechanism. Under the same pollution event, the comparative example system output the source tracing results at 11:20, with a positioning deviation of 2.8 kilometers, a release time error of 1.5 hours, and an emission flux estimation deviation of 31%. Moreover, it could not provide a credible chain of evidence for cross-administrative region responsibility determination.

[0042] The table below summarizes the key performance indicators of the examples and comparative examples: | Indicator | Example | Comparative Example | Source Tracing Response Time | 9 minutes | 125 minutes | |---|---|---|---| | Pollution Source Location Error | 0.3 km | 2.8 km | | Release Time Error | 12 minutes | 90 minutes | | Emission Flux | Relative Error | 6.5% | 31% | | Data Storage Integrity | Full-process immutability | No storage mechanism | | Cross-regional responsibility accounting capability | Supported | Not supported | | Relative error of future 12-hour prediction | 7.2% | 22.4% | In some implementations, the Kriging interpolation algorithm in the hydrological parameter acquisition and fusion module adopts a spherical variogram model, with the nugget effect set to 0.05, the sill value to 0.85, and the range to 1200 meters, to adapt to the watershed scale and monitoring station density. The SUNTANS model output has a time resolution of 10 minutes, an average spatial grid size of 50 meters, and three vertical layers to ensure that the details of the three-dimensional flow field are fully depicted.

[0043] In some implementations, in the three-dimensional particle tracking model module, Values The vertical velocity component uz is determined by the hydrostatic pressure gradient and turbulent mixing, and is directly assigned by the vertical velocity profile output by the SUNTANS model. In addition to riverbed adsorption, the particle mass decay mechanism also introduces a first-order degradation term, with a half-life of 48 hours to simulate the biochemical degradation process.

[0044] In some implementations, the composite inversion algorithm module uses tetrahedral elements for finite element spatial discretization, with the element size adaptively refined within a 5-kilometer radius of the monitoring point, and the minimum element side length being 20 meters. The accompanying model efficiently calculates the gradient using automatic differentiation technology, avoiding numerical noise caused by finite differences. The quasi-Newton method uses the BFGS update formula, and the line search satisfies the Wolfe condition to ensure global convergence.

[0045] In some implementations, the Hyperledger Fabric channel in the blockchain notarization and smart contract module is configured for three-organization consensus, and the endorsement policy requires signatures from at least two organization nodes before a transaction can be submitted. The smart contract is written in Go, deployed in a Docker container, and subscribes to real-time monitoring data streams in an MQTT message queue via an event listener.

[0046] In some implementations, the digital elevation model in the cross-regional collaborative supervision module is derived from a 1:50,000 topographic map, the confluence path extraction uses the D8 algorithm, and the administrative boundary vector data is uniformly released by the Ministry of Natural Resources. The Cosmos cross-chain protocol is implemented through the IBC (Inter-Blockchain Communication) standard, with each administrative region's supervision system registered as an independent Zone to the Hub, and assets and evidence securely transferred through the Packet mechanism.

[0047] In some implementations, the random forest model training set in the visualization and decision support module contains hydrological-meteorological-diffusion feature vectors of 217 historical pollution events in the past five years. The feature dimension is 38-dimensional, including rainfall intensity, wind speed, flow rate, slope, land use type, etc. The model uses 100 decision trees with a maximum depth of 15, an OOB error of 8.3%, and an online inference latency of less than 200 milliseconds.

[0048] In practical deployment, the system described in this invention can rely on existing water conservancy information infrastructure to transmit automatic monitoring station data back to the regional data center in real time via 4G / 5G or fiber optic leased lines. The SUNTANS model runs on a high-performance computing cluster, providing daily rolling forecasts of the flow field for the next 72 hours. Particle tracking and inversion tasks are dynamically allocated to the GPU server pool by the task scheduler to ensure responsiveness under high concurrency events. Consortium blockchain nodes are deployed in the local data centers of each participating party, and data privacy and compliance are ensured through leased line interconnection.

[0049] In the description of this invention, it should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions provided in this disclosure can be achieved, and no limitation is imposed herein.

[0050] The above description is merely a preferred embodiment of the present invention and does not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A pollution diffusion tracing system integrating hydrological parameters and particle tracking, characterized in that, The system includes a hydrological parameter acquisition and fusion module, a 3D particle tracking model module, a composite inversion algorithm module, a blockchain notarization and smart contract module, a cross-regional collaborative supervision module, and a visualization and decision support module. The hydrological parameter acquisition and fusion module is used to acquire the flow velocity components, diffusion coefficient, and water depth at key cross-sections of the basin in real time. It synchronously calibrates the data from automatic monitoring stations with the background flow field output from the SUNTANS hydrodynamic model in the time dimension on an hourly basis, and maps it to unstructured grid nodes in the spatial dimension through Kriging interpolation, generating a 3D dynamic hydrological field covering the entire basin. The 3D particle tracking model module is based on the Lagrange framework and utilizes... The displacement calculation of virtual particles is driven by the three-dimensional dynamic hydrological field, and the displacement is composed of the superposition of deterministic convection term and random diffusion term. The composite inversion algorithm module constructs an optimization framework with the weighted sum of squares of the residuals between observed concentration and simulated concentration as the objective function. It combines the finite element method to discretize the water flow control equation and uses the quasi-Newton method to iteratively update the pollution source parameter vector. The pollution source parameter vector includes spatial coordinates, release time and emission flux. The blockchain storage and smart contract module stores the original observation data, particle trajectory log and inversion results in an immutable manner based on the consortium blockchain architecture, and triggers the smart contract to start the inversion process when the conductivity of the monitoring point exceeds a preset threshold. The cross-regional collaborative supervision module identifies administrative boundary crossing nodes based on the topology of the confluence path extracted by the digital elevation model, realizes data interoperability between multiple administrative region supervision systems through cross-chain protocol, and calculates the amount of ecological compensation based on the affected area, duration and excess concentration of each administrative region; the visualization and decision support module uses WebGL technology to dynamically render particle trajectories and pollution plume evolution process in three dimensions, and integrates machine learning models to predict future pollution diffusion trends.

2. The pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: In the hydrological parameter acquisition and fusion module, the automatic monitoring stations are spaced one kilometer apart. Flow velocity data is calibrated using an acoustic Doppler current profiler, and the diffusion coefficient... and The values ​​are respectively square meters per second and The sampling rate is 1 square meter per second, with a sampling frequency of no less than once every 15 minutes.

3. The pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: The 3D particle tracking model module initializes one thousand virtual particles, with a time step of [missing information]. Fixed at sixty seconds, the particles in 、 、 The displacements in the directions are respectively determined by: ； ； ； The calculation shows that, among which 、 、 The particles are independent, standard normally distributed random variables; they perform reflection operations at the riverbank boundary and decrease in mass at the riverbed interface according to the adsorption rate.

4. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: In the composite inversion algorithm module, the observed concentration From conductivity After conversion formula The objective function is derived as follows: The initial iteration step size is 500 meters. When the objective function value drops to The iteration will terminate and the pollution source parameters will be output.

5. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: The blockchain evidence storage and smart contract module is deployed on the Hyperledger Fabric consortium blockchain. Participating nodes include ecological and environmental authorities, hydrological monitoring agencies, and third-party testing units. All key data are fingerprinted using SHA-256 hashing and written to the distributed ledger in blocks one per hour. The response time of the smart contract from triggering to generating the traceability report does not exceed ten minutes.

6. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: The cross-regional collaborative supervision module achieves data interoperability between supervision systems in different administrative regions through the Cosmos cross-chain communication protocol. The ecological compensation amount is based on the following formula: ; calculate, where For the first Concentration exceeding the standard in the administrative region For the affected area, For duration, Compensation is based on the unit price per unit area.

7. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: In the three-dimensional particle tracking model module, the vertical dispersion coefficient Dz takes the value of square meters per second, vertical velocity component The vertical velocity profile output from the SUNTANS model is directly assigned, and a first-order degradation term with half-life is introduced to simulate the biochemical decay of pollutants.

8. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: The composite inversion algorithm module uses tetrahedral elements to discretize the three-dimensional water flow control equations in finite element space. The element size is adaptively refined to a minimum side length of 20 meters within a set range around the monitoring point. The accompanying model calculates the gradient of the objective function through automatic differentiation technology. The quasi-Newton method adopts the BFGS update formula and a line search strategy that satisfies the Wolfe condition.

9. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: In the blockchain notarization and smart contract module, the Hyperledger Fabric channel is configured for three-organization consensus. The endorsement policy requires at least two organization nodes to sign before a transaction can be submitted. The smart contract is written in Go and subscribes to the real-time monitoring data stream in the MQTT message queue through an event listener.

10. A pollution diffusion tracing system integrating hydrological parameters and particle tracking according to claim 1, characterized in that: The random forest model integrated into the visualization and decision support module is built on a training set containing multiple historical pollution events, including rainfall intensity, wind speed, flow rate, slope, and land use type.

Images