A water ecological pollution diffusion prediction method and system based on big data

Abstract

The application relates to the technical field of environmental monitoring and discloses a water ecological pollution diffusion prediction method and system based on big data, which comprises the following steps: a sensing network of a water body environment is constructed, real-time water body environment data are acquired, the data are preprocessed by using an edge computing node, and then the data are synchronously transmitted to a central platform; a dynamic prediction model of pollutant boundary condition parameters is established based on historical data and the real-time water body environment data; a pollutant propagation graph is constructed according to the dynamic boundary condition parameters, the spatiotemporal distribution trend of pollutant concentration is analyzed, and the position of a pollution source is identified; according to the pollution source position identification result, a time-effect prediction is performed, and a pollution early warning response is triggered in combination with a set threshold value. The spatial coverage and timeliness of pollution monitoring are improved, the intelligence and credibility of pollution tracking are enhanced, and the intelligence and scientific level of water ecological environment pollution prevention and control are improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring technology, specifically to a method and system for predicting the spread of water ecological pollution based on big data. Background Technology

[0002] The diffusion of pollutants in water bodies such as rivers and lakes exhibits distinct spatiotemporal evolution characteristics and is influenced by multiple factors, including flow velocity, water level, rainfall, and emissions, displaying complex features of strong nonlinearity and dynamic change. Therefore, accurately and timely predicting pollutant diffusion trends, identifying potential pollution sources, and conducting effective early warning systems have become key technological requirements for aquatic ecological environment protection.

[0003] In existing technologies, pollution diffusion models based on fluid dynamics principles (such as two-dimensional hydrodynamic models and finite element method simulations) are commonly used to analyze the migration and diffusion processes of pollutants in water bodies. While these methods possess high theoretical accuracy, they are sensitive to initial boundary conditions, involve complex modeling, have a heavy computational burden, and struggle to respond in real-time to dynamically changing hydrological conditions. Furthermore, pollution source identification often relies on expert experience or fixed rules, lacking the dynamic inversion capability to be linked with big data.

[0004] Furthermore, traditional water quality monitoring relies mainly on a small number of fixed monitoring points, resulting in insufficient spatiotemporal coverage and making it difficult to comprehensively perceive the overall situation of pollution spread. However, with the development of remote sensing technology, IoT sensors, and edge computing, a foundation has been laid for large-scale, high-frequency, and multi-dimensional data collection and processing, driving the transformation of pollution spread prediction from "qualitative monitoring" to "intelligent analysis" and "early warning response."

[0005] However, there is still a lack of intelligent methods and systems capable of integrating multi-source water environment data, dynamically constructing boundary conditions, identifying pollution sources, predicting pollution diffusion paths, and automatically triggering early warnings. Therefore, there is an urgent need to build a water ecological pollution diffusion prediction system based on big data, with real-time processing capabilities and pollution early warning functions, in order to improve pollution prevention and control efficiency and ecological security response capabilities. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by this invention is that existing water pollution diffusion prediction methods rely heavily on boundary conditions, have inaccurate pollution source identification, poor real-time prediction performance, and lack intelligent early warning mechanisms.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a water ecological pollution diffusion prediction method based on big data, comprising: constructing a water environment sensor network, acquiring real-time water environment data, preprocessing the data using edge computing nodes, and synchronously transmitting the data to a central platform;

[0009] A dynamic prediction model for pollutant boundary condition parameters is established based on historical data and real-time water environment data.

[0010] Based on the dynamic boundary condition parameters, a pollutant propagation map is constructed to analyze the spatiotemporal distribution trend of pollutant concentration and identify the location of pollution sources.

[0011] Based on the results of pollution source location identification, timely predictions are performed, and pollution early warning responses are triggered in conjunction with set thresholds.

[0012] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method of the present invention, the real-time water environment data includes water quality monitoring data, hydrodynamic related data, meteorological background data, remote sensing and image data, biological and ecological indicator data, and metadata data.

[0013] The preprocessing includes performing noise filtering and outlier detection, removing false alarm data from sensors, filling in missing data using logical rules and model-driven mechanisms, and completing data cleaning operations to achieve data standardization and normalization; and performing format unification and temporal and spatial dimension alignment processing on data from different sensors.

[0014] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method of the present invention, the dynamic prediction model for establishing pollutant boundary conditions includes: setting a time step, using preprocessed historical data and real-time data from the past n time steps, and organizing a water environment data sequence of n consecutive time steps as input to the dynamic prediction model through a sliding window method.

[0015] The dynamic prediction model adopts a fusion prediction framework, which consists of a parallel linear regression model and a deep learning model.

[0016] The linear regression model captures the linear changing trend of aquatic environmental parameters and outputs prediction results based on the aquatic environmental data sequence of the past n time steps.

[0017] The deep learning model uses a GRU network structure to capture the nonlinear dynamic characteristics of water environment parameters, predicts water environment data sequences, and supports online incremental training to adapt to real-time data changes.

[0018] The predicted outputs of the linear regression model and the deep learning model are weighted and fused to generate the final pollutant boundary condition parameter output.

[0019] The pollutant boundary condition parameters include: water flow rate, flow rate, pollutant concentration, and water level at the boundary.

[0020] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method described in this invention, the dynamic prediction model further includes updating the sliding window input when new real-time water environment data arrives, and updating the coefficients of the linear regression model and updating the parameters of the deep learning model through incremental training based on real-time data, thereby achieving continuous optimization of the model prediction accuracy.

[0021] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method of the present invention, the construction of pollutant propagation map includes generating a hydrological connectivity map based on dynamic boundary condition parameters, combined with water body geographic information and watershed structure. The nodes in the hydrological connectivity map represent monitoring points of key water areas, tributary confluences and drainage outlets, and the edges represent the flow relationships between water bodies.

[0022] The dynamic boundary condition parameters are used to dynamically assign values ​​to the edges in the hydrological connectivity graph, determine the direction of water flow, velocity and flow rate attributes, and the pollutant concentration attributes of the nodes are also determined by the dynamic boundary condition parameters and real-time collected data.

[0023] According to the predetermined time steps, based on the updated dynamic boundary condition parameters, pollutant propagation maps are generated and dynamic map sequences are continuously formed.

[0024] Spatiotemporal aggregation analysis is performed on the dynamic spectrum sequence to extract the trajectory of pollutant concentration changes at each node, determine the abnormal increase and diffusion direction of pollutant concentration in local areas, and form a pollution diffusion path.

[0025] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method of the present invention, the method of identifying the location of pollution sources includes: using a graph source tracing algorithm to trace the pollution diffusion path in reverse, identifying the upstream node combination with the highest path overlap frequency as a candidate pollution source area, and making a judgment based on the time series information of the concentration peak of each node and the direction of water flow.

[0026] For candidate pollution source areas, relevant remote sensing image data is retrieved for auxiliary verification. The information of known sewage discharge facilities in the area is compared with the pre-stored environmental knowledge graph. The credibility of the pollution source identification results is calculated based on path continuity, temporal rationality and hydrodynamic consistency, and the location of the pollution source is finally confirmed.

[0027] As a preferred embodiment of the big data-based water ecological pollution diffusion prediction method described in this invention, the time-based prediction includes: using the geographical location of the pollution source to obtain the pollutant emission intensity and historical diffusion behavior;

[0028] Based on the pollution diffusion path, the expected arrival time of pollutants, the peak concentration of pollutants, and the duration of pollutant diffusion are predicted; by comparing the predicted peak concentration of pollutants at each node in the hydrological connectivity diagram with the preset safety threshold in real time, it is determined whether there is a risk of the peak concentration of pollutants exceeding the standard in the next time step.

[0029] When a pollutant concentration peak exceeds a preset threshold at a future preset time step, an early warning message is automatically generated; the early warning message includes the pollution diffusion path, the expected time of exceeding the standard, the predicted concentration peak, and the corresponding risk level.

[0030] A big data-based water ecological pollution diffusion prediction system, wherein:

[0031] The data module constructs a sensor network for the aquatic environment, acquires real-time aquatic environment data, preprocesses it using edge computing nodes, and transmits it synchronously to the central platform.

[0032] The pollutant boundary condition parameter module establishes a dynamic prediction model for pollutant boundary condition parameters based on historical data and real-time water environment data.

[0033] The pollution diffusion simulation module constructs a pollutant propagation map based on the dynamic boundary condition parameters, analyzes the spatiotemporal distribution trend of pollutant concentration, and identifies the location of pollution sources.

[0034] The early warning module performs pollution diffusion path simulation and time-based prediction based on the pollution source identification results, and triggers pollution early warning response in combination with set thresholds.

[0035] A computer device includes: a memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0036] A computer-readable storage medium having a computer program stored thereon, characterized in that: when the computer program is executed by a processor, it implements the steps of the method described in any one of the present invention.

[0037] The beneficial effects of this invention are as follows: The water ecological pollution diffusion prediction method based on big data provided by this invention achieves comprehensive acquisition of multi-source data such as water quality, hydrodynamics, meteorology, and remote sensing by constructing a sensor network of the water environment, thereby improving the spatial coverage and timeliness of pollution monitoring; data preprocessing and standardization through edge computing nodes reduces the central computing pressure and enhances the real-time response capability of the system; by integrating historical and real-time data to establish a dynamic boundary condition prediction model, the problem of unstable boundary conditions in pollution diffusion simulation is solved, improving the accuracy and robustness of prediction; the construction of pollutant propagation maps and the combination of graph source tracing algorithms for pollution source identification enhance the intelligence and credibility of pollution tracking; through pollution diffusion path simulation and timely prediction, combined with a threshold-triggered early warning mechanism, proactive early warning and rapid response to pollution risks are achieved, comprehensively improving the intelligence and scientific level of water ecological environment pollution prevention and control. Attached Figure Description

[0038] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0039] Figure 1 The first embodiment of the present invention provides an overall flowchart of a water ecological pollution diffusion prediction method based on big data. Detailed Implementation

[0040] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0041] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for predicting the spread of water ecological pollution based on big data is provided, comprising:

[0042] S1: Construct a sensor network for the water environment, acquire real-time water environment data, preprocess the data using edge computing nodes, and then transmit it synchronously to the central platform.

[0043] Fixed sensors are deployed in key areas of the water body to be monitored (including but not limited to river inlets, industrial discharge points, and downstream outlets) and necessary areas. These fixed sensors are used to continuously monitor water quality parameters (such as pollutant concentration, pH value, dissolved oxygen, temperature, etc.) and fluid parameters (such as flow velocity and flow rate).

[0044] In conjunction with static sensors, mobile sensor nodes, including unmanned surface vessels and / or drones, are deployed in monitoring blind spots or areas of boundary change. These nodes are equipped with corresponding water quality, environmental, and image acquisition devices to supplement the real-time monitoring data from fixed sensors.

[0045] The raw data collected by the aforementioned fixed and mobile sensors are transmitted in real time to the near-end edge computing node via the Internet of Things (IoT) communication network.

[0046] Real-time aquatic environment data includes water quality monitoring data, hydrodynamic data, meteorological background data, remote sensing and image data, biological and ecological indicator data, and metadata data.

[0047] Water quality monitoring data includes pollutant concentrations (including but not limited to chemical oxygen demand, ammonia nitrogen, total phosphorus, total nitrogen, heavy metals, organic pollutants, etc.), pH value, conductivity, dissolved oxygen, turbidity, water temperature, and redox potential.

[0048] Hydrodynamic data, including flow velocity, flow direction, water level, water depth, cross-sectional flow rate, rainfall runoff, and tidal data.

[0049] Meteorological background data includes wind speed, wind direction, temperature, humidity, air pressure, precipitation, and solar radiation intensity.

[0050] Remote sensing and image data, including remote sensing images of the water surface, water color change maps, water pollution zone maps, thermal infrared images, and suspended matter images collected by satellites or drones.

[0051] Biological and ecological indicators include plankton abundance, chlorophyll a concentration in water, algal distribution, and cyanobacteria density.

[0052] Meta-information data includes information such as the timestamp of the monitoring data, the identification of the acquisition device, latitude and longitude coordinates, sampling depth and section number.

[0053] The preprocessing includes noise filtering and outlier detection, removal of false alarms from sensors, filling in missing data using logical rules and model-driven mechanisms, and data cleaning to achieve data standardization and normalization. Data from different sensors undergo format unification and temporal-spatial alignment.

[0054] Furthermore, by deploying fixed and mobile sensor nodes in a coordinated manner, the dynamic perception capability of key areas and areas with changing boundaries is enhanced. Edge computing nodes are used to preprocess raw data, including noise filtering, anomaly detection, missing data filling, and format standardization, ensuring the real-time nature, integrity, and consistency of the data. This provides a high-quality, structured data foundation for subsequent pollution diffusion prediction and pollution source identification, thereby improving the real-time response capability and analytical accuracy of the entire system.

[0055] S2: Based on historical data and real-time water environment data, establish a dynamic prediction model for pollutant boundary condition parameters.

[0056] Upon system startup, a fixed-length data buffer is initialized to store monitoring data from the most recent n time steps. This data buffer can be implemented using a first-in, first-out (FIFO) queue structure, with a length equal to the number of time steps (n). Data at each time point includes multiple pre-processed feature variables, such as water flow rate, water volume, pollutant concentration, water level, water temperature, pH value, and dissolved oxygen.

[0057] The system extracts standardized monitoring data from the most recent n time points from the historical database and fills them into a buffer in chronological order to form a complete time series window. This window serves as the first set of inputs to the model, used to generate initial pollutant boundary prediction results.

[0058] As the system continues to run, whenever new real-time monitoring data is received, the system automatically appends the data to the end of the buffer and removes the oldest data from the beginning of the buffer, thus implementing the sliding window update operation. With each slide, the window always retains the n most recent time-step data, ensuring that the model input maintains the latest temporal sequence state.

[0059] The sliding window can be updated based on time periods (e.g., every 5 or 10 minutes) or events (e.g., the detection of sudden changes in pollutant concentration). Each time the window is updated, it is pushed as a new input sequence to the dynamic prediction model, which includes parallel linear regression and deep learning sub-models.

[0060] The input sequence is organized in a two-dimensional data structure, where the first dimension is the time step and the second dimension is multiple feature variables at each time point. This sequence is directly fed into the model for prediction without needing to be split into single-point data, effectively preserving the contextual characteristics of the time series and improving the model's ability to learn the spatiotemporal trends of pollutants.

[0061] The sliding window is continuously updated in sync with the data acquisition process, and the model prediction results are fed back to the pollution diffusion simulation module. Simultaneously, the system supports error tracking and incremental model learning mechanisms. When there is a deviation between the sliding window input and actual observations, the system can dynamically adjust model parameters or trigger online retraining, further enhancing the system's responsiveness to sudden pollution events.

[0062] Set a time step, and use preprocessed historical data and real-time data from the past n time steps to form a water environment data sequence of n consecutive time steps through a sliding window method as input for the dynamic prediction model.

[0063] The dynamic prediction model adopts a fusion prediction framework, which consists of a parallel linear regression model and a deep learning model.

[0064] The linear regression model captures the linear changing trend of aquatic environmental parameters and outputs prediction results based on the aquatic environmental data sequence over the past n time steps.

[0065] The deep learning model utilizes a GRU network structure to capture the nonlinear dynamic characteristics of aquatic environmental parameters, predicts aquatic environmental data sequences, and supports online incremental training to adapt to real-time data changes.

[0066] The predicted outputs of the linear regression model and the deep learning model are weighted and fused to generate the final pollutant boundary condition parameters.

[0067] The output of the linear model is The output of the GRU model is: The fusion output result is:

[0068]

[0069] in, This represents the predicted boundary flow rate. This represents the predicted boundary flow. This indicates the predicted boundary pollutant concentration. This indicates the predicted boundary water level height.

[0070] Where the fusion weight α t It can adapt and change, and is defined as follows:

[0071]

[0072] Where, Δ t β represents the absolute difference between the predictions of the two models at time t, reflecting the degree of bias in the model output. β represents the sensitivity control factor for the fusion weights.

[0073] The pollutant boundary condition parameters include: water flow rate, flow rate, pollutant concentration, and water level at the boundary.

[0074] When new real-time water environment data arrives, the sliding window input is updated, and the coefficients of the linear regression model are updated using the recursive least squares method. The parameters of the deep learning model are also updated through incremental training based on real-time data, thereby achieving continuous optimization of the model's prediction accuracy.

[0075] Furthermore, by constructing a dynamic prediction model that integrates historical and real-time aquatic environmental data, high-precision, real-time updates of pollutant boundary conditions (such as flow rate, flow rate, concentration, and water level) are achieved. A sliding window mechanism ensures the model always runs based on the latest time-series data. A linear regression model captures stable trends in environmental parameters, while a deep learning model (GRU) identifies complex nonlinear variations. The prediction results from both models are then fused using adaptive weights. This design significantly enhances the system's ability to perceive and predict pollution diffusion input conditions, providing stable and accurate data support for subsequent pollution propagation simulations and early warnings. Through error feedback and incremental update mechanisms, the model achieves continuous adaptation and intelligent evolution under sudden environmental changes.

[0076] S3: Based on the dynamic boundary condition parameters, construct a pollutant propagation map, analyze the spatiotemporal distribution trend of pollutant concentration, and identify the location of pollution sources.

[0077] The construction of pollutant propagation maps includes: using dynamic boundary condition parameters as the driving force, and combining water body geographic information and watershed structure to generate a hydrological connectivity map. Nodes in the hydrological connectivity map represent key water area monitoring points, tributary confluences, and drainage outlets; edges represent the flow relationships between water bodies; and the directionality, velocity, and flux of the edges are assigned and updated in real time by the dynamic boundary condition parameters.

[0078] Pollutant boundary condition parameters output in S2 stage In the middle, the predicted pollutant concentration As a node v in the graph i Dynamic properties; predicted water flow rate and traffic Mapped to edge e ij Directionality and transmission capacity; predicted water level height Used to determine the effective flux status of edges, dynamically control the activity of edges, and ensure that the map reflects the current real flow environment of the water body.

[0079] At regular time steps, the system reconstructs and updates the propagation map based on the latest dynamic boundary condition parameters, forming a continuous dynamic map sequence. Real-time collected pollutant concentration data serves as the basis for correcting node attributes, further improving the accuracy of the map in reflecting the actual migration state of pollutants.

[0080] The system performs spatiotemporal aggregation analysis on the dynamic map sequence to extract the migration trajectory and concentration evolution trend of pollutants between nodes. If the pollutant concentration of a node increases significantly in consecutive time steps, and the edge connected to the node flows from upstream to that node, the system marks it as a node with potential pollution impact.

[0081] During the pollution source identification phase, the system uses the current pollution path as a foundation and invokes a graph source tracing algorithm. Based on the diffusion path driven by dynamic boundary condition parameters in the dynamic graph, it traces the pollution migration trajectory backward and identifies upstream nodes with high path overlap frequencies. Combining the peak concentration time of each node, the propagation time required, and the hydrodynamic direction, a set of pollution propagation paths for candidate nodes 's' is selected.

[0082] For candidate pollution source areas, relevant remote sensing image data is retrieved for auxiliary verification. The information of known sewage discharge facilities in the area is compared with the pre-stored environmental knowledge graph. The credibility of the pollution source identification results is calculated based on path continuity, temporal rationality and hydrodynamic consistency, and the location of the pollution source is finally confirmed.

[0083] After completing the initial path tracing, the credibility scoring formula is defined as follows:

[0084]

[0085] in, This represents the set of contamination propagation paths that pass through candidate node s. γ p This represents the maximum pollutant concentration along path p. T p This represents the time required for a pollutant to propagate from the starting point to the end of the path. δ(p,s)∈{0,1} indicates whether the path passes through candidate node s; it is 1 if it does, and 0 otherwise.

[0086] Credibility R s A score greater than 0.8 indicates that the node is a genuine pollution source. The system combines this score with the frequency of overlapping paths in the graph source tracing, the temporal characteristics of concentration, and the knowledge graph matching results as an important reference for calculating the credibility of pollution sources.

[0087] Furthermore, by constructing a pollutant propagation map and combining it with a graph source tracing algorithm, quantitative modeling and visual tracking of the spatiotemporal propagation paths of pollutants are achieved. Based on this, a credibility scoring mechanism based on pollution intensity, propagation timeliness, and path structure is introduced, making the pollution source identification process more scientific, systematic, and quantitatively grounded. By fusing the scoring results with remote sensing images, knowledge graphs, and node concentration time-series information, the accuracy and reliability of pollution source location are effectively improved, avoiding potential misjudgments or omissions that might occur with a single source tracing method, and providing a highly reliable data foundation for subsequent pollution diffusion prediction and early warning.

[0088] S4: Based on the pollution source location identification results, perform timely prediction and trigger pollution early warning response in combination with the set threshold.

[0089] After confirming the location of the pollution source and constructing the propagation path, the system performs a time-series prediction of the pollutant diffusion process in the water body based on the dynamic map evolution results driven by dynamic boundary condition parameters. Specifically, this includes utilizing the location of the pollution source node 's', the pollutant emission intensity, and the corresponding set of historical propagation paths. Combining the attributes of each edge in the path (including water flow rate) flow water level height ) and node pollutant concentration The arrival time and concentration evolution trends of pollutants at each downstream node were extrapolated. All parameters mentioned above are derived from the dynamic boundary condition parameters of the previous stage. The system integrates the predicted output and time-series update process of the graph to ensure consistency between the pollutant diffusion simulation and the overall prediction logic. For each downstream node i in the hydrological connectivity graph, the system calculates the predicted peak pollutant concentration at a future time step t+ΔT. Subsequently, the predicted value is compared with the multi-level safety thresholds θ1, θ2, and θ3 corresponding to the pollutant type to determine the node risk level.

[0090] The predicted peak pollutant concentration at node i at future time step t+ΔT Its risk level formula is expressed as:

[0091]

[0092] Where θ1, θ2, θ3 represent the multi-level safety thresholds corresponding to the pollutant types. i Indicates the warning level (0 = safe, 1 = slightly dangerous, 2 = medium dangerous, 3 = high dangerous), used for multi-channel alarm linkage.

[0093] When L i When the value is 0, the system does not trigger an alarm, but only records data for model learning and posterior verification, updating the spatiotemporal evolution trend of pollutants.

[0094] When L i When the value is 1, the system enters a monitoring and early warning state, marking a yellow warning area on the platform interface. Simultaneously, a notification is pushed to the administrator's app, and local frequency enhancement monitoring of edge nodes is initiated.

[0095] When L i When the value is 2, the system triggers an orange alert, sending an SMS / voice notification to the relevant environmental regulatory authorities. It also recommends manual sampling and verification in the downstream areas of the pollution path.

[0096] When L i When the threshold is 3, the system activates a red high-priority alarm, triggering on-site alarm devices (such as warning lights and broadcasts); it retrieves emergency response suggestions from the database and pushes them to the decision-making system. Simultaneously, it marks high-risk pollution transmission paths and continuously refreshes the prediction window.

[0097] Furthermore, by comparing the future pollution concentrations at each key node with multi-level safety thresholds, a tiered early warning triggering and response linkage is achieved. This mechanism can not only identify risk areas that may exceed standards in advance, but also automatically match corresponding response strategies (such as platform prompts, SMS notifications, on-site alarms, and emergency plan pushes) according to the differences in risk levels, thereby enabling early detection, early warning, and early intervention of pollution incidents, effectively improving the system's dynamic perception and intelligent response capabilities to sudden water pollution.

[0098] Example 2, an embodiment of the present invention, provides a water ecological pollution diffusion prediction system based on big data, comprising:

[0099] The data module constructs a sensor network for the aquatic environment, acquires real-time aquatic environment data, preprocesses it using edge computing nodes, and transmits it synchronously to the central platform.

[0100] The pollutant boundary condition parameter module establishes a dynamic prediction model for pollutant boundary condition parameters based on historical data and real-time water environment data.

[0101] The pollution diffusion simulation module constructs a pollutant propagation map based on the dynamic boundary condition parameters, analyzes the spatiotemporal distribution trend of pollutant concentration, and identifies the location of pollution sources.

[0102] The early warning module performs pollution diffusion path simulation and time-based prediction based on the pollution source identification results, and triggers pollution early warning response in combination with set thresholds.

[0103] Example 3, an embodiment of the present invention, differs from the previous two embodiments in that:

[0104] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0105] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0106] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0107] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0108] Example 4 is an embodiment of the present invention, which provides a method and system for predicting the spread of water ecological pollution based on big data. In order to verify the beneficial effects of the present invention, a simulation experiment is conducted for scientific demonstration.

[0109] This embodiment uses River A in the suburbs of a city as the test area. The total length of the river is approximately 23.5 kilometers, with 4 inlets, 2 industrial discharge points, and 1 drinking water source protection area. To verify the accuracy and response capability of the big data-based water ecological pollution diffusion prediction method proposed in this invention, a complete experimental system was constructed and deployed at key sections and potentially high-pollution areas in the river basin.

[0110] First, 12 fixed sensing nodes were deployed along the river, including conventional water quality sensors (pH, conductivity, dissolved oxygen, COD, ammonia nitrogen, total phosphorus, total nitrogen, and temperature) and flow velocity sensors. Each sensing node is equipped with an edge computing unit to perform preliminary local data processing and filter out anomalies. Two unmanned surface vessels (USVs) equipped with water quality sensors and camera modules were also deployed to compensate for blind spots in the static nodes, particularly for high-frequency patrol monitoring near tributaries and industrial discharge outlets. The system sampling frequency was set to once every 5 minutes.

[0111] The received data undergoes denoising and data cleaning at edge nodes before being transmitted to the central platform. In the prediction model construction, monitoring data from the past 72 hours is used as the historical training sequence, forming a sliding window input sequence with 12 time steps at a time. A linear regression sub-model and a GRU deep learning sub-model are used respectively to predict the trends of pollutant concentration and flow rate, and the outputs are fused. The weight α is dynamically adjusted by the prediction bias. The model adaptively adjusts hourly and supports online incremental training.

[0112] At 08:00 on the 3rd day, the system detected a sharp increase in ammonia nitrogen concentration in an upstream tributary (node ​​N3) from 0.86 mg / L to 2.35 mg / L, which was expected to propagate to downstream node N6 within the next hour. Combining indicators such as pollution path, rate of change of peak concentration, and water flow direction, the system determined that the pollution source was an industrial discharge point upstream of N3. Remote sensing images were retrieved to verify a suspicious color difference in the water body in the area, which overlapped with the location of a dyeing and printing factory registered in the knowledge graph. The pollution source confidence score reached 0.91, thus confirming the pollution source.

[0113] Subsequently, the system performed diffusion predictions for five key nodes downstream of the pollution path. During the simulation time step of 08:10–10:00, the predicted peak concentration at node N6 was 2.14 mg / L, exceeding the regional threshold of 1.50 mg / L. The system automatically triggered an orange alert mechanism, highlighting the risk area on the platform interface and simultaneously pushing alert information to management personnel's mobile phones. It also recommended immediately initiating a downstream backup water source switching operation to mitigate the risk impact.

[0114] The average ammonia nitrogen concentration of N3 in the upstream tributary was 0.82 mg / L in the 24 hours prior to the event trigger, and the value at the time of the abnormal increase was 2.35 mg / L. The proportion of false alarm data removed by edge node preprocessing was 3.67%.

[0115] After model fusion prediction, the N4 node predicted a concentration of 1.82 mg / L at 08:20, while the measured value was 1.79 mg / L.

[0116] The predicted arrival time of pollution at node N6 was 09:05, while the actual concentration exceeding the standard occurred at 09:03, with a deviation of less than 2 minutes.

[0117] The predicted peak value for node N6 was 2.14 mg / L, while the actual measured value was 2.10 mg / L.

[0118] The mean squared error of the GRU model's predictions is 0.084 when used alone, 0.131 for the linear model, and 0.056 for the fusion model.

[0119] The traditional model has a prediction bias of 11.3%, while the model of this invention has a prediction bias of 3.7%, representing a reduction in error of 67.3%.

[0120] The system's early warning trigger lead time is 58 minutes, compared to a 15-minute delay in alarms compared to traditional solutions.

[0121] The overall runtime latency is controlled at 2.65 seconds per iteration, compared to approximately 9.84 seconds for the traditional model.

[0122] The experimental process and data presented above demonstrate that the method of this invention exhibits significantly superior performance and responsiveness compared to existing technologies in predicting pollutant diffusion. Firstly, regarding prediction accuracy, this invention constructs a fusion-based dynamic prediction model. Linear regression is used to capture the fundamental trends of pollutant changes, while GRU is used to learn the complex nonlinear fluctuation characteristics in time series, resulting in stronger generalization ability and adaptability in real-world environments. The fusion mechanism introduces a dynamic weighting factor, adjusting it in real time based on model bias, significantly reducing the risk of deviation from the single model for specific pollution patterns. Data shows that the fusion model's mean squared error is 0.056, a reduction of over 67% compared to traditional models, indicating significant convergence in prediction error.

[0123] Secondly, regarding pollution source identification, this invention constructs a dynamic map and introduces a graph-based source tracing algorithm. This not only considers the frequency of path overlap but also incorporates the temporal sequence of pollutant concentration peaks, water flow direction, and auxiliary verification from remote sensing images, making pollution source location more scientifically based. In this implementation, the pollution source identification reliability score reached 0.91, far exceeding traditional methods based on concentration abrupt change points (scores are generally below 0.75), significantly enhancing the reliability of pollution source judgment.

[0124] Third, regarding early warning and response, this invention employs a dynamic pollution level classification mechanism to trigger an early warning 58 minutes before the pollution prediction result reaches a set threshold, simultaneously pushing the warning to user terminals and on-site emergency equipment, thus forming a closed-loop process from identification to response. Compared to the passive handling solutions caused by delayed alarms in traditional models, this solution provides decision-makers with more emergency response windows.

[0125] Furthermore, the edge computing mechanism of this invention also demonstrates significant practical advantages. During the experiment, the processing latency of the edge nodes for raw data was controlled within 3 seconds, enabling false alarm removal, anomaly identification, and preliminary judgment to be completed immediately, thus providing substantial processing space for backend modeling. In contrast, traditional methods, which process all data centrally, result in significant processing delays and are unsuitable for high-frequency monitoring scenarios.

[0126] In summary, this embodiment demonstrates through real-world scenario deployment and complete data stream verification that the pollution diffusion prediction method proposed in this invention has significant advantages in multiple aspects, including prediction accuracy, early warning timeliness, pollution source location accuracy, and system real-time performance. It also has broad prospects for application in various types of aquatic ecosystems, reflecting the novelty and substantial progress of this method in terms of technical means.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting diffusion of water ecological pollution based on big data, characterized in that, The method comprises the following steps: A sensing network is constructed to obtain real-time water environment data, and the data is preprocessed by an edge computing node to generate preprocessed real-time water environment data and synchronously transmitted to a central platform; Based on historical data and preprocessed real-time water environment data, a dynamic prediction model is established to output dynamic boundary condition parameters, including water flow rate, flow, pollutant concentration and water level at the boundary; According to the dynamic boundary condition parameters, a pollutant propagation graph is constructed to analyze the spatio-temporal distribution trend of pollutant concentration and identify the location of the pollution source; According to the pollution source location identification result, time-effective prediction is performed, and a pollution early warning response is triggered in combination with a set threshold value; The construction of the pollutant propagation graph includes generating a hydrological connectivity graph based on the dynamic boundary condition parameters, combining water geographical information and basin structure, wherein the nodes in the graph represent key water area monitoring points, tributary junctions and drainage outlets, and the edges represent the flow relationship between water bodies; the directionality, flow rate and flux of the edges are real-time assigned and updated by the dynamic boundary condition parameters; In the output dynamic boundary condition parameters, the predicted pollutant concentration is the dynamic attribute of the nodes in the graph; the predicted water flow rate and flow are mapped to the directionality and transmission capacity of the edges; the predicted water level is used to determine the effective flux state of the edges, dynamically control the activity of the edges, and ensure that the graph reflects the current real flow environment of the water body; At a predetermined time step, the pollutant propagation graph is generated based on the updated dynamic boundary condition parameters, and a dynamic graph sequence is continuously formed; The dynamic graph sequence is subjected to spatio-temporal aggregation analysis, the change trajectory of the pollutant concentration of each node is extracted, the abnormal growth and diffusion direction of the pollutant concentration in the local area are judged, and a pollution diffusion path is formed; The identification of the pollution source location includes: using a graph tracing algorithm to reversely track the pollution diffusion path, identifying the upstream node combination with the highest path overlap frequency as a pollution source candidate area, and combining the time sequence information of the concentration peak value of each node and the water flow direction to make a judgment; The corresponding remote sensing image data of the candidate pollution source area is called to assist in verification, and the known pollution discharge facility information in the area is compared with the pre-stored environmental knowledge graph, and the pollution source identification result is subjected to credibility calculation according to the path continuity, time sequence rationality and hydrodynamic consistency, and finally the pollution source location is confirmed. 2.The water ecological pollution diffusion prediction method based on big data according to claim 1, wherein: The real-time water environment data includes water quality monitoring data, water power related data, weather background data, remote sensing and image data, biological and ecological index data, and meta-information data; The preprocessing includes performing noise filtering and outlier detection, eliminating sensor false alarm data, filling in missing data using logical rules and model-driven mechanisms, and completing data cleaning operations to realize data standardization and standardization processing; The data from different sensors are subjected to format unification and time-space dimension alignment processing. 3.The water ecological pollution diffusion prediction method based on big data according to claim 2, characterized in that: The establishment of the dynamic prediction model includes: setting a time step, using the preprocessed historical data and real-time data of the past n time steps to form a water environment data sequence of continuous n time steps as the input of the dynamic prediction model through a sliding window method. The dynamic prediction model adopts a fusion prediction framework composed of a linear regression model and a deep learning model in parallel; The linear regression model captures the linear trend of the water environment parameters and outputs the prediction results based on the past n time steps of water environment data sequences; The deep learning model uses a GRU network structure to capture the nonlinear dynamic characteristics of the water environment parameters, predicts the water environment data sequences, and supports online incremental training to adapt to real-time data changes; The prediction outputs of the linear regression model and the deep learning model are weighted and fused to generate dynamic boundary condition parameter outputs. 4.The water ecological pollution diffusion prediction method based on big data according to claim 3, characterized in that: The dynamic prediction model also includes updating the sliding window input when new real-time water environment data arrives, and updating the linear regression model coefficients using the recursive least squares method and updating the deep learning model parameters based on real-time data. 5.The water ecological pollution diffusion prediction method based on big data according to claim 4, characterized in that: The time-effective prediction includes: obtaining the pollution emission intensity and historical diffusion behavior of the pollution source based on its geographic location; Based on the pollution diffusion path, the arrival time of the pollutant, the peak concentration of the pollutant, and the duration of the pollutant diffusion are predicted; by comparing the predicted peak concentration of the pollutant at each node in the hydrological connectivity graph with the preset safety threshold, it is determined whether there is a risk of exceeding the peak concentration of the pollutant in the future time step; When the future preset time step has a peak concentration of the pollutant exceeding the preset threshold, an early warning information is automatically generated; the early warning information includes the pollution diffusion path, the predicted exceeding time, the predicted concentration peak, and the corresponding risk level.

6. A water ecological pollution diffusion prediction system based on big data, for executing the method of any one of claims 1-5, characterized in that: a data module constructs a sensor network of the water environment, obtains real-time water environment data, and generates preprocessed real-time water environment data after preprocessing the data using edge computing nodes, and synchronously transmits the data to the central platform; a pollutant boundary condition parameter module, based on historical data and preprocessed real-time water environment data, establishes a dynamic prediction model, and outputs dynamic boundary condition parameters; a pollution diffusion simulation module, according to the dynamic boundary condition parameters, constructs a pollutant propagation map, analyzes the spatio-temporal distribution trend of the pollutant concentration, and identifies the location of the pollution source; an early warning module, according to the pollution source location identification result, performs time-effective prediction, and triggers the pollution early warning response combined with the set threshold.

Images